Genotype-Response Correlation in DRIVE PK, a Phase 2 Study of AG-348 in Patients with Pyruvate Kinase Deficiency

Pyruvate kinase (PK) deficiency is a congenital hemolytic anemia caused by mutations in the PKLR gene, leading to a deficiency of the glycolytic enzyme red cell PK (PK-R). AG-348 is an orally available small-molecule allosteric activator of PK-R that activates the wild type and a range of mutant PK-R enzymes associated with PK deficiency (Kung et al. Blood 2017;130:1347-56).

In a phase 2 clinical study in patients with PK deficiency (DRIVE PK, NCT02476916), 26 of 52 patients (50%) experienced a maximum hemoglobin (Hb) increase of >1.0 g/dL (mean maximum increase, 3.4 g/dL; range, 1.1-5.8 g/dL). In most cases, Hb increases were rapid and sustained, and seen across a wide dose range from 5 to 300 mg twice daily. Hemolysis markers (reticulocytes, indirect bilirubin, haptoglobin) improved in patients who experienced a maximum Hb increase of >1.0 g/dL. Hb increases were observed in patients with a variety of PKLR mutations and were associated with the presence of at least 1 missense mutation.

As PK deficiency is a genetically heterogeneous disease, with over 200 described mutations, we sought to understand in greater detail the molecular parameters that were associated with Hb increases in AG-348-treated patients. Here, we have analyzed the relationship between Hb increase and patient genotype, biochemical response to AG-348 treatment, and baseline PK-R protein level. The results are consistent with the proposed mechanism of action of AG-348 through binding and activation of residual mutant PK-R enzyme.

Of the 106 PKLR mutations in the 52 subjects (2 subjects carried 3 mutations), there were 47 unique mutations, including 28 missense and 19 non-missense. Of these mutations, only 10 were present in at least 2 patients, highlighting the intrinsic limitations of findings from this small sample size. That notwithstanding, among these 10 mutations, we observed that some were more likely to be associated with Hb increases (e.g. increases were observed in 5/5 patients with at least 1 R486W mutation and 4/4 patients with at least 1 T384M mutation). Other mutations, including V134D (0/3), E241stop (1/6), and R510Q (5/14), were associated with a lack of Hb increase. Notably, 1/10 patients who had 2 non-missense mutations and 0/5 patients who were homozygous for the R479H "Amish" mutation had Hb increases.

We have previously reported that, in biochemical assays, AG-348 could efficiently activate recombinantly expressed R479H and R510Q mutant enzymes. Although both of these are technically missense mutations, it is well understood that R510Q results in severely decreased protein stability (Wang et al. Blood 2001;98:3113-20), and the R479H mutation, occurring at the splice junction between 2 exons, can deleteriously affect mRNA splicing (van Wijk et al. Br J Haematol 2004;125:253-63). Therefore, as AG-348 works by directly binding and stimulating the activity of residual mutant PK-R enzyme, we explored the hypothesis that PK-R protein levels in patients with these mutations might be too low to allow for adequate target engagement at clinically tested doses.

Using antibody-based capture and detection, we quantitated baseline PK-R protein levels of DRIVE PK subjects. Analysis of these data revealed that patients with Hb increases had, on average, higher PK-R protein levels (49% of reference control compared to 13% in patients without Hb increases). All 10 patients with non-missense/non-missense mutations had nearly undetectable PK-R protein levels (average of 3.7% of reference control), consistent with and confirming the predicted effect of these mutations on protein expression. Baseline PK-R protein levels were also lower in patients with at least 1 R510Q (18% of reference control) or R479H (19% of reference control) mutation compared to patients with other missense mutations (59% of reference control).

Taken together, these analyses demonstrate that Hb increases upon AG-348 treatment are associated with the presence of at least 1 missense PK-R mutant enzyme with residual protein for AG-348 to bind and activate. These data strongly suggest that the hematological effects of AG-348 come from the proposed mechanism of action of restoration of glycolytic pathway activity in PK-deficient red cells. The genetic complexity of PK deficiency, combined with the relatively small sample size, limit extrapolation of these findings to the entire PK deficiency patient population, but may help to inform future studies.