LYFE Sciences · Project HERA
Variant Interpretation · Classification Report
Generated: 2026-07-06
Case ID: NM_002168.3_c.515G_A_20260706_061432
Framework: ACMG/AMP 2015
Variant classification summary

NM_002168.3:c.515G>A

IDH2  · NP_002159.2:p.(Arg172Lys)  · NM_002168.3
GRCh37: chr15:90631838 C>T  ·  GRCh38: chr15:90088606 C>T
Gene: IDH2 Transcript: NM_002168.3
Final call
VUS
PS3 supporting PM1 moderate PM2 supporting PP3 supporting
All criteria require review: For research and educational purposes only.
Variant details
Gene
IDH2
Transcript
NM_002168.3
Protein
NP_002159.2:p.(Arg172Lys)
gnomAD AF
6.19484266958072e-07 (v4.1)
ClinVar
risk factor
OncoKB
Oncogenic
Classification rationale
Interpretation summary
Generated evidence synthesis
1
PM1 (moderate): The c.515G>A (p.Arg172Lys) variant affects a well-established mutational hotspot at residue R172 in the IDH2 active site, confirmed by Cancer Hotspots and recurrently observed in COSMIC (n=467).
2
PM2 (supporting): This variant is extremely rare in population databases, absent from gnomAD v2.1 and gnomAD-Canada, with a single heterozygous observation in gnomAD v4.1 (AF = 6.19e-7; 1/1,614,246 alleles), well below the 0.1% threshold.
3
PS3 (supporting): Multiple independent functional studies across AML, glioma, and cholangiocarcinoma models confirm that IDH2 R172K confers a neomorphic gain-of-function producing the oncometabolite 2-hydroxyglutarate. Enasidenib, a selective IDH2-mutant inhibitor, produces clinical responses in R172K-mutated AML, confirming oncogenic dependency. However, all functional evidence derives from somatic cancer models, limiting strength to supporting in a germline context.
4
PP3 (supporting): REVEL score of 0.733 predicts a deleterious effect on protein function. BayesDel (0.432) does not independently support pathogenicity, but the REVEL prediction provides computational support.
5
Summary: One moderate criterion (PM1) and three supporting criteria (PM2, PS3, PP3) are met. No benign criteria are met. Under generic ACMG/AMP 2015 combination rules (PMID:25741868), this combination does not reach the threshold for Likely Pathogenic (requires 2 moderate OR 1 moderate + 2 supporting). The total evidence of PM1 + PM2 + PS3_supporting + PP3 falls short of a Likely Pathogenic classification. This variant is classified as a Variant of Uncertain Significance (VUS).
6
CAUTION: All pathogenic evidence derives from somatic cancer contexts. IDH2 R172K is a well-established somatic oncogenic driver mutation. Its role in germline disease, if any, is not well characterized. The germline disease context identified for IDH2 (MDS/MPN overlap syndromes, PMID:33275756) does not specifically implicate this variant. Clinicians should interpret this germline VUS classification with the understanding that this variant is oncogenic in somatic tissues.
Final determination: Generic ACMG/AMP 2015 fallback rules do not meet benign, likely benign, likely pathogenic, or pathogenic combination thresholds, so the variant is classified as Variant of Uncertain Significance.
Criteria assessment
ACMG/AMP criteria review
Criteria shown when status is available
All criteria require review: For research and educational purposes only.
Criterion Status Rationale Evidence used
PVS1 N/A This is a missense variant (c.515G>A, p.Arg172Lys). It does not fall into the default generic PVS1 null-variant buckets of nonsense, frameshift, or canonical ±1,2 splice consensus variants per ClinGen SVI PVS1 recommendations (PMC6185798).
pvs1_generic_framework pvs1_variant_assessment
PS1 Not met No alternative nucleotide change at codon 172 (AGA) producing the same p.Arg172Lys substitution has been established as pathogenic. The reference codon AGA at positions c.514-516 can only produce p.Arg172Lys via c.515G>A; no independent nucleotide path to the same amino acid change exists in this reference context.
PS2 Not met No de novo occurrence data available for this variant. No reports of confirmed or unconfirmed de novo observation were identified in ClinVar submissions, literature, or any other source.
PS3 Met Multiple independent functional studies demonstrate that IDH2 R172K produces a neomorphic gain-of-function enzymatic activity, converting α-ketoglutarate to the oncometabolite 2-hydroxyglutarate (2HG). This well-established functional effect has been confirmed across AML (Ward 2010, Paschka 2010), glioma (Jin 2011), and cholangiocarcinoma models (Saha 2014). Enasidenib, a selective IDH2-R172K inhibitor, produces clinical responses in R172K-mutated AML (Stein 2017). However, all functional evidence derives from somatic cancer models and was not designed to assess germline pathogenicity per ACMG/AMP standards, limiting strength assignment to supporting.
PMID:20171147 PMID:21326614 PMID:25043045 PMID:20567020 PMID:28588020 oncokb
PS4 Not met No case-control data demonstrating statistically significant enrichment of this variant in affected individuals versus controls in a germline context. The variant is recurrent in somatic cancers (COSMIC n=467), but somatic prevalence does not satisfy PS4 for germline ACMG/AMP assessment.
PS5 Not met ClinVar lists this variant as 'risk factor' (VariationID: 375987) from a single submitter (Clinical Pathology, Tanta University) with no assertion criteria provided and no valid supporting PMIDs. The single submission uses case-control methodology but lacks ACMG criteria. OncoKB classifies as 'Oncogenic' but this is a somatic cancer knowledgebase, not a germline classification. No reputable germline source has classified this variant as pathogenic.
clinvar oncokb
PM1 Met This variant affects residue R172, located in the active site of IDH2 and a well-established mutational hotspot across multiple cancer types. Cancer Hotspots confirms this residue is statistically significant. COSMIC reports this exact variant 467 times. The R172 residue is the IDH2 analog of the IDH1 R132 hotspot and is one of only two recurrently mutated residues in IDH2 (R140 and R172), both in the enzymatic active site.
PMID:20171147 PMID:20567020
PM2 Met This variant is absent from gnomAD v2.1 and present in gnomAD v4.1 at extremely low frequency (AF = 6.19e-7; 1/1,614,246 alleles, 0 homozygotes). The highest subpopulation frequency is in Ashkenazi Jewish (AF = 3.38e-5; 1/29,606 alleles). These frequencies are well below the 0.1% PM2 threshold for non-VCEP assessment. The variant is also absent from gnomAD-Canada v1.0.
gnomad_v2 gnomad_v4 gnomad_canada
PM5 N/A Automated candidate harvesting did not identify any qualifying comparator variant at residue R172 with a different amino acid change and established pathogenicity. No same-residue pathogenic missense with a different alteration was found in ClinVar or other sources.
pm5_candidates
PM6 Not met No de novo observation data available for this variant, with or without confirmed parentage. No reports of de novo occurrence were identified in ClinVar, literature, or any other source.
PP1 Not met No co-segregation data available for this variant. No family studies or pedigree analyses were identified in any source.
PP2 Not assessed HCI prior not available for IDH2 (gene_not_supported). Without gene-level missense constraint metrics or an established low rate of benign missense variation, PP2 cannot be reliably assessed. While IDH2 is an oncogene where missense mutations at specific hotspots drive disease, the gene-level constraint data needed for formal PP2 application are absent.
PP3 Met REVEL score of 0.733 predicts a deleterious effect on protein function. SpliceAI predicts no significant splice impact (max delta = 0.03), which is neither supportive nor contradictory for a missense variant. BayesDel score of 0.432 falls below the typical pathogenicity threshold (0.5) and does not support a deleterious prediction. Multiple lines of in silico evidence provide partial support for a deleterious effect.
revel bayesdel spliceai
PP4 Not met No specific phenotype or family history data are available for this variant in a germline context. While the variant is associated with multiple cancer types somatically (AML, glioma, IHCC), no germline phenotype specificity data exist.
PP5 Not met ClinVar lists this variant as 'risk factor' from a single submitter with no assertion criteria provided (trust_tier=weak_noisy). PMID:22918138 flagged for PP5 is an Association for Molecular Pathology guideline paper on clinical diagnostic genome sequencing that does not mention this specific variant. No reputable germline source has classified this variant as pathogenic.
clinvar
BA1 Not met The overall allele frequency is 6.19e-7 (0.00006%), far below the 1% BA1 threshold. This variant is not a common polymorphism.
gnomad_v2 gnomad_v4 gnomad_canada
BS1 Not met The overall allele frequency is 6.19e-7 (0.00006%), far below the 0.3% BS1 threshold for non-VCEP assessment. The highest subpopulation frequency (Ashkenazi Jewish, 3.38e-5) is also well below 0.3%. This variant is not observed at a frequency consistent with benign variation.
gnomad_v2 gnomad_v4 gnomad_canada
BS2 Not met No data demonstrating observation of this variant in healthy adult controls at a frequency consistent with benign variation. The single gnomAD v4.1 observation lacks phenotypic annotation and does not constitute BS2-level evidence.
BS3 Not met Well-established functional studies demonstrate that IDH2 R172K has a neomorphic gain-of-function effect (2-hydroxyglutarate production) rather than a benign or neutral effect on protein function. The functional evidence supports a damaging, not benign, interpretation.
PMID:20171147 PMID:21326614 PMID:25043045
BS4 Not met No non-segregation data available. No family studies demonstrating absence of co-segregation with disease were identified.
BP1 N/A IDH2-associated disease, to the extent a germline mechanism exists, involves gain-of-function missense mutations (R140, R172) rather than a truncating-only mechanism. BP1 applies specifically to genes where only truncating variants cause disease, which is not the case for IDH2.
BP2 Not met No data on observation of this variant in trans with a known pathogenic variant. No phase information is available for the single gnomAD observation.
BP4 Not met REVEL score of 0.733 predicts a deleterious effect, not a benign one. BayesDel 0.432 is borderline. SpliceAI (max delta 0.03) predicts no splicing impact, which is neutral. Multiple lines of computational evidence do not support a benign interpretation.
revel bayesdel spliceai
BP5 Not met No alternative molecular basis for disease has been identified in individuals harboring this variant. No data suggesting another causative variant explains the phenotype in carriers.
BP6 Not met This variant has not been reported as benign or likely benign by any reputable source. ClinVar lists it as 'risk factor' (not benign). No expert panel or clinical testing laboratory has classified this variant as benign.
clinvar
BP7 N/A This is a missense variant (c.515G>A, p.Arg172Lys), not a synonymous or intronic variant. BP7 applies only to synonymous variants with no predicted splicing impact. While SpliceAI predicts no significant splice impact (max delta 0.03), BP7 is structurally inapplicable to missense variants.
Disclaimer: The content and results provided by LYFE Sciences are for research and educational purposes only and must not be used as a substitute for professional medical judgment, diagnosis, or treatment. Always consult a qualified healthcare professional before making any clinical decisions.