Natural killer (NK) cell function in paroxysmal nocturnal hemoglobinuria: a deficiency of NK cells, but not an NK cell deficiency

To the editor:

Treatment of the glycosylphosphatidylinositol (GPI) anchor deficiency paroxysmal nocturnal hemoglobinuria (PNH) has been revolutionized by the use of the anti-C5 antibody eculizumab, which blocks complement-mediated hemolysis and the associated pathology.^1,2  In addition to complement susceptibility, GPI anchor deficiency alters cellular function with the potential to further contribute to disease. For example, defective natural killer (NK) cell activity in PNH was first described >3 decades ago.^3,4  NK cell deficiencies are associated with susceptibility to infection,^5,6  suggesting that NK function in PNH should be analyzed in more detail. Here we show that functional defects in NK cell activity in PNH result from reduced NK cell numbers rather than cell intrinsic defects.

Early PNH studies showed that impaired NK cell activity was associated with reduced large granular lymphocyte (LGL) counts.^3,4  However, these reports preceded the definition of NK cells and our understanding of the molecular basis of PNH, prompting reassessment. Mosaicism in PNH¹  allows side-by-side functional comparisons of GPI⁺ and GPI^neg NK cells within individual patients, enabling the assessment of NK cell activity on a per cell basis (Figure 1; supplemental Figure 1, available on the Blood Web site). Despite reports of impaired activity,^3,4  the GPI-deficient NK cells were proficient at target cell-induced granule exocytosis (Figure 1A-B; supplemental Figure 1). Thus, early findings associating reduced NK cell activity with reduced LGL numbers rather than intrinsic cellular activity are correct.⁴  The absolute number of NK cells (and more variably, other lymphocytes) is indeed reduced in PNH⁷ ; in our cohort of 39 patients, two thirds had NK cell counts below the reference range (Figure 1C), and NK cell numbers were not significantly correlated with neutrophil, monocyte, or platelet counts (supplemental Figure 2). The basis for reduced NK cell numbers in PNH is unclear, although this might be related to impaired chemotactic or homeostatic mechanisms, as we recently reported.⁸  Although the activity of GPI-deficient NK cells is unimpaired, a reduction in absolute numbers of NK cells will reduce NK cell activity in the blood as a whole.

Clearly, PNH should not be classified as a functional NK cell deficiency (NKD). Classical NKD is characterized by ∼1/10 the normal number of NK cells, and counts in most of our PNH patients exceeded this (Figure 1C). Furthermore, the term NKD is reserved for where “the impact upon NK cells need represent the major immunological abnormality in the patient.”⁶  In PNH, all hematopoietic lineages are affected because of the presence of PIGA mutations in hematopoietic stem cells.¹  More compelling is the clinical phenotype; the defining feature of NKD is the heightened susceptibility to viruses,^5,6  which has not been observed in PNH.^7,9,10  Instead, infection in PNH is bacterial in origin¹⁰  and likely to be associated with neutropenia secondary to underlying bone marrow failure or associated with use of eculizumab, which increases the risk of infection with encapsulated bacteria normally eliminated by terminal complement components.¹  In summary, the low numbers of NK cells in PNH affect overall cytotoxicity, but this defect is not severe enough to manifest as heightened susceptibility to viral infection as seen in NKD.