Sickle hemoglobin oxygen affinity‐shifting strategies have unequal cerebrovascular risks

Abstract

The accelerating development of potential therapeutic agents for sickle cell anemia (SCA) is welcome and encouraging. Yet, our present comments are prompted by one of these drugs, GBT440,1-5 for which trials in United States and Europe are currently enrolling subjects with SCA. We are somewhat puzzled by the rapid advance of this agent that, in our opinion, has a concerning mode of action that enhances cerebrovascular risk. We here explain this via a linear argument, a simplified (i.e., manageable) model, and an illustrative device. We conclude with specific recommendations to mitigate risk. Other HbS-modifying, affinity-shifting drugs that were previously proposed for SCA (BW12C,6, 7 Tucaresol,8 5-HMF9) also appear to exhibit the same concerning effect we address here. Like GBT440, each forms a Schiff base adduct with globin N-termini, thereby stabilizing hemoglobin its oxy configuration. We use GBT440 as our example because: it is the only such agent now in clinical trials; those studies are enrolling children; its proposed use is to induce ∼30% Hb modification; and relevant data are accessible via three publications1-3 and thirteen meeting posters.4 The purported mechanism of action of GBT440, increasing oxygen affinity, is one of the five categorical strategies that can be used to inhibit HbS polymerization, as recently reviewed by Eaton and Bunn.10 Importantly, they noted that it is difficult to know if the net effect of affinity-shifting agents would be helpful (from decreasing polymerization) or harmful (from compromising oxygen delivery). However, we emphasize that the general concept of affinity-shifting agents can be parsed into different modes of action—and that these have strikingly divergent, and perhaps predictable, risk-to-benefit considerations. A Thought Experiment On Oxygen Delivery We will compare the impacts of two affinity-shifting approaches upon oxygen delivery issues. Recognizing that the complexity of oxygen delivery—from a breath of air to ultimate mitochondrial respiration—renders definitive modeling impossible, we necessarily employ simplifying assumptions that facilitate comparison. All are listed in Table 1, but three are most important. One is that metabolic oxygen demand (and, therefore, oxygen extraction from blood) remains fixed at an average, normal resting value (5 ml O2 per 100 ml blood, i.e., 5 vol%). Another is that both normal and sickle subjects start with whole blood P50 = 27 mm Hg. And another is that the burden of oxygen delivery always must be met without the benefit of increasing microvascular blood flow. The justification for these assumptions, as our starting point, lies in our present primary focus on the brain, as will become apparent later. Nonetheless, farther below we also will describe the impact of re-adding to our model several features of the real sickle milieu. Meanwhile, we suggest that clarity emerges from this simplicity. A Starting Point Given to a normal person, GBT440 (at ∼30% modification) shifts the ODC from curve A to curve E (Figure 1A): P50 ∼18 mm Hg, but biphasic and non-sigmoidal, with a shoulder on the left (the 30% modified, extremely high affinity tetramers) and on the right, emergence of normal oxygen binding by the 70% unmodified tetramers.1 In other words, GBT440 induces the P50 shift because the extreme-affinity component essentially “pulls” the ODC to the left. Interestingly, ODC like curve E are seen in thalassemics with high levels of HbH11 that, like GBT440-modified tetramers, has a myoglobin-like P50, a hyperbolic ODC, without cooperativity or Bohr effect or response to 2,3-DPG.11, 12 The Concept Of Functional Oxygen Content (Foc) To illustrate drug impacts we also plot the oxygen dissociation curves for “Functional Oxygen Content” (FOC; Figure 1B), a device previously used by Bunn and Forget.13 The FOC is what remains of hemoglobin's total oxygen content after the contribution from an extreme-affinity component is subtracted out. So, FOC is the amount of oxygen that can actually be off-loaded at a capillary pO2 that is both achievable and tolerable. They are not the same thing. For skeletal muscle and brain cortex, the critical capillary pO2 threshold is believed to be pO2 ∼15 to 20 mm Hg.14-19 So, myoglobin20 and HbH11,12 and GBT440-modified Hb tetramers1—all of which have P50 ∼2 mm Hg—would be unhelpful for oxygen delivery. (We acknowledge that their contribution would not quite be zero. But a small amount of oxygen unloading from a small proportion of hemoglobin would amount to a very small contribution to oxygen delivery. So we here ignore it in our FOC calculation.) For the normal and sickle patient at baseline, their FOC (Figure 1B) are identical to their ODC (Figure 1A) because all hemoglobin tetramers are fully functional (curves A and B). To assist interpretation of FOC, we display in Table 2 the numerical values for parameters important in sickle cell anemia. Compared to normal (line A), for the anemic sickle patient (line B) to meet metabolic oxygen demand already requires a far lower capillary pO2 and an increased proportionate oxy-to-deoxyHb conversion, with higher ending RBC deoxyHb concentration. Oxygen binding curves. These illustrations of oxygen dissociation curve (ODC, panel A) and functional oxygen content (FOC, panel B) are schematic and intended only to highlight P50 (oxygen affinity) and curve height (amount of bound oxygen, in vol%). These were drawn simply using the graphics function of PowerPoint®. Panel 1A ODC: curve A, normal (Hb=14 g/dl); curve B, sickle baseline (Hb=8 g/dl); curve C, newborn as well as a normal given hypothetical drug “↓DPG”; curve D, HbH and myoglobin and GBT440-modified tetramers; curve E, normal given GBT440 to modify 30% of tetramers; Panel 1B FOC: curve A, normal; curve B, sickle baseline; curve F, sickle given “↓DPG”; curve G, sickle given GBT440 to modify 30% of tetramers. Physiologic Penalties The physiologic implications of the two affinity-shifting strategies are strikingly different. [a] ↓DPG, our imaginary drug, given to the sickle patient shifts FOC from curve B to curve F (Figure 1B), a left shift that retains functionality and entails no loss of maximal oxygen carrying capacity. Per Table 2 (lines B vs F) this does require a further—but manageable—lowering of capillary pO2 to instigate oxygen off-loading, but it does not require any increase in oxy-to-deoxyHb conversion. Yet, this is not without physiological impact. Long ago we demonstrated that a P50 left-shifted from 27 to ∼18 mm Hg is enough to impair maximal exercise ability when ambient oxygen is plentiful.21 Superimposed on a significantly anemic patient, it is quite possible that ↓DPG effects would be noticeable at only moderate exertion levels. On the good side, a left-shift of this magnitude is protective when oxygen availability is limiting,21 as for sickle patients with impaired oxygenation. [b] GBT440, at 30% tetramer modification, eliminates 30% of oxygen delivery capability, superimposed upon the sickle baseline anemia. The net effect is shifting the FOC from curve B to curve G (Figure 1B), with substantial physiologic penalty (Table 2, lines B vs G): further lowering of maximal O2 content, that requires an even lower capillary pO2, to induce a far greater proportionate oxy-to-deoxyHb conversion, resulting in no improvement in the ending deoxyHbS concentration. Of note, this degree of GBT440 modification renders the patient functionally equivalent to having Hb = 5.6 g/dl. Risk Concerns The Perplexing Complexity Of Reality Cerebrovascular Jeopardy The brain comprises ∼2% of body weight but demands ∼20% of oxygen consumption.18 An increase in microvascular flow, requiring arteriolar vasoregulation, is normally its major regulatory solution to insufficient oxygen provision.16 Yet, the systemic endothelial dysfunction of sickle cell anemia impedes normal arteriolar vasoregulation.26 The elevated flow seen in the Circle of Willis, even absent vasculopathic lesions, is deceiving since the sickle patient's brain—like other organs—displays the “perfusion paradox” 27 with macrovascular hyperperfusion but microvascular hypoperfusion. Therefore, sickle patients at baseline already reside in a state of high brain vulnerability, with marginal cerebrovascular blood flow and an impaired ability to supply need.28 Indeed, it is believed that this may underlie their alarming ∼50% lifetime incidence of silent stroke. If a GBT440-like drug is added to the mix, absence of the normal compensatory flow increases would make it challenging to offset the drug-induced reduction of FOC. We, therefore, are concerned that, in sickle context, GBT440-like drugs would impose a state of enhanced cerebrovascular risk. Recommendations Our analysis has assumed GBT440 drug dosing to achieve 30% modification, as proposed by the drug's developers.1 It is possible that there may be a “sweet spot” of lower modification that entails a less unfavorable risk-to-benefit ratio, if there indeed is clinical benefit. Obviously, the balance of benefit and risk can only be truly revealed by clinical studies. But we suggest that these should be advanced with heightened caution. Conclusion We urge our colleagues to be circumspect in application of affinity-modulating drugs in sickle cell anemia—and to recognize that different modes of action may have divergent risk-to-benefit considerations. Clinical studies and care should be conducted with awareness of physiological potential for inadvertent harm, in particular in the vulnerable brain. Silent disease is silent—if not specifically sought, it will not be recognized. And regarding available data, it is paramount to remember that: absence of evidence is not the same thing as evidence of absence. Both authors have advised multiple companies regarding potential therapeutics for sickle disease; neither has a conflict of interest at this time. RPH and BEH wrote this manuscript.