Prevalence and disease predisposition of p.A91V perforin in an aged population of European ancestry

TO THE EDITOR:

Perforin is an essential effector protein of cytotoxic T lymphocytes and natural killer cells. When released into an immunological synapse formed with a virus-infected cell, perforin binds to the target cell membrane and polymerizes into transient transmembrane pores, allowing the diffusion of proapoptotic serine proteases (granzymes) into the cytosol, where they initiate rapid apoptosis.¹ Germline mutations that impact negatively on perforin secretion or function can lead to often-fatal systemic inflammation and immune dysregulation in syndromes collectively known as familial hemophagocytic lymphohistiocytosis (FHLH).²

FHLH has been considered to be exclusively a disease of neonates and infants, but it is now widely recognized that carriers of hypomorphic mutations in FHLH-causing genes may present in adolescence or even adulthood with less prominent, but nevertheless equally life-threatening, disease manifestations.^3,4 The common PRF1 polymorphism, c.272C>T (p.A91V; rs35947132), carried by 8% to 9% of whites,⁵ has been the subject of considerable research over the last 15 years, as biallelic p.A91V inheritance, and, far less frequently, compound heterozygosity (p.A91V inherited with a much rarer loss-of-function PRF1 mutation) has been associated with HLH,^6,7 hematological malignancies,^8,9 multiple sclerosis,¹⁰ and other, less well-defined chronic inflammatory conditions¹¹ in adults, suggesting a genetic cause. Biochemical and cell biology studies revealed impaired stability and abnormal trafficking of the mutant protein, as well as its compromised function in various in vitro experimental settings.^12,13 Given the potential role of p.A91V in predisposing to these serious pathologies or to disorders of immune homoeostasis more generally, some have even suggested that broader-based population screening might be warranted.¹⁴

Here, we investigated the population prevalence of p.A91V, in both the heterozygous and homozygous state, in a sequenced cohort of over 13 000 healthy elderly individuals (the Aspirin in Reducing Events in the Elderly [ASPREE] study^15-19). Ethical approval for the genetic research analysis of the ASPREE biobank was provided by the Alfred Hospital Human Research Ethics Committee (HREC), project number 390/15, and research was conducted in accordance with the Declaration of Helsinki. The study population showed no reduction in the expected number of homozygous carriers of the mutation, and all 24 individuals were healthy at the age of >70 years. These results suggest that, in isolation, the p.A91V mutation does not confer an increased risk of cancer or serious hematological or inflammatory disease.

Severe FHLH, which occurs when an individual inherits 2 null PRF1 alleles, is a life-threatening condition and carries significant mortality, even when a diagnosis is made and the recommended therapy is given in an institution with special interest in the disease. A number of studies examining a potential role for p.A91V, including several of our own, have been conducted in relatively small cohorts of patients previously diagnosed with cancer, FHLH, or other serious inflammatory conditions, but these studies were necessarily retrospective and had low statistical power due to their size. For example, retrospective analysis of the reported cases of A91V homozygosity revealed over 70% penetrance of immunopathologies (13 of 18 reported cases of A91V homozygosity).¹ Indeed, we and others have always advised caution in generalizing the conclusions of these studies,^9,20 and considered a large, population-based prospective study to be the ideal methodology to resolve the issue.

Therefore, to explore whether p.A91V homozygosity predisposes to serious health consequences and premature mortality, we performed 2 studies at the population level. First, we interrogated the Genome Aggregation Database (gnomAD) database²¹ (Table 1), and confirmed that the prevalence of p.A91V in whites is consistent with Mendelian inheritance and follows Hardy-Weinberg equilibrium. Thus, among 102 830 evaluable whites, 7903 (7.686%) were heterozygous for p.A91V. On this basis, it was predicted that the frequency of homozygosity should be 0.137%, corresponding to 141 individuals in the population. In fact, 167 individuals (0.162%) carrying 2 copies of p.A91V mutations were found (P = .155). This was the first line of population genomic evidence that suggested p.A91V may not predispose to serious disease. It is worth noting that the A91V allele is far rarer among African, east Asian, and south Asian populations²¹; consequently, these ethnic groups were excluded from the analysis.

Table 1.

Frequency of p.A91V polymorphism in gnomAD^* and ASPREE populations

Cohort	PRF1WT/WT, N observed (%)	PRF1WT/A91V, N observed (%)	PRF1A91V/A91V, N observed (%)/ N expected (%)
gnomAD	102 830 (91.240)	7903 (7.686)	167 (0.162)/141 (0.137)
ASPREE	12 090 (92.072)	1017 (7.706)	24 (0.182)/20 (0.148)

^*gnomAD (Karczewski et al²¹).

An important caveat remained: the age of the gnomAD cohort was widely distributed. Because immune-mediated pathologies associated with hypomorphic mutations in the FHLH-causing genes can present at any age, an overrepresentation of younger people might not allow sufficient time for severe inflammatory diseases, cancer, or autoimmunity to have reduced the prevalence of pA91V homozygosity. To address this issue, we investigated the homozygous carrier rate for p.A91V in an independent cohort study (ASPREE^15-18), in which healthy Australian individuals of white background over the age of 70 years (average age, 75 years) were recruited, randomized to daily low-dose aspirin, and monitored over an average of 5 years of follow-up (until the average age of 80 years) to prospectively determine risks and benefits in the primary prevention setting. All participants of ASPREE were healthy at time of enrolment, with no current diagnosis or history of cardiovascular disease, life-threatening cancer diagnosis, or evidence of dementia.

The PRF1 gene was sequenced in 13 131 healthy elderly donors from ASPREE (Table 1). This identified 1017 heterozygous carriers of p.A91V (7.706%), and 24 study participants who were homozygous for p.A91V (0.182%). Based on the frequency of p.A91V heterozygosity and Hardy-Weinberg equilibrium, the number of homozygotes in the cohort was expected to be 20 (0.148%). As with the gnomAD study, the observed number of homozygotes in ASPREE (N = 24) exceeded the predicted number, but the variance did not reach a statistically significant level (P = .652). Examination of the detailed clinical records of ASPREE participants revealed that, by comparison with the ASPREE cohort average, carriers of biallelic p.A91V had no statistically significant increase in incidence of cancers (including hematological malignancies and melanoma that are controlled by cytotoxic T lymphocytes/natural killer cells), serious inflammatory or autoimmune disease, and death due to any cause.

In the past, several FHLH patients were found to be carriers of bigenic mutations in PRF1 (usually A91V) together with another FHLH-causing genetic abnormality. However, based on our current observations, we suggest that it is unlikely that that there is a causative link between these bigenic mutations and disease. Also, a number of patients with immune-mediated disease have been reported as PRF1 compound heterozygotes, with 1 allele being A91V and the other encoding a rare loss-of-function mutation. As various in vitro studies had demonstrated abnormal function of the A91V mutant, there is a possibility that such inheritance may predispose to immunopathology later in life (beyond infancy). However, these variants are extremely rare in the population (supplemental Table 1, available on the Blood Web site) and the ASPREE cohort does not provide an adequate setting or sufficient statistical power to address the interaction between A91V and other rare variants in PRF1. In fact, by taking the most conservative view that variants of unknown significance were all pathogenic, we identified 137 carriers of such alleles in ASPREE (1.04%). Based on this number, <3 individuals who would have coinherited A91V with any of these potentially deleterious mutations were expected to be found among our 13 131 participants. This number does not have the statistical power (at least 30 000 participants, or more than twice as many mutations would be required) to validly address the important question of whether compound heterozygosity combining A91V with another loss-of-function mutation in PRF1 can predispose to immune-mediated disease.

Although we cannot rule out rare instances to the contrary, our data strongly suggest that p.A91V is a nonpathological polymorphism whose inheritance does not affect the health or longevity of carriers, even those who inherit 2 copies. We therefore provide population-based evidence to refute an association of p.A91V homozygosity with FHLH, cancer, or other previously reported immunopathologies in whites.

Supplementary Material

The online version of this article contains a data supplement.

Authorship

Contribution: I.V. and P.L. conceived and conducted the study and wrote the paper; H.S.-I.J., S.L.F., and I.K. diagnosed and treated FHLH patients and contributed to writing the manuscript; T.F., K.T., and T.N. conducted laboratory-based experiments and contributed to writing the manuscript; M.R., R.S., and E.E.S. conducted DNA sequencing and variant and bioinformatics analyses; and J.J.M. and J.A.T. conceived the study and wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Joseph A. Trapani, Peter MacCallum Cancer Center, 305 Grattan St, Melbourne, VIC 3000, Australia; e-mail: joe.trapani@petermac.org; or John J. McNeil, Department of Epidemiology and Preventive Medicine, School of Public Health and Preventive Medicine, Monash University, 553 St Kilda Rd, Melbourne, VIC 3004, Australia; e-mail: john.mcneil@monash.edu.