Abstract
Objective
Steroid‐induced osteonecrosis of the femoral head (SONFH) is a refractory skeletal disorder influenced by genetic and environmental factors. However, conclusive pathogenic genetic evidence remains elusive due to the limited exploration of rare damaging variants. In this study, we aimed to identify rare variants associated with SONFH.
Methods
We conducted whole‐exome sequencing in an SONFH case‐control study comprising 174 patients with systemic lupus erythematosus (SLE) taking steroids, followed by comparing the cases with an ethnically matched healthy population and validation in an additional SONFH cohort of 246 patients without SLE. Rare damaging variants were identified via genetic burden analysis and confirmed by Sanger sequencing and pedigree studies. The functional assessments of the erythrocytes on target variants were performed.
Results
We identified 10 heterozygous ANK1 and EPB41 rare variants in 9 (10.8%) of 83 patients with SONFH compared with 0 of the 91 non‐SONFH controls (0%). The burden effect of them remained robust after adjusted analysis (Firth's logistic regression‐adjusted odds ratio [aOR] ANK1 11.3; aOR EPB41 32.3; both P < 0.05). The variants were detected in 10 of the validated SONFH cohort (4.1%) with 246 non‐SLE conditions. Functional analyses revealed that the variants result in membrane dysfunctions in the patients’ erythrocytes, characterized by reduced ANK1 expression, abnormal morphology, and increased hypotonic hemolysis.
Conclusion
These findings suggest that the rare variants in ANK1 and EPB41 are novel SONFH disease risk factors that compromise erythrocyte membrane integrity, exacerbating microcirculatory damage in the presence of glucocorticoid as a second hit.
INTRODUCTION
Steroid‐induced osteonecrosis of the femoral head (SONFH) is a leading cause of nontraumatic ONFH. 1 , 2 Primarily affecting young to middle‐aged adults, SONFH disrupts subchondral bone integrity due to impaired microcirculation. This leads to collapse and progressive secondary arthritis, becoming a mounting global health challenge. 3 Differences in epidemiologic survey methodologies preclude direct comparison of global incidence or prevalence rates for ONFH; nonetheless, available data suggest geographic disparities, with notably higher reported rates in certain Asian countries. For example, the overall prevalence rate of nontraumatic ONFH in China was 725 per a population of 100,000, whereas the average annual prevalence of overall (including both traumatic and nontraumatic) ONFH in South Korea over a five‐year survey period was 28.91 per 100,000. 4 , 5 The incidence of nontraumatic ONFH in Japan was 2.51 per 100,000, whereas incidence data from the United Kingdom show an even lower figure, with an overall ONFH incidence of 2.3 per 100,000. 6 , 7 The cause of SONFH is complex, involving both genetic and environmental factors. 1 , 8 The association between environmental factors (such as glucocorticoid [GC] use and autoimmune conditions) and SONFH is well‐established. 1 , 8 , 9 Although genetic predisposition is known to play a role in the pathogenesis of SONFH, the exact genetic variants underlying SONFH development in patients exposed to steroid administration remains elusive.
Common variants associated with the development of SONFH were previously identified to affect GC metabolism, 10 , 11 , 12 coagulation, 13 osteogenic and adipogenic differentiation, and immunoregulation. 14 It is important to note that although many of these common variants have been identified as risk factors linked to SONFH, they only explained a small part of the high heritability, and they may not necessarily have causal effects as they often localize in noncoding regions. Some rare variant defects in families with autosomal dominantly inherited genetic ONFH have been identified as disease causing, such as defects in collagen type II alpha 1 chain (COL2A1) 15 or transient receptor potential cation channel subfamily V member 4 (TRPV4). 16 However, no rigorously designed studies have yet demonstrated a significant burden of rare variants in SONFH‐associated genes, nor has any evidence conclusively linked ethnic disparities in SONFH prevalence to population‐specific minor allele frequency (MAF) variations of such variants.
Rare damaging variants detected by whole‐exome sequencing (WES), which significantly modify protein sequences and function and potentially establish a direct causal link to disease risk, play pivotal roles in the etiology of complex human diseases. 17 We hypothesized that using WES and burden analysis might uncover new, rare damaging variants as pathogenic genetic risk factors for SONFH. We provide data herein suggesting rare damaging variants in ANK1 and EPB41 both encoding (red blood cell membrane) cytoskeletal proteins as two of the novel SONFH pathogenic genetic risk factors.
MATERIALS AND METHODS
Study design and participants
The study was designed as a two‐phase study based on WES. In the discovery stage, a case‐control analysis of GC‐treated patients with systemic lupus erythematosus (SLE; SONFH vs non‐SONFH) screened candidate rare variants via gene‐based collapsing. These variants were identified by comparing their allele frequencies in both groups to those in an ethnicity‐matched reference population from the HUABIAO project. 18 In the validation stage, a cohort of non‐SLE patients with SONFH was validated for sequencing. The workflow for design and analysis is shown in Supplementary Figure S1. All participants were of Han Chinese ethnicity. Between February 2016 and December 2019, we recruited 219 GC‐treated patients with SLE from (a) our institutional ONFH biobank and (b) a previous cohort study. 9 Based on post‐GC SONFH development status, 95 were classified as cases (asymptomatic early stages were confirmed by magnetic resonance imaging [Supplementary Figure S2] and advanced stages were confirmed by x‐ray or computed tomography according to the clinical guideline 19 ) and 124 as controls.
A detailed list of inclusion and exclusion criteria is described in the Supplementary Methods section of the Supplementary Appendix. Finally, a total of 177 patients (85 patients with SONFHs and 92 non‐SONFH patients) were included and underwent WES of blood samples. Between May and July 2024, we sequenced samples of another cohort of 246 non‐SLE patients with SONFH in our database and biobank. This study was approved by the Ethics Committee of the Shanghai Sixth People's Hospital (No.2016‐KY‐001[K]). All participants provided written informed consent.
In the case‐control study of 177 sequenced samples (three samples were removed for the final analysis due to failing variant quality control), the mean depth of WES was 139× with 99.4% coverage of target nucleotides, and in the validation group of 246 patients with SONFH with non‐SLE diseases, the mean depth and coverage of target nucleotides were 185× and 99.6%, respectively.
Bioinformatic processing of WES data
The raw sequencing data (FASTQ format) from all participants were processed through a standardized bioinformatic pipeline based on Genome Analysis Toolkit best practice with the GRCh37 human reference genome. The specific processing procedures, including variant calling and data quality control, were detailed in the Supplementary Methods (Supplementary Appendix). The frequency distribution of variants after quality control is shown in Supplementary Figure S3A. Frameshift indels were not analyzed in this study due to their limited detection in the samples.
PLINK version 1.9 (https://www.cog-genomics.org/plink/1.9/) was used to estimate autosomal heterozygosity, relatedness, and population outliers (via principal component analysis [PCA]). Ensembl Variant Effect Predictor (version 113) was applied to functionally annotate and prioritize genetic variants. Data were filtered to retain coding variants, classified as synonymous, missense, or loss‐of‐function (LoF; including splice acceptor/donor, stop gained/lost, and initiator codon), with an MAF <0.1% in both Genome Aggregation Database (gnomAD; r2.1, http://gnomad.broadinstitute.org) and 1000Genomes populations. Polymorphism Phenotyping version 2 (PolyPhen‐2), Sorting Intolerant from Tolerant (SIFT; version 5.2.2), and Combined Annotation Dependent Depletion (CADD; GRCh37‐v1.7) were used to predict the variant pathogenicity. Variants predicted as probably/possibly damaging by PolyPhen‐2 and deleterious by SIFT were classified as deleterious missense (D‐mis); among these D‐mis variants, those with CADD scores ≥15 were designated as potential pathogenic variants. 20
Gene‐level collapsing analysis and gene set burden analysis
We first identified potential SONFH‐associated rare variants via Fisher exact test. We then compared the frequencies of these variants with those in a healthy population of the same ethnicity, using the HUABIAO project (https://www.biosino.org/wepd), as additional healthy controls and performed P value adjustment for multiple comparisons with Bonferroni correction. The large‐scale WES database of healthy Han Chinese individuals contains deep sequencing (>100×) of over 9,900 unrelated healthy samples (as of 2020), mainly from three representative Han Chinese populations in north China, east China, and south China. 18 After excluding variants with no significant differences, we obtained SONFH‐related risk variants. To address the unbalanced case‐control distribution, we performed gene‐level collapsing analysis of these selected risk variants using Firth's bias‐reduced logistic regression, 21 , 22 , 23 adjusting for sex, age of GC pre‐exposure, initial intravenous GC, GC daily dose, and GC exposure time. Based on these risk variants identified from the gene‐level burden analysis, we performed an exploratory functional enrichment gene set burden analysis using Gene Ontology (GO) pathways and Firth's bias‐reduced logistic regression.
Key novel variants validation and their functional study
We confirmed the rare variants in ANK1 and EPB41 by Sanger sequencing and pedigree analysis following the routine protocols and validated them in an independent SONFH group with non‐SLE patients. To assess how the variants influenced the morphology and susceptibility to hemolysis of red blood cells (deleterious functional consequence on the erythrocyte membrane), we evaluated the blood smear and then used imaging flow cytometry, Western blot, and flow‐cytometric osmotic fragility test, respectively (Supplementary Methods in Supplementary Appendix).
Statistical analysis
Quantitative data are expressed as means with SDs or medians with interquartile ranges, which were compared using parametric or nonparametric tests as appropriate. Categorical variables were presented as absolute numbers and percentages, which were compared using the chi‐square or Fisher exact tests. Gene‐level collapsing analysis between groups was performed with Fisher exact test. Multivariable analyses were conducted using Firth's bias‐reduced logistic regression to assess the independent effects of single genes on SONFH, adjusting for sex, age of GC pre‐exposure, initial intravenous GC, GC daily dose, and GC exposure time following methods described in previous similar study. 21 , 22 , 23 Principal components were not included in covariate adjustment because the PCA results showed no population stratification in our study population (Supplementary Figure S3B).
P values were two sided with statistical significance defined at ≤0.05. Where appropriate, P values were adjusted using the Bonferroni correction or false discovery rate (FDR). Following multiple comparisons of the frequencies for these 35 potential variants between the case‐control and additional healthy control (HUABIAO project) groups, the P values were adjusted using the Bonferroni correction (threshold P ≤ 0.05/35/2 = 0.00071). The threshold for significant enrichment in pathway analysis was defined as FDR q‐value ≤0.05. Statistical analyses were performed using R package (version 4.5.1, The R Foundation, http://www.R-project.org).
Data availability statement
The analysis data sets supporting the conclusions of this article are included within the article and its supplementary appendixes. Additional data are available through reasonable request to the corresponding author.
RESULTS
Characteristics of the case‐control patients
In a case‐control study of 219 patients with SLE treated with GCs, a total of 174 cases were selected for the final analysis after exclusion of 45 cases (34 patients did not meet clinical inclusion criteria, eight samples were excluded due to low DNA concentration, and three samples were excluded due to their poor single‐nucleotide polymorphism quality control; Supplementary Figure S1). Among the 174 patients, 83 patients developed SONFH, and 91 did not develop SONFH and served as control. The patients with SONFH were relatively younger (age of blood sampling: mean 31.0 ± SD 11.1 years vs mean 43.0 ± SD 13.0 years, P < 0.001), with relatively more men than compared to the control group (16.9% vs 4.4%; Table 1). The SONFH group developed osteonecrosis after a median exposure to GCs over 26 months with a cumulative dose equivalent to 9.9 g of prednisone (Table 1), whereas the control group did not show any SONFH after an even longer period of time (123 months) and higher dosage (25.2 g).
Table 1.
Characteristics of patients with SLE with GC exposure between the SONFH and non‐SONFH groups*
| Characteristic | Non‐SONFH (n = 91) | SONFH (n = 83) | P value |
|---|---|---|---|
| Male patients, n (%) | 4 (4.4) | 14 (16.9) | 0.007 |
| Age of blood sampling, mean ± SD, y | 43.0 ± 13.0 | 31.0 ± 11.1 | <0.001 |
| Age of GC pre‐exposure, mean ± SD, y | 31.7 ± 12.7 | 27.9 ± 10.4 | 0.03 |
| GC exposure | |||
| Initial intravenous GC, n (%) | 26 (28.6) | 55 (66.3) | <0.001 |
| GC exposure time, median (IQR), mo | 123.0 (47.0–204.0) | 26.0 (15.0–51.0) | <0.001 |
| GC total cumulative dose, median (IQR), g | 25.2 (12.5–43.3) | 9.9 (7.4–15.6) | <0.001 |
| GC daily dose, median (IQR), mg/d | 8.1 (5.5–11.9) | 12.7 (8.4–17.7) | <0.001 |
| Any aPLs, n (%) | 10 of 79 (12.7) | 7 of 18 (38.9) | 0.02 |
| Key clinical manifestations, n (%) | |||
| Skin manifestation | 40 of 91 (44.0) | 35 of 62 (56.5) | 0.13 |
| Lupus nephritis | 45 of 91 (49.5) | 35 of 62 (56.5) | 0.40 |
| Arthrosis | 49 of 91 (53.8) | 22 of 62 (35.5) | 0.03 |
| Serositis | 5 of 91 (5.5) | 3 of 62 (4.8) | 1.00 |
| Neurologic disorder | 3 of 91 (3.3) | 0 of 62 (0.0) | 0.27 |
| Hematologic disorder | 33 of 91 (36.3) | 22 of 62 (35.5) | 0.92 |
| Only one manifestation | 32 of 91 (35.2) | 18 of 62 (29.0) | 0.43 |
| Any two manifestations | 31 of 91 (34.1) | 15 of 62 (24.2) | 0.19 |
| Equal or more than three manifestations | 28 of 91 (30.8) | 29 of 62 (46.8) | 0.04 |
| Rare ANK1 and EPB41 variants, n (%) a | 0 (0.0) | 9 (10.8) | 0.001 |
| ANK1 variants | 0 (0.0) | 5 (6.0) | 0.02 |
| EPB41 variants | 0 (0.0) | 5 (6.0) | 0.02 |
aPLs, antiphospholipid antibodies; GC, glucocorticoid; IQR, interquartile range; SLE, systemic lupus erythematosus; SONFH, steroid‐induced osteonecrosis of the femoral head.
One patient had each variant in both ANK1 and EPB41 genes.
Identification of rare damaging variants in ANK1 and EPB41
A total of 162,008 variants were retained for the purpose of the final data analysis in the 174‐sample discovery cohort. Gene collapsing analysis of the case‐control cohort initially identified 35 rare variants. After comparing them with the healthy population with the same ethnic background, we identified rare variants in a total of 29 genes significantly associated with SONFH (Figure 1A), which included previously reported APOB (Supplementary Table S1) and the other 28 novel genes with newly identified rare variants (Supplementary Table S2). Using multivariable Firth's logistic regression at the single‐gene level, we found that 10 of them were closely associated with SONFH (P < 0.05; Figure 1A, Supplementary Table S3), which were present in 44 of 83 cases of SONFHs (53.0%). Subsequently, GO enrichment analysis revealed that 27 of 35 gene sets (including collagen‐containing extracellular matrix, fatty acid metabolic process, lipid catabolic process, regulation of calcium ion transport, and structural constituent of cytoskeleton) were significantly associated with SONFH risk (FDR < 0.05). Among them, two rare variants (ANK1 and EPB41) encoding cytoskeletal protein with spectrin binding were significantly associated with SONFH development (FDR < 0.01; Figure 1, Supplementary Appendix 2).
Figure 1.

Genetic burden of rare damaging variants associated with SONFH. (A) The circular plot shows 29 genes harboring filtered candidate rare damaging variants. Ten of these gene (in red) carry rare damaging variants closely associated with SONFH. (B) The forest plot shows the GO pathway enrichment analysis with log(OR) values (Firth's regression‐adjusted) quantifying disease risk association strength per pathway gene set. BP, biologic process; CC, cellular component; CH, Carbon‐Hydrogen bond; FDR, false discovery rate; GO, Gene Ontology; MF, molecular function; mRNA, messenger RNA; OR, odds ratio; SnRNP, small nuclear RNP; SONFH, steroid‐induced osteonecrosis of the femoral head.
Ten heterozygous variants of ANK1 (Figure 2A) and EPB41 (Figure 2C), localizing in the codons encoding highly conserved amino acids (Figure 2B and 2D, Supplementary Figure S4), were found in nine patients with SONFH in the case group (Supplementary Table S4, the primers for Sanger sequencing are listed in Supplementary Table S5). In contrast, they were not found in the patients who did not develop SONFH. The results indicated that the patients with the variants of ANK1 and EPB41 have a high risk of developing SONFH in association with SLE. Consistently, multivariable analysis showed that the rare variants of ANK1 (Firth's logistic regression‐adjusted odds ratio [aOR] 11.3, 95% confidence interval [CI] 1.1–1518.1) and EPB41 (aOR 32.3, 95% CI 2.0–4,883.0) were two of the independent risk factors associated with SONFH (Supplementary Table S3). And enrichment of the combined variants in these two genes conferred a greatly elevated risk for SONFH (aOR 36.3, 95% CI 2.8–5394.6; Supplementary Appendix 2).
Figure 2.

ANK1 and EPB41 variants in the case‐control study of patients with SLE‐SONFH. (A) A schematic diagram of ANK1 protein. The “n” indicates case numbers. (B) The variant sites (green boxes) are evolutionally conserved in ANK1 of human (NP_000028.3), mouse (NP_001104253.1), rat (NP_001382491.1), cattle (XP_059738357), rhesus monkey (XP_015000781.2), and African clawed frog (XP_041443016.1). (C) A schematic diagram of EPB41 protein. The “n” indicates case numbers. (D) The variant sites (green boxes) are evolutionally conserved in EPB41 of human (NP_001159477.1), mouse (NP_906273.3), rat (XP_017449316.1), cattle (NP_001415051.1), rhesus monkey (XP_028692906.1), and African clawed frog (XP_041436831.1). 4_1_CTD, 4.1 protein C‐terminal domain; ANKYR, ankyrin repeat; B41, band 4.1 homologues; Death, death domain of Ankyrin‐1; FA, FERM (4.1 [four point one] and ERM [ezrin, radixin, moesin] proteins) adjacent; FERM_C, FERM domain C‐lobe of Protein 4.1 family; SAB, spectrin and actin binding domain; SLE, systemic lupus erythematosus; SONFH, steroid‐induced osteonecrosis of the femoral head; UPA_2, UPA domain; WTX, domain of Wilms tumors protein; ZU5, domain present in ZO‐1 and Unc5‐like netrin receptors. Color figure can be viewed in the online issue, which is available at http://onlinelibrary.wiley.com/doi/10.1002/art.70153/abstract.
The probability of being LoF intolerant (pLI) ranges from 0.0 to 1.0. A higher score means a greater degree of intolerance for LoF variants in a healthy population. Therefore, based on data from the gnomAD database (https://gnomad.broadinstitute.org/), ANK1 is highly intolerant to haploinsufficiency as indicated by pLI = 1.0 (Observed/expected ratio [o/e] = 0.24 [95% CI 0.19–0.30]), whereas EPB41 is not (pLI = 0, o/e = 0.59 [95% CI 0.48–0.74]). These heterozygous variants were further validated by Sanger sequencing (Supplementary Figure S5). Furthermore, the pedigree analysis revealed that the variants in ANK1 and EPB41 were inherited from the paternal or maternal side (Supplementary Figure S6).
The variants are associated with SONFH independent of SLE status
To determine whether the variants were associated with SONFH independent of SLE status, we analyzed the rare damaging variants in another cohort of 246 patients with SONFH without SLE diseases (Supplementary Table S6). Overall, the burden of rare variants in these 10 genes was successfully replicated in an independent validation cohort, observed in 28.5% (70 of 246) of cases. This confirmed the initial finding from the case‐control study (53.0% of cases; Supplementary Table S7). Indeed, we identified 10 cases (4.1%) with six ANK1 variants and four EPB41 variants (Supplementary Table S7, Supplementary Figure S7A). Three ANK1 variants (p.S1507P and p.L1194P, Supplementary Table S8) were identical to previously identified ones, whereas the other three variants were different but in the conserved sequences (Supplementary Figure S7B). All the EPB41 variants (p.R700Q, Supplementary Table S8) were identical to previously identified ones. In summary, our findings suggest that ANK1 and EPB41 variants are relatively universally associated with SONFH and are not restricted to patients with SLE (Supplementary Table S7 and S8). Notably, all patients with rare variants in this study did not exhibit clinical symptoms of jaundice or hemolytic anemia, nor was there any documented evidence of hereditary spherocytosis or elliptocytosis. (Supplementary Table S8).
Morphologic alterations in erythrocytes of the patients
We first checked if the expression of ANK1 and EPB41 on red blood cell membrane was influenced by these rare variants. Although case 1‐2 (p.Leu1194Pro variant in ANK1 and p.Arg700Gln variant in EPB41) and case 1‐7 (p.Arg700Gln variant in EPB41) shared the same variant in EPB41, their EPB41 expression did not decrease (Figure 3A). However, both patients had decreased ANK1 expression (Figure 3A), no matter if they had the ANK1 variant or not. The consistent results were observed in their family members (Figure 3A) according to the pedigree analysis (Supplementary Figure S6). Our results suggested that the variants in ANK1 and EPB41 may influence red blood cell morphology by impairing the stability of cytoskeleton through decreased ANK1 levels. However, the morphologic changes of red blood cells in these patients were usually neglected because they were not easily distinguished and quantified on the blood smears by light microscopy (data not shown). So we decided to use imaging flow cytometry to examine the morphologic changes of the patients’ peripheral red blood cells (Figure 3B). Discocytes and nondiscocytes could be divided by analysis of the label‐free images of red blood cells using a series of features/masks of the analytical software (Supplementary Figure S8). Indeed, the proportions of discocytes in red blood cells of patients (mean 40.7% ± SD 9.8%, n = 3) were significantly lower (P < 0.05) than that of the healthy controls (mean 56.1% ± SD 5.3%, n = 4). This suggested that there are more abnormal morphologies of red blood cells in patients carrying the variants.
Figure 3.

The rare variants of ANK1 and EPB41 impair erythrocyte membrane integrity and function. (A) Western blot of membrane proteins in erythrocytes of the patients and their family members. Control 1 is the healthy control; Patient 1 carries the p.Leu1194Pro variant in ANK1 and the p.Arg700Gln variant in EPB41; Patient 2 carries only the p.Arg700Gln variant in EPB41. GAPDH was referred to as internal control. The number under each band of ANK1 and EPB41 was the relative expression level calculated by measuring the band intensity relative to that of GAPDH. The image showed one representative result of two repeats. Western blot results (the right panel) showed that ANK1 expression was significantly lower in cases/carriers compared to controls, whereas EPB41 expression showed no significant difference. (B) The morphologic changes of red blood cells caused by ANK1 and EPB41 variants. The percentage of discocytes was determined by imaging flow cytometry. (C) Flow‐cytometric osmotic fragility test results show the susceptibility of residual red blood cells to hemolysis. The left and middle panels display representative histograms for Control 1 and Patient 2. After exposure to hypotonic solution, 100% of red blood cells remained in Control 1, compared to only 59.2% in Patient 2. The right panel summarized the difference of percentage of residue red blood cells between the controls and patients. Ctrl, control. Color figure can be viewed in the online issue, which is available at http://onlinelibrary.wiley.com/doi/10.1002/art.70153/abstract.
The erythrocytes carrying the variants are susceptible to hemolysis
The osmotic fragility test is one of the screening tests for red blood cell membrane disorders by evaluating the red blood cells’ ability to incorporate hypotonic solutions. 24 Regular erythrocytes can incorporate some quantities of hypotonic solutions without lysis, whereas those with membrane disorders are prone to hemolysis. Therefore, we performed the flow‐cytometric osmotic fragility test to examine the function of the red blood cell membrane (Supplementary Figure S9). We found that the residue cells after spike‐in water (hypotonic solution) were significantly reduced in the red blood cells of the patients compared to the healthy controls (P < 0.01; Figure 3C, Supplementary Table S9). Taken together, the mutant red blood cells had morphologic changes and were susceptible to hemolysis, most likely due to the disruption of cytoskeletons by ANK1 and EPB41 variants.
DISCUSSION
To the best of our knowledge, this study represents the first identification of rare genetic variants, particularly heterozygous variants in ANK1 and EPB41, associated with a substantially higher risk of SONFH (aOR ANK1 = 11.3, aOR EPB41 = 32.3). We have shown that the burden of rare variants in ANK1 and EPB41 was significantly increased in cases (10.8%) versus controls (0.0%) in steroid‐administrated patients with SLE (aOR = 36.3). Our findings were supported by pedigree analysis, functional study, and validation in an independent cohort of 246 non‐SLE patients with SONFH.
Ankyrin 1 (encoded by ANK1) and membrane protein band 4.1 (encoded by EPB41) connect the plasma membrane with the spectrin‐actin cytoskeleton in eukaryotic cells like erythrocytes. 25 Therefore, ANK1 and EPB41 play crucial roles in regulating red blood cell shape and mechanical integrity. Defects or mutations in components of the complexes consisting of ANK1 and EPB41 are known to cause hemolytic anemias, including hereditary spherocytosis and elliptocytosis. 26 In our study, however, the rare damaging variants in EPB41 did not influence its own protein expression (Figure 3A). Instead, the protein level of ANK1 in the cytoskeleton was significantly decreased (Figure 3A). It is most likely that the rare damaging variants in EPB41 altered the conformation of the protein, which inhibits anchoring of ANK1 through indirectly impairing the stability of the cytoskeleton network. Despite the morphologic and functional defects of the patients’ red blood cells (Figure 3BC), no clinical symptoms of jaundice or hemolytic anemia were observed (Supplementary Table S8). In addition, the functional defects were observed in the patient's son (73.7% residue cells in osmotic fragility test), who carried the same variant but did not have SLE (case 1‐7 in Supplementary Figure S6). Taken together, the novel variants in ANK1 and EPB41 mildly damaged the erythrocyte membrane–cytoskeleton complex and posed a risk for developing osteonecrosis without florid hemolysis, 26 which required a secondary hit—in this case, GC use.
The heterozygous variants are likely to have causal effects because ANK1 is highly intolerant to haploinsufficiency (but not reaching statistical significance for EPB41). Consistently, we observed an increase in abnormal red blood cell morphology in peripheral blood samples of the patients who developed SONFH (Figure 3B), which, however, did not meet the diagnostic criteria for spherocytes or elliptocytes. The deformability of red blood cells is vital for their ability to pass through capillaries that are only about one‐third the diameter of the cells themselves. 27 In sickle cell disease (SCD), stiff red blood cells impair the microcirculation and cause distal tissue ischemia by vaso‐occlusion, which eventually leads to ONFH. 28 The adverse effects of ONFH from SCD can be exacerbated under GC exposure. 29 In analogy to SCD, ANK1 and EPB41 variants rendered red blood cells more susceptible to hemolysis during osmotic fragility testing (Figure 3C) and were associated with a slight elevation in reticulocyte levels (Supplementary Table S8). Indeed, osteonecrosis was documented in a few case series of hereditary spherocytosis. 30 , 31 Therefore, we propose that heterozygous variants of ANK1 and EPB41 can alter the morphology and function of red blood cells, resulting in increased number of red blood cells with nonbiconcave shapes and enhanced fragility or rigidity. These changes decrease erythrocyte deformability and impair the transit efficiency of the cells through capillaries. 32 , 33 This may compromise microvascular flow dynamics, increasing friction and adhesive damage to the microvessel wall 28 , 34 ; when individuals undergo steroid therapy due to some diseases (eg, SLE, thrombocytopenia, allergic purpura, etc), endothelial damage exacerbates microvascular congestion according to the multihit hypothesis of GC effects, 1 , 35 ultimately leading to microcirculatory disturbances and local ischemia and causing the osteonecrosis (Figure 4). Furthermore, we noticed that the family members who carry the heterozygous variants without being exposed to steroids had intact femoral heads (data not shown). A previous study has suggested that ANK1 plays a significant role in coronary microvascular dysfunction. 36 ANK1 expression in human osseous tissues remains limited. In publicly available data sets such as the GTEx database and the Human Protein Atlas, ANK1 expression is primarily detected in hematopoietic cells within bone marrow. Our examination of publicly available immunohistochemistry data further supports the absence of ANK1 expression in bone marrow–adjacent tissues. We acknowledge a recent study 37 that reported ANK1 as differentially expressed in bone and will investigate the possibility of additional mechanisms by which ANK1 could contribute to SONFH in the future.
Figure 4.

The diagram shows ANK1 and EPB41 as susceptible genes for SONFH. The effects of ANK1 and EPB41 variants lie on a phenotypic spectrum, ranging from nondamaging and asymptomatic on one end of the spectrum to severe clinical hemolytic anemia on the other end of the spectrum. The variants identified in our study represent a milder intermediate phenotype. The carriers of a single altered copy (heterozygous state) demonstrate increased osmotic fragility and reduced red blood cell deformability compared to noncarriers, which do not produce hemolysis or anemia but represent a subclinical phenotype. However, this partial impairment appears sufficient to predispose individuals to the development of osteonecrosis in the presence of the second hit (ie, glucocorticoid exposure). RBC, red blood cell; SONFH, steroid‐induced osteonecrosis of the femoral head.
SONFH is a multifactorial disease partly resulting from genetic factors, although its pathoetiology is still not fully elucidated. To date, some gene polymorphisms have been identified to be associated with SONFH, however, usually by the biased screening of genes previously found in either SLE or SONFH. For example, NOS3 (encoding nitric oxide synthase 3, associated with ONFH) 38 and CR2 (encoding complement C3d receptor 2, associated with SLE) 39 were identified in a Korean SLE‐SONFH cohort with both healthy and SLE controls by TaqMan Assay. A recent genome‐wide association study of Japanese and Korean SLE‐ONFH cohorts discovered three novel candidates MIR4293 (encoding microRNA 4293), NAALAD2 (encoding N‐acetylated alpha‐linked acidic dipeptidase 2), and MYO16 (encoding myosin XVI); these variants were located in nonprotein coding regions, and their biologic functions in SONFH still need to be further validated. 40 Collectively, prior studies have neither systematically assessed the burden of rare coding‐region variants nor identified cytoskeleton‐related susceptibility genes for SONFH.
To identify novel susceptible genes for SONFH, we applied approaches different from those in the previous studies. First, we increased the possibility by recruiting patients with SLE‐SONFH with GC administration and confirmed the robustness of the associations via multiple adjusting analytical approaches. The overall prevalence of ONFH is less than 1% worldwide. 19 However, we previously investigated the cumulative incidence of SONFH events in a cohort of patients with SLE exposed to GCs, which reached 9.1% (ranging from 4.2% to 43.1%). 9 Therefore, the ANK1 and EPB41 variants were highly enriched (10.8%, 9 of 83) in the case‐control study, although their allele frequencies in the general population are very low according to the gnomAD or the HUABIAO database (Supplementary Table S10). The elevated SONFH incidence in Chinese populations, evidenced by current biomedical and epidemiologic data, 7 , 19 , 41 may partly reflect the increased allele frequencies of such rare variants (eg, ANK1, EPB41, etc) in Han Chinese individuals relative to other ethnic groups in the world (Supplementary Table S10). Second, we analyzed the WES data with a primary focus on rare damaging variants, which significantly alter protein function and are considered more likely to be causative than other types of genetic variants. 42 , 43 Third, we also validated the results in an SONFH cohort of 246 non‐SLE patients treated with steroids. We identified 10 cases (4.1%) of ANK1 and EPB41 variants (Supplementary Table S7). Taken together, we identified ANK1 and EPB41 as two pathogenic susceptibility genes for SONFH.
The variants identified in this study were not associated with prominent clinical dominant features, and such recessive silence is often overlooked by physicians. A query of private insurance claims in the United States between 2012 and 2014 revealed that up to 21% of patients received at least one prescription for GCs. 44 The discovery of novel genetic susceptibility to steroid‐induced osteonecrosis is highly significant, given the widespread use of GCs. Many of their adverse effects are already addressed early in treatment, such as the use of Pneumocystis jirovecii prophylaxis and measures to prevent osteoporosis, gastroduodenal, and other systemic complications. 45 Therefore, identifying such genetic risk factors for osteonecrosis offers clinicians an additional pharmacogenomic tool to predict which patients may develop this complication. This could guide a more personalized approach, favoring steroid‐sparing strategies for those at higher risk.
Our study had several limitations. First, the data sets suffered from small sample size and some variable mismatches due to limited patient sample availability. To strengthen the reliability, we employed external healthy controls, functional enrichment analysis of the gene set, and the adjusted P values with Bonferroni or FDR. This limitation could be ultimately addressed by large‐scale, multicenter cohort studies in future. Second, ANK1 and EPB41 variants were identified in Han Chinese individuals. The results need to be re‐examined in different ethnicities. Lastly, our functional evaluation was limited to a subset of morphologic, protein expression, and osmotic features of erythrocytes in vitro. More mechanistic validations in different tissues and organs including in vivo experiments should be performed in future.
In conclusion, we identified the rare variants in ANK1 and EPB41 as novel pathogenic genetic risk factors for SONFH. This finding sheds light on the pathogenesis and pharmacogenetics of SONFH and provides potential targets for prevention, screening, and risk stratification when using GC treatment in the future.
AUTHOR CONTRIBUTIONS
All authors contributed to at least one of the following manuscript preparation roles: conceptualization AND/OR methodology, software, investigation, formal analysis, data curation, visualization, and validation AND drafting or reviewing/editing the final draft. As corresponding author, Dr C. Zhang confirms that all authors have provided the final approval of the version to be published and takes responsibility for the affirmations regarding article submission (eg, not under consideration by another journal), the integrity of the data presented, and the statements regarding compliance with institutional review board/Declaration of Helsinki requirements.
Supporting information
Disclosure Form:
Figure S1. Workflow for design and analysis of rare damaging variants in this study.
Figure S2: Early ONFH of stage I diagnosed by MRI.
Figure S3: Quality analysis of the WES data of the case‐control study.
Figure S4: The alignment of ANK1 and EPB41 protein sequences by constraint‐based multiple alignment tool (COBALT).
Figure S5: Sanger sequencing that validated ANK1 and EPB41 variants in the case‐control study.
Figure S6: The pedigree trees of case 1‐2 and case 1‐7 in the case‐control study.
Figure S7: Sanger sequencing that validated ANK1 and EPB41 variants in the cohort of SONFH without SLE diseases.
Figure S8: Label free analysis of red blood cell membrane disorders by imaging flow cytometry.
Figure S9: The representative results of FCM‐OFT.
Table S1: Frequency distribution of previously reported candidate genes linked to ONFH according to rare variant criteria in the study.
Table S2: The total rare damaging variants associated with SONFH which were identified in the study.
Table S3: The burden of the rare damaging variants associated with SONFH analyzed by Firth's logistic regression.
Table S4: The rare damaging variants of ANK1 and EPB41 detected in the case‐control study.
Table S5: Primers for sanger sequencing.
Table S6: Characteristics of the validated cohort of 246 non‐SLE SONFH.
Table S7: Association of ANK1 and EPB41 variants with SONFH in the study.
Table S8: Individual clinical characteristics of ANK1 and EPB41 variants in the study.
Table S9: The “% residual red blood cells” calculated in FCM‐OFT.
Table S10: The frequency distribution of rare variations of ANK1 and EPB41 among major global populations (ethnic groups).
Supplementary Appendix 2. Gene‐set burden analysis adjusting for covariates based on 29 rare damaging variants associated with SONFH.
ACKNOWLEDGMENTS
We thank all individuals who participated in this study. We thank Ethan Tyler in the NIH Medical Arts department for modifying the diagram of Figure 4. We thank Yaoyi Zhu, Haiyan He, and Qinqing Liu at Six People's Hospital and Ping Ye at Renji Hospital for collecting the blood samples and follow‐up, and we also thank Prof Guoqing Zhang (Bio‐Med Big Data Center, Shanghai Institute of Nutrition and Health, Chinese Academy of Sciences) and Dr. Lichun Jiang (Biomedical big data platform, The Shanghai Institute for Advanced Immunochemical Studies, ShanghaiTech University) for their academic consultation and support in this study.
Supported by the Shanghai Municipal Health Commission Joint Research Project (grant 2014ZYJB0301), Shanghai Municipal Health Commission key priority discipline project (Shanghai Spinal Disease and Trauma Orthopedics Research Center, grant 2022ZZ01014), and the Three‐Year Action Project in Municipal Hospitals of Shanghai Hospital Development Center (grant SHDC2023CRS049).
1Department of Orthopaedic Surgery and Institute of Microsurgery on Extremities, Shanghai Sixth People's Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, National Center for Orthopaedics, Shanghai, China; 2Shanghai Institute for Advanced Immunochemical Studies, ShanghaiTech University, Shanghai, China; 3Guangzhou Women and Children's Medical Center, Guangzhou Medical University, Guangzhou, China; 4Clinical Research Center, Shanghai Sixth People's Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, China; 5National Institute of Diabetes and Digestive and Kidney Diseases, NIH, Bethesda, Maryland; 6Department of Rheumatology, Shanghai Jiao Tong University School of Medicine Affiliated Renji Hospital, Shanghai, China; 7Department of Laboratory Medicine, Shanghai Sixth People's Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, China; 8National Clinical Research Center for Blood Diseases, State Key Laboratory of Experimental Hematology, Haihe Laboratory of Cell Ecosystem, Institute of Hematology & Blood Diseases Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, Tianjin, China; 9National Clinical Research Center for Hematologic Diseases, Jiangsu Institute of Hematology. Institute of Blood and Marrow Transplantation, Collaborative Innovation Center of Hematology, The First Affiliated Hospital of Soochow University, Suzhou, China; 10Eunice Kennedy Shriver National Institute of Child Health and Human Development, NIH, Bethesda, Maryland; 11Neuro‐Oncology Branch, National Cancer Institute, Center for Cancer Research, NIH, Bethesda, Maryland.
Drs Shengbao Chen, Xianmin Zhu, Zhongzhong Chen, and Wei Chen are co‐first authors and contributed equally to this work. Drs Dongxu Jin, Chunde Bao, and Zhengping Zhuang also contributed equally to this work.
Drs Dongxu Jun, Chunde Bao, and Zhengping Zhuang are co‐corresponding authors.
Additional supplementary information cited in this article can be found online in the Supporting Information section (https://acrjournals.onlinelibrary.wiley.com/doi/10.1002/art.70153).
Author disclosures are available at https://onlinelibrary.wiley.com/doi/10.1002/art.70153.
[Correction added on 27 July 2026, after first online publication: The article title has been corrected.]
Correction added on 9 September 2026, after first online publication: Dongxu Jun, Chunde Bao, and Zhengping Zhuang have been designated as co‐corresponding authors in this version.
References
- 1. Chang C, Greenspan A, Gershwin ME. The pathogenesis, diagnosis and clinical manifestations of steroid‐induced osteonecrosis. J Autoimmun 2020;110:102460. doi: 10.1016/j.jaut.2020.102460 [DOI] [PubMed] [Google Scholar]
- 2. Xiang D, Zhou Y, Wang H, et al. Exosomal SPRY4 from adipogenic BMSCs impairs angiogenesis via the PTPRB/TIE2/PI3K axis in Steroid‐induced osteonecrosis of the femoral head. Stem Cell Res Ther 2025;16(1):375. doi: 10.1186/s13287-025-04505-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Lamb JN, Holton C, O'Connor P, et al. Avascular necrosis of the hip. BMJ 2019;365:l2178. doi: 10.1136/bmj.l2178 [DOI] [PubMed] [Google Scholar]
- 4. Zhao DW, Yu M, Hu K, et al. Prevalence of nontraumatic osteonecrosis of the femoral head and its associated risk factors in the Chinese population: results from a nationally representative survey. Chin Med J (Engl) 2015;128(21):2843–2850. doi: 10.4103/0366-6999.168017 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Kang JS, Park S, Song JH, et al. Prevalence of osteonecrosis of the femoral head: a nationwide epidemiologic analysis in Korea. J Arthroplasty 2009;24(8):1178–1183. doi: 10.1016/j.arth.2009.05.022 [DOI] [PubMed] [Google Scholar]
- 6. Yamaguchi R, Yamamoto T, Motomura G, et al. Incidence of nontraumatic osteonecrosis of the femoral head in the Japanese population. Arthritis Rheum 2011;63(10):3169–3173. doi: 10.1002/art.30484 [DOI] [PubMed] [Google Scholar]
- 7. Cooper C, Steinbuch M, Stevenson R, et al. The epidemiology of osteonecrosis: findings from the GPRD and THIN databases in the UK. Osteoporos Int 2010;21(4):569–577. doi: 10.1007/s00198-009-1003-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Mont MA, Salem HS, Piuzzi NS, et al. Nontraumatic osteonecrosis of the femoral head: where do we stand today?: A 5‐year update. J Bone Joint Surg Am 2020;102(12):1084–1099. doi: 10.2106/JBJS.19.01271 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Chen S, Cai Q, Xu Y, et al. Associations between glucocorticoids, antiphospholipid antibodies and femur head necrosis in patients with SLE: a directed acyclic graph‐based multicentre study. Ther Adv Musculoskelet Dis 2021;13:1759720X211002677. doi: 10.1177/1759720X211002677 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Zhao Z, Xue Y, Hong D, et al. Polymorphisms in the glucocorticoid receptor gene and associations with glucocorticoid‐induced avascular osteonecrosis of the femoral head. Genet Test Mol Biomarkers 2017;21(5):322–327. doi: 10.1089/gtmb.2016.0260 [DOI] [PubMed] [Google Scholar]
- 11. Zhou Z, Hua Y, Liu J, et al. Association of ABCB1/MDR1 polymorphisms in patients with glucocorticoid‐induced osteonecrosis of the femoral head: evidence for a meta‐analysis. Gene 2015;569(1):34–40. doi: 10.1016/j.gene.2015.03.023 [DOI] [PubMed] [Google Scholar]
- 12. Tamura K, Nakajima S, Hirota Y, et al. Genetic association of a polymorphism of the cAMP‐responsive element binding protein‐binding protein with steroid‐induced osteonecrosis after kidney transplantation. J Bone Miner Metab 2007;25(5):320–325. doi: 10.1007/s00774-007-0770-z [DOI] [PubMed] [Google Scholar]
- 13. French D, Hamilton LH, Mattano LA Jr, et al; Children's Oncology Group . A PAI‐1 (SERPINE1) polymorphism predicts osteonecrosis in children with acute lymphoblastic leukemia: a report from the Children's Oncology Group. Blood 2008;111(9):4496–4499. doi: 10.1182/blood-2007-11-123885 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Yu Y, Zhang Y, Wu J, et al. Genetic polymorphisms in IL1B predict susceptibility to steroid‐induced osteonecrosis of the femoral head in Chinese Han population. Osteoporos Int 2019;30(4):871–877. doi: 10.1007/s00198-019-04835-9 [DOI] [PubMed] [Google Scholar]
- 15. Liu YF, Chen WM, Lin YF, et al. Type II collagen gene variants and inherited osteonecrosis of the femoral head. N Engl J Med 2005;352(22):2294–2301. doi: 10.1056/NEJMoa042480 [DOI] [PubMed] [Google Scholar]
- 16. Mah W, Sonkusare SK, Wang T, et al. Gain‐of‐function mutation in TRPV4 identified in patients with osteonecrosis of the femoral head. J Med Genet 2016;53(10):705–709. doi: 10.1136/jmedgenet-2016-103829 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Flannick J, Mercader JM, Fuchsberger C, et al; Broad Genomics Platform; DiscovEHR Collaboration; CHARGE; LuCamp; ProDiGY; GoT2D; ESP; SIGMA‐T2D; T2D‐GENES; AMP‐T2D‐GENES . Exome sequencing of 20,791 cases of type 2 diabetes and 24,440 controls. Nature 2019;570(7759):71–76. doi: 10.1038/s41586-019-1231-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Hao M, Pu W, Li Y, et al. The HuaBiao project: whole‐exome sequencing of 5000 Han Chinese individuals. J Genet Genomics 2021;48(11):1032–1035. doi: 10.1016/j.jgg.2021.07.013 [DOI] [PubMed] [Google Scholar]
- 19. Zhao D, Zhang F, Wang B, et al. Guidelines for clinical diagnosis and treatment of osteonecrosis of the femoral head in adults (2019 version). J Orthop Translat 2020;21:100–110. doi: 10.1016/j.jot.2019.12.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Belot A, Rice GI, Omarjee SO, et al; FREX Consortium; GENIAL Investigators; UK JSLE Study Group . Contribution of rare and predicted pathogenic gene variants to childhood‐onset lupus: a large, genetic panel analysis of British and French cohorts. Lancet Rheumatol 2020;2(2):e99–e109. doi: 10.1016/S2665-9913(19)30142-0 [DOI] [PubMed] [Google Scholar]
- 21. Hop PJ, Lai D, Keagle PJ, et al; Project MinE ALS Sequencing Consortium . Systematic rare variant analyses identify RAB32 as a susceptibility gene for familial Parkinson's disease. Nat Genet 2024;56(7):1371–1376. doi: 10.1038/s41588-024-01787-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Li YJ, Phillips EJ, Dellinger A, et al; Drug‐induced Liver Injury Network . Human leukocyte antigen B*14:01 and B*35:01 are associated with trimethoprim‐sulfamethoxazole induced liver injury. Hepatology 2021;73(1):268–281. doi: 10.1002/hep.31258 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Bonfiglio F, Lasorsa VA, Cantalupo S, et al. Inherited rare variants in homologous recombination and neurodevelopmental genes are associated with increased risk of neuroblastoma. EBioMedicine 2023;87:104395. doi: 10.1016/j.ebiom.2022.104395 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. King MJ, Garçon L, Hoyer JD, et al. ICSH guidelines for the laboratory diagnosis of nonimmune hereditary red cell membrane disorders. Int J Lab Hematol 2015;37(3):304–325. doi: 10.1111/ijlh.12335 [DOI] [PubMed] [Google Scholar]
- 25. Vallese F, Kim K, Yen LY, et al. Architecture of the human erythrocyte ankyrin‐1 complex. Nat Struct Mol Biol 2022;29(7):706–718. doi: 10.1038/s41594-022-00792-w [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Perrotta S, Gallagher PG, Mohandas N. Hereditary spherocytosis. Lancet 2008;372(9647):1411–1426. doi: 10.1016/S0140-6736(08)61588-3 [DOI] [PubMed] [Google Scholar]
- 27. Mohandas N, Gallagher PG. Red cell membrane: past, present, and future. Blood 2008;112(10):3939–3948. doi: 10.1182/blood-2008-07-161166 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Ware RE, de Montalembert M, Tshilolo L, et al. Sickle cell disease. Lancet 2017;390(10091):311–323. doi: 10.1016/S0140-6736(17)30193-9 [DOI] [PubMed] [Google Scholar]
- 29. Darbari DS, Fasano RS, Minniti CP, et al. Severe vaso‐occlusive episodes associated with use of systemic corticosteroids in patients with sickle cell disease. J Natl Med Assoc 2008;100(8):948–951. Published online August 01, 2008 doi: 10.1016/S0027-9684(15)31410-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Afshar A. A case of Kienböck's disease and hereditary spherocytosis anaemia. J Hand Surg Eur Vol 2012;37(5):466–467. doi: 10.1177/1753193412437636 [DOI] [PubMed] [Google Scholar]
- 31. Schilling RF, Gangnon RE, Traver MI. Delayed adverse vascular events after splenectomy in hereditary spherocytosis. J Thromb Haemost 2008;6(8):1289–1295. doi: 10.1111/j.1538-7836.2008.03024.x [DOI] [PubMed] [Google Scholar]
- 32. Boas LV, Faustino V, Lima R, et al. Assessment of the deformability and velocity of healthy and artificially impaired red blood cells in narrow polydimethylsiloxane (PDMS) microchannels. Micromachines (Basel) 2018;9(8):384. doi: 10.3390/mi9080384 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Arakawa C, Gunnarsson C, Howard C, et al. Biophysical and biomolecular interactions of malaria‐infected erythrocytes in engineered human capillaries. Sci Adv 2020;6(3):eaay7243. doi: 10.1126/sciadv.aay7243 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Pernow J, Mahdi A, Yang J, et al. Red blood cell dysfunction: a new player in cardiovascular disease. Cardiovasc Res 2019;115(11):1596–1605. doi: 10.1093/cvr/cvz156 [DOI] [PubMed] [Google Scholar]
- 35. Mont MA, Cherian JJ, Sierra RJ, et al. Nontraumatic osteonecrosis of the femoral head: where do we stand today? A ten‐year update. J Bone Joint Surg Am 2015;97(19):1604–1627. doi: 10.2106/JBJS.O.00071 [DOI] [PubMed] [Google Scholar]
- 36. Lu Y, Wang Y, Xin Q, et al. Metabolic and proteomic profiling of coronary microvascular dysfunction: insights from rat models. Biomolecules 2024;14(10):1305. doi: 10.3390/biom14101305 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Goldoni I, Ibelli AMG, Fernandes LT, et al. Comprehensive analyses of bone and cartilage transcriptomes evince ion transport, inflammation and cartilage development‐related genes involved in chickens’ femoral head separation. Animals (Basel) 2022;12(6):788. doi: 10.3390/ani12060788 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Kim HS, Bae SC, Kim TH, et al. Endothelial nitric oxide synthase gene polymorphisms and the risk of osteonecrosis of the femoral head in systemic lupus erythematosus. Int Orthop 2013;37(11):2289–2296. doi: 10.1007/s00264-013-1966-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Kim TH, Bae SC, Lee SH, et al. Association of complement receptor 2 gene polymorphisms with susceptibility to osteonecrosis of the femoral head in systemic lupus erythematosus. BioMed Res Int 2016;2016:9208035. doi: 10.1155/2016/9208035 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Suetsugu H, Kim K, Yamamoto T, et al. Novel susceptibility loci for steroid‐associated osteonecrosis of the femoral head in systemic lupus erythematosus. Hum Mol Genet 2022;31(7):1082–1095. doi: 10.1093/hmg/ddab306 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Sodhi N, Acuna A, Etcheson J, et al. Management of osteonecrosis of the femoral head. Bone Joint J 2020;102‐B(7_Supple_B):122–128. doi: 10.1302/0301-620X.102B7.BJJ-2019-1611.R1 [DOI] [PubMed] [Google Scholar]
- 42. Chen Z, Finnell RH, Lei Y, et al. Progress and clinical prospect of genomic structural variants investigation. Sci Bull (Beijing) 2024;69(6):705–708. doi: 10.1016/j.scib.2024.01.035 [DOI] [PubMed] [Google Scholar]
- 43. Chen Z, Lei Y, Zheng Y, et al. Threshold for neural tube defect risk by accumulated singleton loss‐of‐function variants. Cell Res 2018;28(10):1039–1041. doi: 10.1038/s41422-018-0061-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Waljee AK, Rogers MA, Lin P, et al. Short term use of oral corticosteroids and related harms among adults in the United States: population based cohort study. BMJ 2017;357:j1415. doi: 10.1136/bmj.j1415 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Malpica L, Moll S. Practical approach to monitoring and prevention of infectious complications associated with systemic corticosteroids, antimetabolites, cyclosporine, and cyclophosphamide in nonmalignant hematologic diseases. Hematology (Am Soc Hematol Educ Program) 2020;2020(1):319–327. doi: 10.1182/hematology.2020000116 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Disclosure Form:
Figure S1. Workflow for design and analysis of rare damaging variants in this study.
Figure S2: Early ONFH of stage I diagnosed by MRI.
Figure S3: Quality analysis of the WES data of the case‐control study.
Figure S4: The alignment of ANK1 and EPB41 protein sequences by constraint‐based multiple alignment tool (COBALT).
Figure S5: Sanger sequencing that validated ANK1 and EPB41 variants in the case‐control study.
Figure S6: The pedigree trees of case 1‐2 and case 1‐7 in the case‐control study.
Figure S7: Sanger sequencing that validated ANK1 and EPB41 variants in the cohort of SONFH without SLE diseases.
Figure S8: Label free analysis of red blood cell membrane disorders by imaging flow cytometry.
Figure S9: The representative results of FCM‐OFT.
Table S1: Frequency distribution of previously reported candidate genes linked to ONFH according to rare variant criteria in the study.
Table S2: The total rare damaging variants associated with SONFH which were identified in the study.
Table S3: The burden of the rare damaging variants associated with SONFH analyzed by Firth's logistic regression.
Table S4: The rare damaging variants of ANK1 and EPB41 detected in the case‐control study.
Table S5: Primers for sanger sequencing.
Table S6: Characteristics of the validated cohort of 246 non‐SLE SONFH.
Table S7: Association of ANK1 and EPB41 variants with SONFH in the study.
Table S8: Individual clinical characteristics of ANK1 and EPB41 variants in the study.
Table S9: The “% residual red blood cells” calculated in FCM‐OFT.
Table S10: The frequency distribution of rare variations of ANK1 and EPB41 among major global populations (ethnic groups).
Supplementary Appendix 2. Gene‐set burden analysis adjusting for covariates based on 29 rare damaging variants associated with SONFH.
Data Availability Statement
The analysis data sets supporting the conclusions of this article are included within the article and its supplementary appendixes. Additional data are available through reasonable request to the corresponding author.
