Red blood cell pyruvate kinase deficiency (PKD) is an autosomal recessive disorder caused by PKLR gene mutations, resulting in nonspherocytic hemolytic anemia¹ (see S10–S12 in supplementary material). PKLR encodes the erythrocyte pyruvate kinase enzyme, a rate-limiting reaction in the glycolytic pathway essential for red-cell adenosine triphosphate (ATP) production. Homozygous or compound heterozygote PKLR variants range from those of unknown significance to missense mutations of variable severity to nonsense variants and massive deletions (see S3, S4 in supplementary material). Heterogenous clinical manifestations include severe transfusion-dependent anemia in early childhood, which may persist through adulthood or variably resolve; milder cases include incidentally discovered hemolysis with minimal or no transfusion requirements.¹

PKD therapy consists of transfusions, iron chelation, and splenectomy, which is associated with modest hemoglobin increases and reduced transfusion requirements (see S5 in supplementary material). A small-molecule allosteric enzyme activator, mitapivat, is currently under FDA review, and is associated with increased hemoglobin and reduced hemolysis in subsets of non-transfusion-dependent and transfusion-dependent adult patients² (see S6, S7 in supplementary material). Allogeneic hematopoietic stem-cell transplant has been curative in some cases, although toxicity precludes its consideration as a standard-of-care (see S8 in supplementary material). Gene therapy utilizing lentiviral transduction of autologous hematopoietic stem and progenitor cells (HSPCs) has yielded results in two initial adult patients suggesting potential for extensive correction of anemia and hemolysis³ (see S9 in supplementary material). As gene therapy is evaluated in additional adult and pediatric patients, there is a need to more extensively categorize severe PKD. Categorization has been challenging because of disease heterogeneity, but is now facilitated by PKD Natural History Study (PKD NHS) results and patient-reported outcome research. We summarize recent literature, and propose a potential expanded categorization of severe PKD to facilitate identification of PKD patients most appropriate for ongoing gene- and cell-based evaluations.

PKD-associated molecular and clinical diversity is evident from over 350 reported pathogenic mutation variants, with increases anticipated from enhanced disease awareness and diagnostic methods enabling identification of cryptic splice site or intronic variants (see S4, S10, S11 in supplementary material). Aspects of phenotype–genotype correlation have been generalizable — for example, a more frequent severe phenotype arises from nonsense or missense mutations affecting active sites or protein stability; more robust response to splenectomy is observed in patients with bi-allelic or compound heterozygote missense mutations (see S3, S13, S14 in supplementary material). Nonetheless, more severely afflicted individuals may harbor diverse mutation profiles, and substantial phenotypic variability has been observed within cohorts harboring the same mutation⁴ (see S3, S15 in supplementary material). The limited genotype–phenotype correlation (and discordant symptom severity and hemoglobin levels) might result from variably increased 2,3-diphosphoglycerate levels and additional metabolic or epigenetic polymorphisms, not yet extensively categorized. Genetic analysis — frequently vital for diagnosis and prognostic in specific instances — is not a consistently independent component of severity assessment⁵ (see S3 in supplementary material). More precise RPK protein quantification may become more clinically accessible in coming years, and would facilitate severity assessment (see S16, S17 in supplementary material).

PKD severity definitions were historically fluid; the frequently cited Zanella classification (Table 1) is not a point-based, rigid staging system, but a spectrum on which definable elements may be characterized (see S3, S18 in supplementary material). This is understandable, since PKD frequently does not represent an immediate survival risk and available therapies have been supportive. Additional PKD classifications have been posited recently, predominantly based on transfusion requirements, which may not be consistently indicative of morbidity or disease burden.^{4, 6}

PKD severity classification is also complicated because there is frequently no strong correlation between potential severity parameters. Patients may have different tolerance to similar anemia levels, transfusion requirements differ because of practice patterns; iron overload may not correlate with transfusion requirements or hemoglobin.^{1, 4} Splenectomy — a standard of care for patients with childhood transfusion requirements — confers variable benefit with incomplete correlation between patients' pre- and post-splenectomy status.⁴ The disease burden itself may alter over time, with some adults demonstrating increasing symptoms and decreased anemia tolerance with advancing age and diminished cardiopulmonary reserves.¹ A demarcated classification system has been less critical because no disease-targeted therapies were available until recently. Enhanced PKD severity categorization becomes more important with the development of allosteric enzyme activators, and particularly necessary with the introduction of gene therapy investigations.

The first-in-human multinational gene therapy trial enrolled 2 splenectomized adult patients who received gene-modified autologous CD34+ enriched HSPCs in 2020. Initial results demonstrate hemoglobin increases in both patients from baseline levels of 7.4 and 7.0 g/dL to 13.3 and 14.8 g/dL at 12 months following therapy, respectively, with concomitantly reduced hemolysis markers, including indirect bilirubin, erythropoietin, and reticulocytes. No severe adverse events related to the investigational infusion were reported as of November 2021.³ Pediatric evaluation is underway.

In these revised definitions, anemia and fatigue related thresholds were based on NCI CTCAE 5.0 grades. Iron-overload thresholds were based on criteria utilized in recent publications (adapted from β-thalassemia guidelines) with additional corroboration via hematopoietic transplant iron overload guidelines¹⁰ (see S29–S31 in supplementary material). Based on expert consensus including an advisory meeting, subsequent review and concordance involving all co-authors, the working criteria delineated in Table 1 are proposed as those indicative of severe PKD. These criteria are further categorized in Table 2 as Major (based on either transfusion-dependence or severe anemia in the presence of additional phenomena) or Additional, including those pertaining to fatigue or energy symptom burden, iron overload, or additional manifestations. As indicated in Tables 1 and 2, severity classifications to date have been based on transfusion requirements and hemoglobin levels. The proposed Additional Criteria may enhance identification of severely-afflicted patients, and may be considered for implementation in forthcoming investigations as the benefit/risk of gene therapy is further delineated.

Allosteric enzyme activators have been associated with improved hemolytic anemia in adult patients with transfusion-independent and transfusion-dependent PKD, specifically in those with at least one missense PKLR variant. In phase 2 and 3 clinical trials, mitapivat (AG-348), an oral small-molecule red-cell pyruvate kinase activator, conferred hemoglobin increases greater than 1.5 g/dL in 37% and 40% of adult non-transfusion-dependent PKD patients, respectively² (see S6 in supplementary material). In regularly transfused patients, a ≥33% reduction in transfusion burden was reported in 37% and transfusion independence was observed in 22% over 24-week follow-up (see S7 in supplementary material). Based on this agent's evolving role in adults (and potential pediatric investigation), severity definitions include the terminology “despite available medical therapy” in order to account for this agent's likely availability.

Definitions of severity in PKD are further conditioned by splenectomy status. Although the proposed classification incorporates aspects of clinical PKD in both pre- and post-splenectomy settings, the scenarios provided are most relevant to splenectomized children and adults. Splenectomy is currently a standard-of-care and is frequently performed during childhood (at age 4–5 years or older) for patients with significant anemia or transfusion requirements. Splenectomy is associated with an average 1.6 g/dL hemoglobin increase in PKD and with frequent decreased transfusion requirements.⁴ Reductions in fatigue or other energy-related symptoms may be disproportionately higher than the relatively modest hemoglobin increases; this may result from diminished hemolysis and concomitant hematopoietic and metabolic requirements. Factors associated with persistent post-splenectomy hemoglobin less than 8 g/dL include pre-splenectomy levels less than 7 g/dL, higher pre-splenectomy bilirubin, and non-Amish ethnicity.⁴ Splenectomy does not ameliorate the underlying disorder or concomitant iron overload, and is associated with a risk of significant infectious and thrombotic complications.

If the efficacy and safety demonstrated over 1-year in initial adult patients is replicated in the pediatric setting, evaluation of gene therapy in non-splenectomized patients may be warranted. This will require judicious determination of the pre-splenectomy population with the highest likelihood of ongoing severe PKD manifestations.

During recent years, the PKD NHS and additional research efforts have enabled improved understanding of the complex and varied PKD phenotypes. Although transfusion-dependence and anemia parameters factor substantively into severity assessment, no single parameter allows consistent identification of patients most extensively afflicted. Optimal investigation of PKD-focused therapies necessitates severity classification encompassing the entirety of the disorder, including the multiple components that impair patient well-being. We therefore developed this proposed severity classification to provide a broader potential definition of the most severely-afflicted PKD population. This is a narrative expert proposal and limitations include a methodology that did not incorporate formal development guidelines for expert-consensus recommendations (see S32 in supplementary material); we anticipate a subsequent and more exhaustive methodology involving a broader range of disease and therapeutic-area experts as additional adult and pediatric results become available. It is likely that the classification system presently proposed will undergo refinement guided by additional natural history and interventional study results.