Cipaglucosidase alfa plus miglustat: linking mechanism of action to clinical outcomes in late-onset Pompe disease

Abstract

Enzyme replacement therapy (ERT) is the only approved disease-modifying treatment modality for Pompe disease, a rare, inherited metabolic disorder caused by a deficiency in the acid α-glucosidase (GAA) enzyme that catabolizes lysosomal glycogen. First-generation recombinant human GAA (rhGAA) ERT (alglucosidase alfa) can slow the progressive muscle degeneration characteristic of the disease. Still, most patients experience diminished efficacy over time, possibly because of poor uptake into target tissues. Next-generation ERTs aim to address this problem by increasing bis-phosphorylated high mannose (bis-M6P) N-glycans on rhGAA as these moieties have sufficiently high receptor binding affinity at the resultant low interstitial enzyme concentrations after dosing to drive uptake by the cation-independent mannose 6-phosphate receptor on target cells. However, some approaches introduce bis-M6P onto rhGAA via non-natural linkages that cannot be hydrolyzed by natural human enzymes and thus inhibit the endolysosomal glycan trimming necessary for complete enzyme activation after cell uptake. Furthermore, all rhGAA ERTs face potential inactivation during intravenous delivery (and subsequent non-productive clearance) as GAA is an acid hydrolase that is rapidly denatured in the near-neutral pH of the blood. One new therapy, cipaglucosidase alfa plus miglustat, is hypothesized to address these challenges by combining an enzyme enriched with naturally occurring bis-M6P N-glycans with a small-molecule stabilizer. Here, we investigate this hypothesis by analyzing published and new data related to the mechanism of action of the enzyme and stabilizer molecule. Based on an extensive collection of in vitro, preclinical, and clinical data, we conclude that cipaglucosidase alfa plus miglustat successfully addresses each of these challenges to offer meaningful advantages in terms of pharmacokinetic exposure, target-cell uptake, endolysosomal processing, and clinical benefit.

Keywords: Pompe disease; enzyme replacement therapy; glycogen storage disease type II; lysosomal storage disorders; n-butyldeoxynojirimycin.

Figure 1

Summary of challenges in delivering rhGAA for Pompe disease.

Figure 2

Proposed approach to overcoming key challenges in delivering rhGAA.

Figure 3

Cipaglucosidase alfa has increased competence for intracellular delivery versus alglucosidase alfa because of enhanced bis-M6P glycosylation. Cipaglucosidase alfa (A) or alglucosidase alfa (B) was loaded onto a CI-MPR column to assess the relative capacity for each enzyme mixture to interact with the receptor through M6P-bearing N-glycans. Following binding, rhGAA was eluted with increasing concentrations of free M6P (dashed red lines). Eluted rhGAA was collected in fractions and levels measured with a 4-MU-α-D-glucopyranoside (4-MU-Glc) enzyme activity assay (solid blue and green lines). The vast majority (~95%) of cipaglucosidase alfa bound tightly to the CI-MPR column, eluting only at high M6P concentrations. In contrast, less alglucosidase alfa (27%) bound to the CI-MPR column, reflecting a lack of M6P-bearing N-glycans on most enzyme molecules. (B) Used with permission from AME Publishing PTE LTD: ‘Challenges in treating Pompe disease: an industry perspective’; Do et al. (24). Permission conveyed through Copyright Clearance Center, Inc.

Figure 4

Cipaglucosidase alfa binds to CI-MPR with significantly better affinity than alglucosidase alfa. Cipaglucosidase alfa and alglucosidase alfa were incubated with immobilized soluble CI-MPR receptor in a 96-well plate-based binding assay to determine K_D. Cipaglucosidase alfa and alglucosidase alfa at concentration ranges of 0.027–180 and 0.74–539 nM, respectively, were incubated with CI-MPR. Unbound GAA was washed from the wells, and binding was quantified by residual GAA enzyme activity in situ (see Supplementary material for methodology). The equilibrium dissociation constant, K_D, in nM, was determined from the non-linear regression analysis of the binding curve that yielded the maximum binding (B_max). K_D is defined as the concentration of GAA that results in 0.5 B_max. The K_D values for cipaglucosidase alfa and alglucosidase alfa are 2.8 and 46.9 nM, respectively. Note that saturation was not achieved for alglucosidase alfa at the concentrations tested, and the true K_D is most likely to be higher.

Figure 5

CI-MPR-mediated uptake of cipaglucosidase alfa is more efficient than that of alglucosidase alfa in vitro. Cipaglucosidase alfa or alglucosidase alfa at 0–500 nM was incubated with fibroblasts from patients with Pompe disease (see Supplementary material for methodology). Uptake was measured by GAA enzyme activity in cell lysates after washout of unincorporated rhGAA (47). GAA activity was plotted against rhGAA concentration and fitted with a one-site binding with saturation model from which maximal uptake and K_uptake (the concentration of rhGAA that yields half-maximal uptake) were calculated by non-linear regression analysis. The shaded area represents the estimated approximate concentration of rhGAA ERT in the interstitial space available for uptake into cells (24).

Figure 6

(A) Model-predicted plasma GAA protein profiles for participants in the PROPEL trial and (B) infographic representation of rhGAA uptake from the bloodstream into muscle. (A) Total GAA protein was measured by signature peptide mass spectrometry at week 52, just before the start of GAA infusion (time 0) and at 1, 4, 6, 12, and 24 h after infusion. A population PK model is shown, with solid lines reflecting predicted median total GAA protein concentrations and shaded regions indicating 5th and 95th percentiles (47). While both drugs reached peak concentrations by approximately the end of infusion (4 h), predicted median total GAA concentrations declined at a higher rate for cipaglucosidase alfa during the distribution phase of terminal elimination, consistent with greater bis-M6P levels facilitating more efficient uptake through the CI-MPR receptor. (B) rhGAA distribution from systemic circulation to the interstitial space in muscle tissue, with cipaglucosidase alfa plus miglustat shown on the left and alglucosidase alfa shown on the right. While similar concentrations of both drugs are achieved shortly after administration in blood, the increased bis-M6P content of cipaglucosidase alfa facilitates higher-affinity interaction with the target CI-MPR receptor, leading to faster clearance and greater uptake into muscle fibers. Used with permission from AME Publishing PTE LTD: ‘Challenges in treating Pompe disease: an industry perspective’; Do et al. (24). Permission conveyed through Copyright Clearance Center, Inc.

Figure 7

Cipaglucosidase alfa undergoes full proteolytic processing in cellulo. Fibroblasts from individuals with Pompe disease were treated with 0 (untreated) or 20 nM cipaglucosidase alfa for 18 h at 37°C. After 18 h, the uptake medium was replaced with growth medium, and cells were harvested at the indicated time points (see Supplementary material for methodology). The Western blot shows analysis of rhGAA, processed over a 24 h time course, compared with precursor enzyme (lane 1). The blot was probed with a primary antibody against GAA (top) or actin (bottom). Precursor, intermediate, and mature bands are highlighted by arrows, and time points of harvest are shown above each lane.

Figure 8

Miglustat increases cipaglucosidase alfa exposure in adults with Pompe disease. Individuals with LOPD from study ATB200-02 (NCT02675465; n = 19) were treated with a single dose of 20 mg/kg intravenous cipaglucosidase alfa alone or co-administered with 130 mg or 260 mg miglustat. Blood samples were collected just before the start of infusion and at 1, 2, 3, 3.5, 4, 4.5, 5, 6, 8, 10, 12, and 24 h after infusion. (A) Total rhGAA protein levels were measured in plasma with a validated liquid chromatography–tandem mass spectrometry signature peptide assay. While similar peak rhGAA concentrations were observed with and without miglustat, miglustat increased cipaglucosidase alfa exposure in the distribution phase in a dose-dependent manner. (B) Data were analyzed with ANOVA and geometric LS mean ratios for C_max, AUC_0–t, and AUC_0–∞. Neither parameter was significantly increased for cipaglucosidase alfa with 130 mg miglustat versus cipaglucosidase alfa alone. However, the geometric LS mean ratios for AUC_0–t and AUC_0–∞ indicated that cipaglucosidase alfa with 260 mg miglustat was significantly increased from that for cipaglucosidase alfa alone, while C_max was not. (A) Adapted from Byrne et al. (46), used under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). *An upper limit of 90% CI >125% indicated a statistically significant increase from the reference treatment. ANOVA, analysis of variance; AUC_0–t, AUC from time zero to the last quantifiable time point; AUC_0–∞, AUC from time zero to infinity; CI, confidence interval; C_max, maximum concentration; LS, least-squares.

Figure 9

(A) Co-administration of miglustat with cipaglucosidase alfa improved treatment outcomes compared with vehicle control, whereas miglustat alone had no impact on glycogen reduction in the muscles of Gaa −/− mice, and (B) no reduction in LAMP1 or LC3 II was observed in Gaa −/− mice treated with miglustat alone. (A) Gaa −/− mice (15–19 weeks old; n = 8/group) received two biweekly IV administrations of vehicle, 20 mg/kg alglucosidase alfa, 20 mg/kg cipaglucosidase alfa + 10 mg/kg miglustat (administered orally 30 min prior to cipaglucosidase alfa IV injection), or 10 mg/kg miglustat alone (30). Tissues were collected 14 days after the second administration, homogenized, and assayed for glycogen content. Glycogen was significantly reduced in the quadriceps and triceps of animals treated with cipaglucosidase alfa + miglustat compared with vehicle-treated animals and those treated with miglustat alone. One-way ANOVA with Fisher’s least significant difference post hoc test: *p < 0.05; **p < 0.005; ****p < 0.00005. (B) Gaa −/− mice (15–19 weeks old; n = 8/group) received two biweekly IV administrations of vehicle, 20 mg/kg alglucosidase alfa, 20 mg/kg cipaglucosidase alfa + 10 mg/kg miglustat (administered orally 30 min prior to cipaglucosidase alfa IV injection), or 10 mg/kg miglustat alone (30). Tissues were collected 14 days after the second administration and qualitatively analyzed with immunohistochemistry. Accumulation of LAMP1-positive lysosomes is a pathological hallmark of LOPD. LC3 II localizes to autophagosome membranes, and LC3 II positive aggregates in vehicle-treated Gaa −/− mice reflect autophagic build-up. Cipaglucosidase alfa plus miglustat-treated mice displayed a reduction in LAMP1 and LC3 II signals in quadriceps compared with alglucosidase alfa treatment, and no reduction was observed in mice treated with miglustat alone. Representative images from seven or eight animals analyzed per treatment group, 200x magnification. IV, intravenous; KO, knockout; LAMP1, lysosomal-associated membrane protein 1; LC3, microtubule-associated protein light chain 3; WT, wild type.

Figure 10

Hex4 levels are improved in patients treated with cipaglucosidase alfa plus miglustat compared with alglucosidase alfa plus placebo. Hex4 is a breakdown product of glycogen and biomarker surrogate for glycogen storage metabolism. (A) Percentage CFBL in urine Hex4 in the combined PROPEL study population (ERT-experienced plus ERT-naïve patients). Week 52 data LOCF (55). (B) Hex4 (mmol/mol creatinine) CFBL calculations at week 52 performed using an ANCOVA model with LOCF, with associated LS mean difference and nominal p value. Figure adapted from Schoser et al. (55), with permission from Elsevier. ANCOVA, analysis of covariance; CFBL, change from baseline; LOCF, last observation carried forward; SD, standard deviation; SE, standard error.

Figure 11

CK levels are reduced in patients with Pome disease treated with cipaglucosidase alfa plus miglustat compared with those treated with alglucosidase alfa plus placebo. (A) Percentage CFBL in serum CK in the combined PROPEL study population (ERT-experienced plus ERT-naïve patients). Week 52 data LOCF. (B) CK (U/L) CFBL calculations at week 52 performed using an ANCOVA model with LOCF, with associated LS mean difference and nominal p-value. Figure adapted from Schoser et al. (55), with permission from Elsevier.

Figure 12

Muscle-function assessments (wire hang and grip strength) in Gaa −/− mice. Gaa −/− mice (15–19 weeks old; n = 8/group) received biweekly IV administrations of vehicle, 20 mg/kg alglucosidase alfa, 20 mg/kg cipaglucosidase alfa alone, or 20 mg/kg cipaglucosidase alfa + 10 mg/kg miglustat (administered orally 30 min prior to cipaglucosidase alfa IV injection). Muscle function was assessed once a month for 5 months, 7 days after administration of drug. The wire-hang latency assay (left) was conducted once on two consecutive days; the average of two assessments is shown. Maximum grip strength (right) was measured three times on the same day; the average of those three assessments is shown. Each point represents mean ± SE of 15 animals/group up to 3 months and eight animals/group for the remaining 3 months (seven animals sacrificed after 3 months). One-way ANOVA with Dunnett’s post hoc test. *p < 0.05 versus alglucosidase alfa. Figure adapted from Xu et al. (25), used under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/).

Figure 13

Change from baseline at week 52 of PROPEL – effect of cipaglucosidase alfa plus miglustat compared with alglucosidase alfa plus placebo in key efficacy outcomes. (A) Forest plot illustrating mean estimated treatment differences between cipaglucosidase alfa plus miglustat versus alglucosidase alfa plus placebo and corresponding 95% CIs are shown for the combined PROPEL study population for each outcome, with units as indicated on the x-axes. For all outcomes, right-sided directionality of treatment differences indicates favorable outcomes for cipaglucosidase alfa plus miglustat compared with alglucosidase alfa plus placebo. (B) The table shows baseline mean values and Week 52 CFBL values for cipaglucosidase alfa plus miglustat and alglucosidase alfa plus placebo. Shaded CFBL indicates nominally significant improvement (green) or nominally significant worsening (red) from baseline (i.e., the 95% CI does not include zero) within each treatment group. The p-values (two-tailed LS mean difference) shown in the far-right column are for the between-group treatment differences illustrated in the forest plot.

Figure 14

Summary of how the mechanism of action of cipaglucosidase alfa plus miglustat addresses the key challenges with Pompe ERTs and translates into improved outcomes for patients.