One Variant, Two Names: How a Numbering Quirk Obscured the Alzheimer’s Risk Gene

A routine genetic test in a Turkish clinic has exposed a surprisingly simple problem lurking inside modern genomic medicine: the same DNA change can carry two entirely different names, depending on how scientists count the amino acids of the protein it alters. In a case report published in Molecular Biology Reports, researchers from Selcuk University and Necmettin Erbakan University in Konya describe a 57-year-old man with early-onset Alzheimer’s disease whose sequencing results initially flagged an unfamiliar variant in the APOE gene, the most important genetic risk factor for late-onset Alzheimer’s. Only after careful cross-checking did the team realize that the mysterious variant was, in fact, the famous epsilon 4-defining change that geneticists have known about for more than four decades. The episode, the authors argue, is a warning about how easily clinically meaningful findings can be mislabeled as uncertain when annotation conventions collide.

The patient at the center of the report underwent next-generation sequencing as part of the diagnostic workup for his early-onset dementia. When the laboratory report came back, it listed a homozygous variant in the APOE gene described as p.Cys130Arg, meaning that the 130th amino acid of the apolipoprotein E protein, a cysteine, had been replaced by an arginine. For a clinician reading the report, this annotation did not immediately correspond to any of the well-known APOE isoforms, and the variant was accordingly classified as a variant of uncertain significance, or VUS. That classification, standard practice under the American College of Medical Genetics and Genomics framework when a variant cannot be confidently linked to disease, effectively told the clinical team that the finding’s meaning was unknown.

The problem, as the researchers discovered when they revisited the result, was not the biology but the bookkeeping. The APOE protein exists in three common human isoforms, epsilon 2, epsilon 3, and epsilon 4, which differ from one another by single amino acid substitutions at two positions. In the protein sequence that has been used in the Alzheimer’s literature since the pioneering sequencing work of the early 1980s, the epsilon 4-defining change is a cysteine-to-arginine swap at position 112, conventionally written as Cys112Arg. But the numbering used in the clinical report counted from a different starting point, placing the identical substitution at position 130 instead. The two labels, p.Cys130Arg and Cys112Arg, describe exactly the same physical change: the rs429358 single nucleotide variant that defines the epsilon 4 allele.

The discrepancy traces back to how the reference sequence is defined. The mature apolipoprotein E protein, as it circulates in plasma and the brain, is 299 amino acids long, but the protein is initially produced with an 18-amino-acid signal peptide that is cleaved off during processing. Depending on whether a numbering scheme counts from the first amino acid of the nascent protein or from the first residue of the mature protein, every position shifts by that offset. The classic Alzheimer’s literature, following the original protein sequencing studies of Weisgraber, Rall, and Mahley, uses mature-protein numbering, so the epsilon 4-defining substitution sits at position 112. Modern transcript-based annotation systems, which describe variants relative to the messenger RNA and the full precursor protein, place the same change at position 130. Both are internally consistent; neither is wrong. The trouble arises when the two systems meet in a clinical report without an explicit bridge between them.

Once the research team integrated information from the transcript annotation, the protein-level description, and the dbSNP database entry, the picture resolved completely. The variant listed as p.Cys130Arg corresponded to rs429358, one of the two single nucleotide polymorphisms that together determine a person’s APOE genotype. Because the patient carried the change in a homozygous state, he was effectively epsilon 4/epsilon 4, the genotype associated with the highest risk of late-onset Alzheimer’s disease and, in population studies, with an earlier average age of onset. What had entered the chart as an uncertain finding was in fact one of the most extensively characterized genetic risk factors in all of human medicine, documented in thousands of studies since a landmark 1997 meta-analysis confirmed the association between APOE genotype and Alzheimer’s disease across age, sex, and ethnic groups.

The clinical stakes of such confusion are considerable. APOE status informs risk counseling, shapes expectations about disease course, and increasingly matters for the emerging generation of Alzheimer’s therapies, some of which carry genotype-specific safety considerations. A variant of uncertain significance, by contrast, is a dead end in the diagnostic narrative: it prompts neither action nor reassurance, and it may trigger unnecessary follow-up testing or family studies. If the epsilon 4-defining variant, arguably the single most recognizable variant in dementia genetics, can be rendered unrecognizable by a numbering convention, the authors suggest, less famous variants are likely to suffer the same fate far more often, silently inflating the VUS rate and eroding clinician confidence in genomic reports.

The case also illustrates a broader and growing pain point in clinical genomics: nomenclature inconsistency across databases and annotation tools. A recent survey of ClinVar pathogenic variants found widespread discrepancies in how the same variants are described by different software pipelines, from differences in transcript selection to divergent syntax conventions. The Human Genome Variation Society has published detailed recommendations for describing sequence variants precisely, but those guidelines do not eliminate the underlying multiplicity of reference sequences, transcripts, and protein isoforms. Each laboratory pipeline makes choices, and each choice produces a label. When a clinician or even a genetics specialist encounters a label that does not match the one they know from the literature, the gap must be bridged manually, transcript by transcript, position by position.

For the APOE gene specifically, the mismatch is particularly treacherous because the epsilon nomenclature itself is a third layer of description layered on top of the others. Clinicians and researchers speak fluently of epsilon 4, but a sequencing pipeline reports nucleotide changes such as c.388T>C relative to a specific transcript, and protein changes relative to a specific reference sequence. The connection between the familiar isoform name and the pipeline output requires exactly the kind of multi-level reasoning the Turkish team performed: checking the transcript annotation, the protein consequence, and the dbSNP identifier, then confirming that all three converge on rs429358. Without that triangulation, the report’s VUS classification would have stood, and the patient’s most significant genetic finding would have remained buried in plain sight.

The authors’ recommendation is straightforward: clinical genomic interpretation should routinely integrate genomic coordinates, transcript-level annotations, protein-level descriptions, and database cross-references rather than relying on any single label. In practice, that means laboratories should map variants to dbSNP identifiers and ClinVar entries as a standard step, and reports should explicitly note when a variant corresponds to a well-known functional allele under an alternative numbering convention. It also means that education matters: the epsilon 4 story is taught in every neuroscience course, but the arithmetic that connects Cys112Arg to p.Cys130Arg is rarely taught anywhere, leaving even experienced professionals vulnerable to the mismatch. As whole-genome and whole-exome sequencing becomes routine in dementia care, such bridging knowledge will only grow more essential.

What makes the case resonate beyond its single patient is the reminder that genomic medicine’s greatest challenges are not always at the frontier of discovery. Sometimes they lie in translation, in the mundane conventions that determine whether decades of accumulated knowledge actually reach the clinic. A 57-year-old man with early-onset Alzheimer’s carried the strongest known genetic risk factor for the disease, documented beyond dispute since 1981, and his clinical report still called it uncertain. Fixing that kind of failure requires no new science, only better alignment between the ways different communities count, name, and describe the same molecular event, and a healthy institutional habit of skepticism whenever a familiar gene produces an unfamiliar label.

Subject of Research: APOE ε4 variant interpretation and the protein numbering discrepancy that led to its misclassification as a variant of uncertain significance in early-onset Alzheimer's disease

Article Title: APOE ε4 variant interpretation and protein numbering discrepancy: a case report

Article References: Dagdelen, B., Ozguncu, C., Zamani, A. G., & Arikoglu, H. (2026). APOE ε4 variant interpretation and protein numbering discrepancy: a case report. Molecular Biology Reports, 53(1), Article 1608. https://doi.org/10.1007/s11033-026-12809-4

Image Credits: AI Generated

DOI: 10.1007/s11033-026-12809-4

Keywords: Alzheimer's disease, APOE ε4, rs429358, protein numbering, variant interpretation, next-generation sequencing, variant of uncertain significance, clinical genomics, genetic nomenclature, dbSNP, early-onset dementia, case report