A ‘chief conductor’ driving cartilage formation and skeletal development, Runt-related transcription factor 2 (RUNX2) is a gene with at least 200 known mutations. Several such variants can manifest as cleidocranial dysplasia, a condition with delayed fontanel closure, enlarged head circumference, clavicular dysmorphia, abnormal dentition, short stature, or hand defects.[1] The frequency of RUNX2 variants in the general population may be as low as 0.4%.[2] The spalt-like transcription factor 1 (SALL1) gene broadly controls DNA packaging and chromatin remodeling. This gene also coordinates female urogenital structure, and Mullerian defects can occur with its mutation.[3] SALL1 also modulates human tumorigenesis with >70 variants now associated with Townes-Brocks syndrome (i.e., otic, limb, renal, and anal anomalies).[4] Such mutations occur with uncertain frequency, but from >3,000 targets with putative association to single gene disorders, SALL1 was among the least common.[5] The gene for sterile α motif domain-containing protein 9 (SAMD9) localizes to a region frequently involved in myeloid malignancies, blood dyscrasias, and childhood immune impairments. That SAMD9 might be causative in certain hematologic disorders was first advanced by Asou et al. [6] who linked microdeletions at this locus with myeloid malignancies. Germline missense SAMD9 changes were subsequently implicated in a constellation of features termed MIRAGE syndrome,[7] consisting of myelodysplasia, infection, restricted growth, adrenal hypoplasia, genital anomaly, and enteropathy. Fewer than 50 such cases have been described.[8] The major pathobiological consequences of RUNX2, SALL1, or SAMD9 variants limit the opportunity for longitudinal follow-up, and alteration of all 3 in combination has only been encountered once.[9] The current project was undertaken to enlarge the clinical characterization of this atypical genetic confluence during the adolescent-to-adulthood transition.

As the 3 newly discovered variants here were not identified together in any public genome library and the closest partial intersection involving these genes was reported from an animal model,[10] a priori pathogenicity estimates could not be calculated. Precautionary bone marrow biopsy was advised to provide tissue for genome-wide single nucleotide polymorphism (SNP) microarray analysis (Illumina Infinium CytoSNP-850Kv1.2 BeadChip; Illumina, San Diego, CA, USA). With disease-associated gene-centric content of >800 K SNP loci from chromosomes 1-22, >29 K and >2 K loci from X and Y chromosomes, respectively, this platform offered a resolution of ~1 probe every 4 kb with an extra 110,000 probes interrogating at least 4 pathology-associated genes, as per ABMG consortium opinion. Nucleotide positions were mapped using the February 2009 assembly of the human reference sequence.[11] The proband’s half-sibling also underwent bi-directional sequence analysis for the newly discovered variants in RUNX2, SALL1, and SAMD9 but was negative for all 3 (Table 1). Total body bone density was measured on the Lunar Prodigy Advance device (GE Medical Systems, Madison, WI, USA), calibrated at 37 μGy.

Medical information for this intake was gathered after proband’s high school graduation to prepare for her relocation to university. Laboratory tests, pathology slides, X-rays, and chart notes for the period corresponding to ages 13 to 18 years were matched with key health events based on patient narrative, contemporary notes, discussion with care team members, and parental interviews. Exome sequence analysis was completed in 2021 during COVID-19 hospitalization and intensive care unit transfer, followed by renal biopsy. Molecular characterization for the patient and both non-consanguineous parents provided an informative pedigree to document variants in RUNX2, SALL1, and SAMD9, as previously described.[9]

Complete blood count (CBC) data were analyzed from routine check-ups between 2016 and 2021, when the patient’s most recent body mass index was 17.8 kg/m². While this included marginally abnormal indices and occurred with no discernable pattern, the exception was mean corpuscular volume (MCV) which was consistently elevated. The patient had long taken multivitamin supplements for this macrocytosis, and serum folate was normal at 777 ng/mL. Platelet (PLT) concentration was always within the standard range during the review period although mean PLT volume (MPV) was not measured at every visit. When available, MPV was slightly low (data not shown). Flow cytometry immune testing identified nominally elevated CD3⁺ (ratio) and CD8⁺ (absolute) cells, an upward shift in the CD4⁺/CD8⁺ ratio, and markedly suppressed natural killer cells. Serum immunoglobulin (Ig) A and IgG levels were also marginally reduced, despite normal total Ig and IgM levels (Table 2). Serum blood urea nitrogen (BUN) and creatinine (Cr) levels were unrelated to concurrent CBC fluctuations (Fig. 1A, B).

Whole blood analysis on morning of bone marrow biopsy showed erythrocyte (RBC), hemoglobin, and hematocrit at 3.3×10⁶/μL. 11.8 g/dL, and 36.1%, respectively. Leukocyte count (white blood cell) was normal at 5.7 K/μL (reference range, 4.4-9.7). Both MCV and mean corpuscular hemoglobin were elevated at 106.8 fL and 34.9 pg. PLT count was normal at 182 K/μL. Generalized bone marrow hypocellularity was observed with trace storage iron, decreased erythroid precursors with mildly megaloblastoid maturation, decreased granulocyte precursors with full maturation, and normal megakaryocytes (Fig. 2A, B).

The myeloid:erythroid precursor ratio was normal on core analysis, and no increased fibrosis was evident on reticulin stain. Marrow samples showed high lymphocytes and monocytes (43.5 and 6%, respectively), low bands (5.5%; reference range, 17-33) and nominally elevated orthochromatic normoblasts (7.5%; reference range, 1-5). Fluorescent in situ hybridization demonstrated a normal signal with no copy number variance in nuclei (N=200). Bone marrow cytogenetics verified a non-mosaic 46,XX karyotype, aligning with results obtained earlier from peripheral (somatic) samples. The fragile X carrier panel was negative. At age 15 years, her bone age via standard radiograph was below mean, consistent with age 13.5 years. Bone density assessment identified low bone density, with height-adjusted Z-scores of −0.7 and −1.6 for axial (spine) and total (sub-cranial) body, respectively.

By age 7 years, all deciduous teeth were succeeded by permanent teeth, although anomalous non-descent of left cuspid was incidentally identified during pre-orthodontic screening. In addition, a panoramic X-ray at age 17 years revealed multiple underdeveloped roots and secondary loss of right lateral incisor. Surgery to remove all 4 impacted wisdom teeth (Fig. 3) is scheduled.

Menarche was at age 11 years but only one spontaneous menses followed; within 2 years menses had ceased. At age 15.5 years, low-dose oral contraceptives were initiated to restore cyclicity, and no other hormonal therapy was ever prescribed. Serum follicle-stimulating hormone was consistently >100 mIU/mL with anti-Müllerian hormone below measurement threshold (<0.015 ng/mL). Small ovaries were seen on abdominal ultrasound, including sparse but active bilateral follicular response on the clomiphene challenge test, and pre-replacement serum estradiol levels were regularly <5 pg/mL, all indicative of diminished ovarian reserve.[12] Anti-ovarian antibody test was negative. At age 16, a solitary 2 cm right anterior neck mass was evaluated; laboratory tests were negative for autoimmune involvement and found no evidence of abnormal thyroid or parathyroid function. The structure was nontender and was palpable only after singing. Fine needle aspiration of the lesion showed Bethesda category III cytology; molecular testing found a missense c.1358T>C mutation in the thyroid-stimulating hormone (TSH) receptor gene with a variant allele frequency of 46%, negative for gene fusions or copy number variants.

Transcription factors set the pace and pattern of nuclear DNA conversion to mRNA. From amidst more than 1,500 such factors, RUNX2, SALL1, and SAMD9 are especially prominent. The genes coding for these products thus controls cell division, migration, body plan architecture, and apoptosis. While disruption of these factors typically is poorly tolerated, our patient had an active, healthy childhood while carrying variants in all 3. Remarkably, the most dramatic highlight against a rather ordinary pediatric background was her hospitalization for COVID-19. Attribution of findings discussed here to genetic etiologies vs. COVID-19 is challenging, given the unclear status of the specified variants and the expansive, evolving nature of ‘Long COVID’. Curiously, the high MPV seen with COVID-19 infection documented by others [13] was absent here. Perhaps the 3 variants dampened MPV by altering individual PLT responsiveness, even though overall PLT number was unaffected. Likewise, the observed low serum IgA and IgG levels alongside normal total Ig may reflect a synthesis of coordinated action by these variants.

Some 150 mutations in SAMD9 have been associated with outcomes ranging from spontaneous remission to malignant progression.[14] Most SAMD9 variants presage early death from myeloid dysplasia, immune system imbalance, adrenal insufficiency, or chronic undernourishment from feeding difficulty.[15] SAMD9 mutations via germline transmission predispose to low PLTs, acquired monosomy 7, constitutional abnormalities (e.g., ambiguous genitalia) and immune dysfunction.[16] Less is known about SAMD9 variants which appear as de novo events. Computer modeling has envisioned a protein-protein interaction network of differentially expressed genes, and human SAMD9 was among the ‘hub genes’ having special relevance after SARS-CoV-2 infection.[17] Thomas et al. [18] assessed the functional impact of wild-type and mutant SAMD9 in primary mouse or human hematopoietic stem and progenitor cells. Using protein interactome analyses, transcriptional profiling, and functional validation, it was concluded that SAMD9 mutations tend to favor interference with DNA damage repair and ultimately apoptosis in hematopoietic cells.[18] The hypocellular terrain noted here on bone marrow biopsy may be a partial expression of this SAMD9 influence.

RUNX2 is considered a centrally regulating transcription factor for osteoblast and chondrocyte differentiation and overall skeletal architecture.[19,20] Some 80 variants in RUNX2 have been identified [1] and while heterozygous loss of function mutations can lead to cleidocranial dysplasia, this is inconsistent. Triplication or quadruplication of RUNX2 accompanies more serious syndromic phenotypes, including coronal/sagittal synostosis or pan-craniosynostosis, suggesting a dosage effect. For our patient, sequence analysis did identify a 3′ duplication involving at least exons 6 through 9 of RUNX2,[9] as plausible contributors both to microscopic (hypocellular) and macroscopic (reduced density) osseous features observed here.

In late adolescence or early adulthood third molar impaction is not unusual, and others have investigated differential expression of RUNX2 regarding tooth location before surgical removal. While RUNX2 was not implicated, another transcription factor (muscle segment homeobox 1 [MSX1]) was differentially expressed depending on the depth of molar impaction, and RUNX2 is partially controlled by MSX1. As alveolar bone remodeling is central to orthodontic tooth movement, a novel regulatory mechanism whereby FOXO3 induces osteocalcin transcription by promoter activation in concert with RUNX2 could help explain dental features in this case.

The T>C mutation in codon 453 of the TSH receptor (TSHR) gene found here was previously reported in the setting of nonautoimmune hyperthyroidism—but without RUNX2, SALL1, or SAMD9 involvement. Wide tissue expression of TSHRs is well established and includes brain, bone marrow, lymphocytes, pituitary, thymus, testes, kidney, adipose tissue, and fibroblasts. While oocyte growth is influenced via TSHR/cyclic adenosine monophosphate signaling, there has been no reason to obtain ovarian tissue sampling in this patient. A pelvic ultrasound will be useful to identify any changes in gross ovarian anatomy. Similarly, tracking thyroid size, nodularity, and tenderness, as well as the thyroid laboratory panel will be important.

Concerning SALL1, some 50 different mutations are currently known. SALL1 and RUNX2 may have special relevance in human reproduction. Mammalian estrogen receptor (ER)-β is required for ovarian follicles to advance past the antral stage, and work in rat granulosa cells has placed RUNX2 within the ER-β-regulated genes directing folliculogenesis, oocyte maturation and ovulation. While SALL1 is essential for stem cell maintenance in kidney, heart, and spermatogonial progenitors,[4] its role in human ovarian tissue awaits better characterization.

The Venn diagram for worldwide clinical experience with RUNX2, SALL1, and SAMD9 returned a null union set prior to this patient. The detrimental consequences of mutation of these genes in isolation may have been mitigated by the chance occurrence of all 3 variants in concert—an offset possibly enabled by cross-gene effects or epigenetic silencing.[9] Monitoring for patients with a SAMD9 variant includes CBC with differential every 6 months, and repeat bone marrow aspirate/biopsy (and karyotype) should anemia, thrombocytopenia, or neutropenia develop.[8] Red blood cell dysplasia, agglutination, or fragmentation are unlikely given the low/normal red-cell distribution width measured here. Serum BUN and Cr levels evidenced no discernable pattern whenever these were abnormal but as cystatin-C may be superior to BUN/Cr to detect early-stage disease, a screening panel including all 3 will guide the need for repeat renal biopsy. To guard against further loss in bone mineral content, the current oral contraceptive pills regime will be maintained with annual DXA. As this case incorporates a de novo variant, the risk to siblings is considered somewhat greater than the background population [8] and cytogenetic testing for an older half-brother was reassuring. Clinical guidelines for these variants have not been developed due to their exceptionally low frequency, and this underscores an important limitation of our report—it is impossible to assign specific genetic causality to such rare events. Because not all instances of marrow hypocellularity will render immediate effects, periodic reassessment is essential.