GATA2 at 14: genotype-phenotype correlations

Abstract

GATA2 mutations cause adult-onset bone-marrow failure characterized by cytopenias, infections and increased risk of myeloid malignancy. We reviewed hospital records and referrals of 232 GATA2-mutated individuals from 122 families to gather hematopoietic and syndromic features. Mutations were categorized by their effect on GATA2 protein: mutation after the second zinc-finger (C-term, N=10); missense in the second zinc-finger (ZF2, N=104); mutations producing a stable truncated protein (Truncation, N=22); Null alleles (mRNA instability, large deletions, N=46); or Enhancer mutations (N=50). Regression models for symptom onset identified earlier onset and increased hazard ratios (HR) between Truncation, Null or ZF2 mutations (13 years, HR=5.00; 17 years, HR=3.60; and 22 years, HR=2.23, respectively) and Enhancer mutations. Commonly mutated ZF2 amino acids also stratified patients: those with R396 or T354 presented earlier with increased hazard ratios (16 years, HR=2.96 and 19 years, HR=2.16, respectively) compared to those with R398 (34 years). Mutation groups with a median onset before 20 years had higher prevalences of cytopenias, infections, human papilloma virus (HPV), thromboses and hearing loss. Warts/anogenital HPV were common (93/232, 40.1%). Non-hematologic manifestations included hearing loss (45/232; 19.4%), lymphedema (30/232; 12.8%), and non-hematologic malignancies (46/232, 19.8%), the majority being HPV-related anogenital cancers. Leukemia or lymphoma occurred in 25/232 patients (10.8%). Functional analysis of patients’ mutations showed impaired transactivation. Males had more monosomy 7 (17/73 males, 6/110 females, P=0.0016). Non-tuberculous mycobacterial infection was the presenting symptom in 14/73 males versus 7/110 females (P=0.0151). GATA2 mutation type and specific mutations affect penetrance and expression. Sex is an independent risk for specific infectious presentations and cytogenetic abnormalities. ClinicalTrials.gov identifier: NCT01905826.

Introduction

GATA2 deficiency was described phenotypically in 2010 as a constellation of symptoms including infections, monocyte, B, and natural killer (NK) cell cytopenias, and pulmonary manifestations with sporadic or dominant inheritance.¹ The following year the hematopoietic stem-cell transcription factor, GATA2, was identified as the causative gene² with phenotype extended to familial leukemia,³ loss of dendritic cells,⁴ lymphedema⁵ and myelodysplastic syndrome (MDS).⁶ Non-hematopoietic findings included deafness and thrombotic events.^7-11

In the 14 years since GATA2 deficiency was identified, phenotypic characterization has expanded with GATA2 mutations identified in registries of neutropenia⁷ and pediatric¹⁰ and adult¹¹ MDS. While early studies focused on unifying phenotypes, which defined the typical patients with GATA2 deficiency seen in the clinic, more recent studies have catalogued somatic mutations affecting bone marrow hypoplasia, MDS and leukemia. A survey of 400 published GATA2 mutations did not correlate clinical presentation with underlying mutation.¹² To address issues of penetrance and expression we took advantage of our large, longitudinal single-center cohort.

Methods

Patients referred to the National Institutes of Health (NIH) for genetic diagnosis or clinical management provided written informed consent prior to enrolling in Instituional Review Board-approved protocols. Chart review was performed for all records at NIH and from referring clinicians through June, 2020. Detailed descriptions of experimental methods used are provided in the Online Supplementary Appendix.

Results

Patients

Our cohort of patients consisted of 124 probands from 122 families, including two pairs of identical twins, referred to the NIH for genetic diagnostics or clinical management. All patients were confirmed to carry GATA2 mutations; one patient had a full gene deletion and one had uniallelic GATA2 expression (patient 23.I.3 previously reported^{9,1 3}). The majority of mutations overlap with the 480 mutations in the literature recently reported with associated American College of Medical Genetics and Genomics (ACMG) variant criteria.¹² At-risk relatives of mutation-positive individuals, including all first-degree relatives or surviving descendants, were offered genetic testing. This led to the identification of 108 additional mutation-positive individuals totaling 232 individuals. Chart review and clinical history were used to identify initial symptoms, age at presentation, and additional clinical findings. There was a broad age range for initial symptom presentation (<1 year to 85 years); similarly, the initial presenting symptoms were varied, spanning hematopoietic and non-hematopoietic findings as well as infections (Online Supplementary Figure S1). Of the 232 individuals, 104 have not undergone hematopoietic stem cell transplantation (HSCT), 73 are post-HSCT, and 55 have died, 18 of whom underwent HSCT (Online Supplementary Tables S1 and S2; Online Supplementary Figure S2).

Classification of GATA2 mutations

GATA2 mutations are frequently private to a family, although several recurrent mutations occur at CpG dinucleotides, which undergo C>T transitions after demethylation. As a result, these mutations may arise sporadically and are not part of a larger population haplotype. Supporting the lack of a GATA2 population haplotype, among 52 sets of parents of probands available for genetic testing, 21 probands had mutation-negative parents (21/52, 40.4%) including four with R396 and one with R398 mutations.

We classified the mutations’ predicted effect on the GATA2 protein (Figure 1A) and grouped patients accordingly. Mutations in the intronic enhancer element (N=50) reduce transcription of the mutant allele by ~60%. However, these transcripts encode wild-type (WT) protein,¹³ yielding a decreased total amount of normal GATA2 protein. Since expression occurs from both alleles, the total GATA2 level should still be ~70% of WT. Null mutations (N=46) included full-gene or large, partial-gene deletions and premature stop codons. Splicing or frameshift mutations causing nonsense-mediated decay of the transcript result in loss of expression from the mutant allele; therefore, they are also classified as Null. These mutations have transcript and protein only from the intact WT allele, causing haploinsufficiency.¹³ One family in our cohort carried a synonymous mutation at the last base of exon 5, c.1017G>T, p.L339L, which altered proper splicing, similar to other reported patients.¹⁴ Transcript analysis demonstrated utilization of a cryptic intronic splice site causing a 16-base insertion. Since the mutant transcript had significantly lower expression than the WT allele (Online Supplementary Figure S3), the transcript was determined to be unstable, placing this family in the Null category. The remainder of the mutations are described with respect to the critical second zinc finger (ZF2), which binds DNA during transcriptional activation of target genes. Truncations (N=22) include stop codons, splice mutations or frameshifts occurring late enough in the transcript to prevent nonsense-mediated decay, having stable mRNA and the potential for a protein lacking some or all of the ZF2. Truncation could also remove the nuclear localization signal which occurs after the ZF2. Since there is evidence for chromatin binding even in the absence of the nuclear localization signal,¹⁵ we treat Null and Truncation mutations as separate categories for this analysis. Two cases (c.1018-1G>A; c.1018-50_1143+247del) in which cDNA analysis demonstrated in-frame exon skipping deleting most of the ZF2 are classified as Truncations due to the loss of the majority of ZF2. Missense changes or small, in-frame deletions wholly contained within the ZF2 are grouped together as ZF2 mutations (N=104). Lastly, C-terminal mutations (C-term, N=10) include those missense or frameshift mutations occurring after the ZF2 region. The mechanism by which these C-term mutations alter GATA2 function is unresolved; however, there are two families with multiple individuals carrying C-term mutations and phenotypes consistent with GATA2 deficiency. These last three mutation categories (Truncation, ZF2 and C-term) produce stable mRNA and would be predicted to have roughly half WT and half mutant GATA2 protein. Unique mutations from the cohort and the associated functional classifications are listed in Table 1 and shown in Figure 1B. Aggregating patients by mutation category allowed group comparisons. We observed a striking difference between penetrance in the Null and Truncation patients (95.5% each) compared to the ZF2 and Ehancer patients (81.7% and 54%, respectively) (Table 2). Age variation between symptomatic and asymptomatic individuals did not account for the presence of symptoms; the asymptomatic group had a higher median age (44.6 years) than the symptomatic group (21.0 years) providing sufficient time for phenotypes to develop (Online Supplementary Figure S4A). Comparing across mutation groups, the median age of asymptomatic individuals is the same as or higher than that of the symptomatic patients (Online Supplementary Figure S4B). The C-term mutations showed an 80% penetrance, with the majority of individuals with these mutations coming from a single family.^16,17 Additional genetic factors within this family cannot be ruled out as a cause of the altered penetrance observed.

Next, we examined differences in age at first reported symptom by mutation category. For this analysis, we considered clinical presentations sufficiently abnormal to be treated or noted in a chart and excluded minor signs found during annual surveillance. Symptoms included both immune/hematologic manifestations (cytopenias, infections, abnormal bone marrow, hematologic malignancies) as well as non-hematologic manifestations (solid tumors, lymphedema, hearing issues, erythemanodosum). Kaplan-Meier estimates for symptom onset showed clear differences across mutation categories, consistent with the observed differences in penetrance (Figure 2A, Online Supplementary Figure S5). The most penetrant mutation categories also had the lowest median ages at symptom onset: Truncation (13 years), Null (17 years), ZF2 (22 years), C-term (31 years), and Enhancer mutations (45 years, with 35% of Enhancer mutation-positive individuals remaining asymptomatic in to their 7^th decade). Using a mixed-model Cox regression analysis, which accounted for relatedness among patients, we compared patients with Enhancer mutations to those with the other categories of mutation. Significant hazard ratios (HR) were found for Truncation (HR=5.63, corrected P<0.001), Null (HR=3.55, corrected P<0.001) and ZF2 mutations (HR=2.24, corrected P=0.006) (Figure 2B). Pairwise comparisons across all mutation categories showed significantly increased hazard ratios between those with Truncation and ZF2 mutations (HR=2.52, P=0.0487), while statistical significance was not reached for any of the remaining categories (Online Supplementary Table S2). No statistically significant differences were observed between patients with Null and Truncation mutations.

Figure 1.GATA2 cohort mutation classifications and locations. (A) Mutations were grouped into categories based on the effect of the mutation on mRNA (Enhancer and Null mutations) or GATA2 protein (Truncation, ZF2, and C-term). The wild-type allele is depicted at the top with amino acid numbers of defined domains. Enhancer mutations cause reduced levels of wild-type transcript and protein from the mutant allele. ZF2 mutations are small missense or in-frame deletions wholly within ZF2. Truncation mutations are predicted to have a stable protein but delete all or part of ZF2. C-term mutations occur after ZF2. Null mutations lack transcript and protein from the mutant allele. Predicted GATA2 functional level from mutant allele (left) or mutant plus wildtype alleles (right). (B) Mutation diagram. The genomic GATA2 organization is shown at the top, the linear protein in the middle, and an enlargement of ZF2 at the bottom. Mutations are shaped by mutation type and color coded by mutation classification. Numbers within symbols indicate independent probands with the mutation. AA: amino acid; #: number; ZF: zinc finger; C-term: C-terminal; WT: wild-type; indel fs: framesirt insertions/deletions; exp: exposure; del: deletion.

Table 1.GATA2 mutations and classifications.

Table 2.Penetrance by mutation class or amino acid.

Given the large number of ZF2 mutation-positive individuals, we identified four recurrently mutated sites, T354, R361, R396, and R398. There were 17 individuals with mutations at T354 (T354M N=15, 8 probands; T354K N=2, 1 proband); 14 with R361 mutations (R361C N=5, 3 probands; R361H N=5, 4 probands; R361P N=3, 1 proband; R361delRNAN N=1); 18 with R396 mutations (R396Q N=11, 6 probands; R396W N=6, 5 probands; R396G N=1 proband); and 33 with R398 mutations (R398W N=31, 12 probands; 1 proband each with R398L and R398Q) (Online Supplementary Figure S6). Kaplan-Meier estimates for age at symptom onset grouped by affected amino acid (Figure 1C, Online Supplementary Figure S7) showed striking differences among the groups: those with R396 and T354 mutations presented in the second decade (16 and 19 years, respectively) reminiscent of the Truncation and Null mutation groups described above. In contrast, patients with R361 and R398 mutations had later median onsets (22 and 34 years, respectively). Applying a mixed-effect Cox regression analysis comparing patients with R396, T354 or R361 to those with R398 demonstrated significantly increased hazard ratios for those with R396 and T354 mutations (HR=2.67, corrected P=0.004; HR=2.16, corrected P=0.028, respectively). There was not a significant difference between patients with the R361 mutation and those with a R398 mutation (HR=1.63, P=0.186) (Figure 2B).

Penetrance across recurrently mutated amino acids also mirrored symptom onset: 100% of individuals carrying R396 mutations had some disease manifestations while only 24/33 (72.7%) of R398 mutation carriers were symptomatic (Table 2). In fact, across all mutation groups there was a strong, negative correlation between median age at first symptom onset and penetrance (R²=0.8908, P=0.0001) (Figure 2D).

Figure 2.Differential age of symptom onset by GATA2 mutation. (A) Kaplan-Meier plot of symptom onset (age at presenting symptom) across mutation groups. The table below indicates number of at-risk individuals at each 5-year age increment. (B) Hazard ratios of mutation groups compared to Enhancer (top) or recurrently mutated amino acids compared to R398 (bottom). (C) Kaplan-Meier plot of symptom onset for patients with mutations at one of four recurrently mutated ZF2 amino acids. The table below indicates number of at-risk individuals at each 5-year age increment. (D) Correlation between median age of symptom onset and penetrance by mutation category or recurrently mutated amino acid. (R² =0.8908, P=0.0001: P value and R² calculated using simple linear regression). C-term: C-terminal mutation; ZF: zinc finger.

Functional assessment of GATA2 mutations

To assess the transactivation potential of GATA2 we transfected a luciferase GATA2-reporter construct¹³ with decreasing amounts of GATA2-WT (Figure 3A) demonstrating a clear dose-dependent response to GATA2 levels. Reduction of GATA2-WT by 25%, similar to that caused by Enhancer mutations, produced significantly less luciferase than did WT (Figure 3A). Similarly, transfecting only 50%, akin to haploinsufficiency, further decreased luciferase production (Figure 3A). Several patients’ variants were analyzed in the same system (Figure 3B). While the somatic, activating mutation L359V induced higher luciferase, all of the ZF2 mutations showed decreased transactivation compared to the WT. The C-term mutation, S447R was not significantly different from WT (Figure 3B). Three rare variants selected from gnomAD, had normal (T347N, A364T) or slightly increased (G392R) activity.

Infectious phenotypes

The most common presenting symptom or sign within our cohort was infection (81/183 symptomatic individuals, 44.3%); cytopenias were the next most frequent (47/183, 25.7%) (Online Supplementary Table S3). However, over the course of disease, infections were observed in the majority (145/183 symptomatic individuals, 79.2%) while cytopenias were even more prevalent (161/183, 88.0%). As patients lost NK cells, viral infections became more pronounced.¹⁸ Within our cohort there were 114 symptomatic individuals with viral infections (114/183, 62.3%). Genital or extragenital human papilloma virus (HPV) infections, including cutaneous warts, were seen in 94 patients; 22 patients had Epstein-Barr virus, including three with chronic Epstein-Barr virus infections; 19 had herpes simplex, while nine had cytomegalovirus infections. Additional viral infections cited in records included H1N1 influenza (N=2), varicella-zoster (N=8), measles (N=2) and one case each of mumps, respiratory syncytial virus, influenza (other than H1N1) and BK virus. Bacterial infections were seen in 74 patients (74/183, 40.4%) and were predominantly non-tuberculous mycobacteria (NTM, 53/183, 29.0%), as well as staphylococcal and streptococcal infections in five patients each (2.7%). Fungal infections were seen in 44 patients (24.0%), most commonly Candida (N=18) and Aspergillus (N=12). Histoplasmosis (N=6), blastomycosis (N=2) and tinea (N=3) were also seen. Of the 145 patients with infections, 136 had documented cytopenias (93.8%) with the remaining nine individuals being relatives with no lymphocyte phenotyping available. These findings are consistent with previous reports of co-occurrence of cytopenias and major complications of GATA2 deficiency.⁹

Hematologic phenotypes

In 2016 GATA2 deficiency was recognized as a separate entity within myeloid neoplasms by the World Health Organization.¹⁹ In view of the high rate of myeloid dysfunction and leukemia, we routinely conduct bone marrow surveillance on mutation-positive individuals. Of 167 patients with bone marrow analysis, some abnormality was reported in 142 (142/167, 85.0%) including hypocellularity, atypical megakaryocytes, absent hematogones or cytopenias. MDS was documented in 102 patients (102/167, 61.1%), 116 of whom had bone marrow samples available for in-house review by a hematopathologist. MDS-multilineage dysplasia was the predominant classification (N=43) followed by MDS-with excess blasts 1 and MDS-single lineage dysplasia (4 each), MDS-myeloproliferative neoplasm (N=3), and MDS-unclassifiable, refractory cytopenia in childhood, and chronic myelomonocytic leukemia (2 each). Within the cohort, 26 patients had documented leukemia or lymphoma, some first identified during asymptomatic periods.

Figure 3.GATA2 transactivation is dose-dependent and decreased by the patients’ mutations. (A) Effect of decreasing amounts of wild-type (WT) GATA2 on luciferase production in transfected HEK cells. Differing amounts of WT GATA were transfected into HEK cells. After 20 hours, cells were lysed and renilla-normalized luciferase activity was measured. (B) GATA2-induced luciferase by empty vector (gray), somatic, activating (L359V red), patient (blue), or gnomAD (white) variants normalized to WT. Data are the results of 5-13 independent experiments of each variant; within each experiment, samples were tested in triplicate and the results were averaged. P values were calculated using an unpaired t test comparing to WT GATA2 500 ng (A) or Kruskal-Wallis with Dunn’s multiple comparison test (B). *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001. Error bars represent mean + standard error of mean. Luc: luciferase; Ren: renilla; EV: empty vector.

Non-hematopoietic phenotypes of GATA2 deficiency

Lymphedema with hearing loss and infections, first described as Emberger syndrome, was recognized to be caused by GATA2 mutations in 2011.⁵ PROX1 is a critical transcription factor for lymphatic development with Prox1 haploinsufficiency causing disruption of lymphovenous valves.²⁰ While it was suggested that GATA2 null mutations were more common in lymphatic dysfunction and lymphedema,^9,21 the Emberger cohort missense mutations (R361L, C373R, R396Q) display reduced binding to the PROX1 -11 kb element and significantly reduced DNA-binding and transactivation of the PROX1 +4.5 regulatory region.²² Twenty-nine individuals in our cohort (15.8% of 183 symptomatic individuals) exhibited some degree of lymphedema: ten patients from the Null group, three from the group with Truncation mutations, four from the group with Enhancer mutations, 11 from those with ZF2 mutations and one from the group with C-term mutations. Of the 11 individuals with lymphedema and missense mutations, six carried mutations demonstrated to affect PROX1 promoter binding²² (4 R361, 1 C373, and 1 R396).

Hearing loss in patients with GATA2 deficiency was first described as part of Emberger syndrome in 2011 ⁵ followed by identification in a broader group of GATA2-mutated patients in 2014,⁹ in whom it was hypothesized to be related to aminoglycoside exposure. There were 45 individuals within our cohort with documented hearing loss (45/232, 19.4%); eight cases documented by the age of 3 years (8/232, 3.4%), and an additional three by the age of 10 (11/232, 4.8%), an order of magnitude larger than the Center for Disease Control and Prevention estimate of 5-7/1000 children.²³ Hearing loss after amikacin exposure is recognized but is dependent on dosage and duration. As a result, we separated hearing loss by documented or suspected (after NTM infection) amikacin treatment. Congenital or bilateral hearing loss before onset of infections was seen in eight individuals, while an additional 20 had sensorineural or conductive hearing loss or tinnitus prior to potential aminoglycoside exposure (28/232, 12.1%). An additional 17 (17/232, 7.4%) had been treated for NTM or Pseudomonas, with the onset of hearing loss after infection, possibly compatible with antibiotic exposure. These findings support a critical role for GATA2 in the development and maintenance of the inner ear. In fact, dominant GATA3 mutations lead to hypothyroidism, deafness and renal insufficiency syndrome.²⁴ There is clear overlap between Gata2 and Gata3 expression in the inner ear.²⁵ Examination of the Encyclopedia of DNA Elements (ENCODE) (encodeproject.org) track on the UCSC genome browser (genome.ucsc.edu) identified GATA2 binding to an intronic enhancer element within GATA3 (Online Supplementary Figure S8A), suggesting direct regulation of GATA3 by GATA2. Additionally, GATA2 binds to both the promoter and an intronic enhancer within GDF6 (Online Supplementary Figure S8B) which, when deleted, disrupts cochlear development.²⁶ Together, these data suggest that some GATA2 transcriptional targets are critical for development and/or maintenance of the inner ear and may account for the hearing impairment seen with GATA2 deficiency.

Additional non-hematopoietic phenotypes included thrombotic events, identified in 15 individuals, eight of whom had more than one incident, including deep vein thromboses, pulmonary embolism, stroke, renal infarct and portal vein thrombosis. Nine women reported miscarriages, six having had more than one.

Non-hematologic malignancies

Non-hematologic malignancies occurred in 46/232 members of the cohort (19.8%) including five individuals with two separate non-hematologic malignancies. Of these, the vast majority (34/46, 73.9%) had genital dysplasia or cancer. In addition, 12/46 (26.1%) had squamous cell carcinoma involving the skin (N=5), oral mucosa (N=2), or both (N=1). Additionally, 4/46 (8.7%) had breast cancer, 4/46 (8.7%) had basal cell carcinomas, 2/46 (4.3%) had renal cancer, and we found one case each of lung cancer, desmoid tumor, leiomyosarcoma, metastatic melanoma and pancreatic cancer. The high rate of genital dysplasia/cancer and prevalence of squamous cell carcinoma is consistent with the high rate of HPV infection seen in patients with GATA2 deficiency.

To determine whether the rate of non-hematologic malignancies observed was disproportionately high, we compared the incidence in our cohort with that in the USA reported in the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) database (seer.cancer.gov) for 2020. We grouped our cohort into the same 5-year age brackets used by SEER and compared the proportion of individuals in each interval with the specific malignancy by age. Since leukemic transformation is common in cases of GATA2 deficiency, we first compared rates of leukemia between the two groups. As expected, we observed a much higher proportion of leukemia among our cohort (Figure 4A). Next, we compared the presence of any malignancy, leukemia or non-hematopoietic. Consistent with the increased rate of leukemia and our observation that nearly 20% of patients with GATA2 deficiency were diagnosed with a non-hematopoietic malignancy, the GATA2 cohort demonstrated a higher proportion of malignancies than the general population (Figure 4B). Lastly, we examined the recurrent malignancies - HPV-related anogenital and breast cancer in women (Figure 4C) and anogenital cancers in men (Figure 4D), again finding a much higher proportion among patients with GATA2 deficiency than in the general population. Across all comparisons, there appears to be an earlier age of onset among GATA2-mutated patients, especially across the HPV-related anogenital malignancies. Together these findings reveal a previously unappreciated increased rate of non-hematopoietic malignancies among the GATA2 cohort and possibly an earlier age of onset as well, although larger cohort studies would be needed to confirm this.

Genotype-phenotype correlations

Examination across the assembled cohort of GATA2 mutation-positive individuals showed a median age of symptom onset of 20 years (Online Supplementary Figure S9). We then divided the mutation groups between those with a median symptom onset <20 years (“early”, N=102, including those with Null and Truncation mutations, and the recurrent ZF2 mutations at R396 and T354), or >20 years (“late”, N=107, including those with recurrent ZF2 mutations at R361, R398, C-term and Enhancer mutations).

We performed comparisons between the two groups for the presence and/or age of specific phenotype onset. This analysis found significant differences between the groups for events which occur early in GATA2 deficiency: cytopenias, infections, HPV (including warts), thrombotic events, hearing loss, and hearing loss before documented exposure to antibiotics. In each case these phenotypes had a higher prevalence in the early onset group (Figure 5A). No significant differences in prevalence were observed for MDS, leukemia/lymphoma or non-hematologic malignancies between early and late onset groups. Both cytopenias and MDS occurred at significantly younger ages in the early group than in the late group (cytopenias - 18 years vs. 28.5 years, respectively, P=0.0052; MDS -20 years vs. 31 years, respectively, P=0.0098) (Figure 5B); there was no difference in age of first infection or onset of lymphedema.

Thirty-eight patients had symptom onset recorded before the age of 10 years, 25 of whom (65.8%) carried one of the “early” mutations classified above. The most frequent early-onset symptom was viral infection, usually warts (15/38 patients, 39.5%, one of whom had hearing loss as well). Each of those presenting in childhood with impaired hearing was diagnosed by the age of 3 years and carried a Truncation mutation (5/5). It is noteworthy that four of five patients presenting with lymphedema before the age of 10 had Null mutations; together these findings suggest effects of mutation type on specific pathways driving phenotypes. Larger cohorts will be required to validate these observations.

Figure 4.Increased prevalence of malignancies among the cohort with GATA2 deficiency. (A-D) Patients were grouped by age into 5-year bins and the proportion of individuals with the specific malignancy was plotted for each age (pink) compared to the incidence in the Surveillance, Epidemiology, and End Results (SEER) database (blue). (A) Leukemia. (B) Any malignancy. (C) Female anogenital malignancy (filled circles) or breast cancer (open circles). (D) Male anogenital malignancy.

Sex and GATA2 deficiency

We next asked whether there were differences between age at symptom onset by sex. Examination of the 141 females (60.8%) and 91 males (39.2%) in our cohort found no sex difference in age at symptom onset (Figure 6A). Evaluation of presenting symptoms across 183 symptomatic individuals (110 females, 73 males) showed similar percentages of hematologic (40.9% vs. 41.0%), non-hematologic (12.7% vs. 17.8%) and infectious manifestations (46.4% vs. 41.0%) between sexes. However, NTM infections were the presenting symptom among males 3-fold more than among females (14/73, 19.2% vs. 7/110 6.4%, P=0.0096) (Figure 6B). Conversely, females had almost double the infections other than NTM as their presenting symptoms compared to males (40.0% vs. 21.9% P=0.0155) (Figure 6B), with HPV and other viral pathogens predominating (Figure 6C).

During disease progression, bone marrow stress may lead to clonal hematopoiesis and the acquisition of somatic mutations and cytogenetic abnormalities.^27,28 Among the 183 symptomatic patients, 23/72 (31.9%) males and 29/111 (26.1%) females developed gross chromosomal abnormalities. Consistent with previous reports on GATA2 deficiency⁶ and MDS in general,²⁸ isolated monosomy 7 (-7) (N=15) and trisomy 8 (+8) (N=23) were the most frequent cytogenetic variants. We also observed -7/+8 together (N=5) and +8 with +1 der(1;7)(q10;p10) (N=3) as well as additional isolated cytogenetic variants (Online Supplementary Table S4). Among patients exhibiting symptoms consistent with GATA2 deficiency, 17/72 males (23.6%) developed loss of chromosome 7 (isolated -7, +1der(1;7)(q10;p10), -7/+8) compared to 6/111 females (5.4%, P=0.0004, Fisher exact test). In contrast, the majority of individuals with chromosome 8 abnormalities (+8; +8, +1(der); +8, -11) were female (20/111 vs. 5/72) (P=0.0460, Fisher exact test).

Discussion

Most previous studies in GATA2 deficiency have focused on patients with later-stage GATA2 deficiency,^7,9-11,21,29 successfully describing disease manifestations. Importantly, patients with normal bone marrow lack somatic mutations leading to disease progression including MDS and leukemic transformation.^30-32 Our proactive search for, and inclusion of, all GATA2-mutated relatives allowed us to more fully explore penetrance and genotype/phenotype correlations prior to somatic mutations or malignant transformation. In order to facilitate comparison, we classified mutations by reported residual GATA2 activity. Null and Truncation mutations had the earliest onset and highest penetrance. Enhancer mutations, which reduce but do not eliminate WT GATA2 protein from the mutant allele, were the least penetrant, with a median age of symptom onset of 45 years and a significant proportion remaining healthy into their seventh decade. The ZF2 missense mutations showed differences among the four recurrently mutated amino acids, with patients with R396 and T354 having median ages of symptom onset similar to those of patients with Truncation and Null mutations (16 and 19 years vs. 13 and 17 years, respectively). In contrast, R361 and R398 mutations had median ages of onset a decade later (22 and 34 years, respectively).

Figure 5.Increased prevalence of specific phenotypes among patients with mutations manifesting early. (A) Mutation categories or recurrently mutated amino acids were grouped by median age of symptom onset being <20 years (“Early”) or >20 years (“Late”). Specific GATA2-related phenotypes occur more frequently in the “Early” mutation group compared to the “Late” group, shown in red. (B) Age of onset of specific phenotypes among patients in the “Early” versus “Late” mutation groups. Cytopenias and myelodysplastic syndrome appear earlier in “Early” patients (P=0.0052 and P=0.0098, respectively). No difference was observed in age of first infection or occurrence of lymphedema. Error bars – median with 95% confidence interval; q values calculated using the Mann-Whitney test with the two-stage linear step-up procedure of Benjamini, Krieger, and Yekutieli, false discovery rate <0.05. HPV: human papillomavirus; MDS: myelodysplastic syndrome; Adj: adjusted; yrs: years; NS: not statistically significant. **P<0.01.

Dickinson et al reported frameshift mutations associated with earlier presentation than substitution mutations.⁸ In their cohort of 30 patients, ten had early frameshifts predicted to cause nonsense-mediated decay, nine carried R398 mutations while 11 carried other ZF2 mutations. The median ages of presentation across their three groups were 17.6, 34 and 21.6 years, respectively. These are closely aligned with our findings of 17 years for patients with Null mutations and 35 years for those with R398 mutations.

Non-hematologic phenotypes reported in other cohorts were similar to ours. Lymphedema (23%) and deafness (9%) were described in a GATA2-mutated, pediatric MDS cohort.¹⁰ A French and Belgian cohort^11,32 reported lymphedema in 15% but only one case of deafness (1.3%). In a follow-up of the same group plus patients from the UK,³² lymphedema occurred in 15.8% but deafness was not independently reported. In comparison, among our symptomatic patients, 15.9% had lymphedema whereas hearing loss occurred in 24.5%. While the incidence of lymphedema was consistent across cohorts, the differences in the prevalence of deafness may reflect the ages at which the patients were studied, as we had several cases of adult-onset deafness; our median age at presentation or symptom-free assessment was 19 years. Furthermore, we included later-onset hearing impairment, while the other two groups only reported congenital deafness. We also employed dedicated audiology assessment for all subjects regardless of reported symptoms, which likely increased sensitivity for mild hearing impairment.

Until now, no systematic review of non-hematologic malignancies in patients with GATA2 deficiency has been reported. We identified a significant risk for HPV-driven anogenital cancers and an increased risk of breast cancers, in addition to the previously known risk of leukemic transformation in GATA2-deficient patients. Whether this reflects impaired NK-cell surveillance, impaired T-cell surveillance, or some other immune or non-immune defect remains to be seen. The very favorable response of anogenital HPV disease to HSCT suggests that restoration of immune function is ameliorative, at least for those malignancies. Given the effectiveness of the HPV vaccine, we typically vaccinate early, prior to exposure to the high-risk HPV strains and while patients are still able to mount an effective immune response. The role of additional cancer screening beyond general recommendations is undefined, but we suggest that a high index of suspicion be maintained.

Figure 6.Sex differences in presenting symptoms. (A) Kaplan-Meier plot of symptom onset (age at presenting symptom) comparing sexes. (B) Patients presenting with infection as the first symptom split by sex and non-tuberculous mycobacterial infection (NTM) or non-NTM infections. P values were calculated using the Fisher exact test with Benjamini, Krieger, Yekutieli two-stage step-up method for a 5% false discovery rate. (C) Breakdown of presenting infections among patients with infection as the first symptom split by sex. Infxn: infection; NOS: not otherwise specified; HPV: human papilloma virus.

GATA2 is tightly regulated with selective pressures against damaging alleles. Among patients with haploinsufficiency due to deletions, nonsense-mediated decay, truncating mutations that removed the ZF2, or missense mutations that directly impact binding to the WGATAR motif (R396³³), 60/65 individuals were symptomatic, although three of the asymptomatic individuals are <13 years old. Missense mutations varied in age of onset and penetrance; those associated with an early median age of onset, T354 and R396, have been shown to have severely reduced transactivation capacity.^29,31 Chong et al. reported differential effects on disease depending on the specific GATA2 amino acid mutated.²⁹ They examined 94 reported individuals with either T354M, R396Q, or R398W for the cumulative incidence of MDS/acute myeloid leukemia or immunodeficiency: disease onset was earliest in those with R396Q and latest in those with R398W, consistent with our findings. While our in vitro assays were informative for decreased transactivation of ZF2 variants, a 1-day transfection assay is unlikely to recapitulate all the nuances of life-long hematopoiesis, which may allow the eventual expression of even mild variants. The latest median age of onset was within the group with intronic Enhancer mutations, in which only 24/50 individuals were symptomatic. Enhancer patients express some level of normal protein from the mutant allele,¹³ providing more WT GATA2 than any other category, but still insufficient for normal, life-long hematopoiesis. This variability across mutation types is concordant with differences in embryonic lethality between Gata2^-/- mice (which died on embryonic day 10.5³⁴) and +9.5 intronic enhancer knockout mice (which died on embryonic day 13.5³⁵) corresponding to different stages of hematopoietic development. Gata2^-/- mice fail to transition from aorto-gonad-mesonephros to fetal yolk sac/fetal liver hematopoiesis, while +9.5^-/- deletion impacts definitive hematopoiesis and vascular integrity. Heterozygous Gata2^+/- mice had reduced numbers of hematopoietic stem cells that fared poorly in competitive transplantation assays.³⁶ These mouse models demonstrate differential requirements for GATA2 function during embryogenesis and may parallel the differences in disease onset observed among patients with different genotypes.

Phenotypic variability within mutation groups suggests additional as yet undiscovered factors, which could include trans-acting genetic variants impacting GATA2 transcription or GATA2-targets. For missense changes, differing residual GATA2 activity may account for the specific phenotypic differences observed. Alternatively, environmental exposures could influence the course of disease in some patients.^37,38 Since our data were derived from patients and families referred to the NIH, in which at least one individual had symptomatic GATA2 deficiency, our data do not include individuals carrying GATA2 variants of uncertain significance yet lacking clinical phenotypes. While mutation-positive relatives of a deceased proband were included after genetic testing, our cohort does not include individual patients with early, fatal disease lacking a genetic diagnosis or their relatives. Additionally, the use primarily of DNA from whole blood does not provide screening for apparent mutation-negative parents who could be germline mosaic, passing the mutation to the affected individual. There is also a small possibility that some singleton cases may carry somatic mutations yet have a high enough variant allele frequency to be mistaken for having germline mutations. This distinction cannot be determined without testing a germline control tissue such as hair follicles or skin fibroblasts. Understanding the impact of mutation type on disease onset and penetrance is critical to determining the best monitoring frequencies and management strategies.

Sex differences were observed within our cohort of GATA2-deficient patients. Consistent with studies of pediatric and adolescent cases of MDS and leukemia,^39,40 partial or complete monosomy 7 was seen more frequently in males than in females. Additionally, while the overall frequency of mycobacterial infections was similar between males and females, males were more likely to have mycobacterial infections as their presenting symptom. HPV and other viral infections were more likely to be the presenting symptoms in females, possibly due to increased surveillance for women. Since most patients with GATA2-deficiency maintain immunoglobulin levels even as B-cell numbers decline,¹¹ vaccination prior to the loss of B cells would likely provide some protection against HPV for those mutation-positive children vaccinated. Almost a decade and a half after the initial clinical descriptions and identification of GATA2 as a disease-causing gene, the phenotype of the disease now includes cytopenias, leukemia and mycobacterial illnesses,^1-4 lymphedema,⁵ hearing loss, thrombotic events,⁹ as well as HPV-related neoplasms.⁹ Cohort studies have shown that HSCT is curative for most hematologic phenotypes.^41,42 However, full resolution of HPV-related disease has been variable,⁴² and non-hematopoietic phenotypes including lymphedema and hearing loss have not be corrected by HSCT. Additionally, cases of donor-derived⁴³ and relapsed⁴⁴ hematologic malignancies after HSCT have been reported. Worse outcomes were associated with pre-HSCT disease severity and monosomy 7.⁴⁵ GATA2 haploinsufficiency is a complex syndrome with clear genotype/phenotype correlations. A genetic diagnosis empowers appropriate prophylaxis, successful hematopoietic transplants, and proactive family screening. The newer generation of GATA2-deficient patients will benefit from improved understanding of disease onset and progression, which will allow pre-emptive intervention prior to development of hematologic malignancies. It will be important to temper our prognoses and treatments with these new understandings and to identify underlying molecular, cellular and environmental events that affect them.

Footnotes

• Received November 19, 2024
• Accepted December 29, 2025

Correspondence

°Current affiliation: Internal Medicine, University of Pennsylvania, Philadelphia, PA, USA

°°Current affiliation: Heart and Vascular Institute, The Johns Hopkins School of Medicine, Baltimore, MD, USA

A.P. Hsu twins@niaid.nih.gov

Disclosures

No conflicts of interest to disclose.

Contributions

APH, EPS and JLo collected clinical data. APH, SP and JLK analyzed data and made the figures. APH, SP, JLK, JLa and PLFJ performed the statistical analysis. LF, CP, CSZ, JRD, JD, KRC, DDH and SMH provided clinical care and patients’ samples. KRC reviewed bone marrow biopsies. APH wrote the manuscript. APH, KRC and SMH revised the manuscript.

Funding

This research was funded in whole or in part by the Division of Intramural Research, National Institute of Allergy and Infectious Diseases, National Cancer Institute, and Clinical Center, National Institutes of Health.