PSEN1 null alleles fail to match the expected Alzheimer’s phenotype
The case for PSEN1 loss-of-function as a direct cause of early-onset Alzheimer’s disease has taken a sharp turn. In a preprint examining population and biobank datasets, investigators identified 45 carriers of predicted PSEN1 loss-of-function variants expected to trigger nonsense-mediated decay, yet none had a recorded case of early-onset dementia. Seventeen of those carriers were older than 65 years at last follow-up and still had no documented dementia or Alzheimer’s disease. The pattern is difficult to reconcile with a model in which simple absence of presenilin-1 protein is sufficient to produce familial early-onset disease.
The implication is biologically specific rather than merely statistical. Presenilin-1 is central to gamma-secretase function and amyloid precursor protein processing, so the classic assumption has been that damaging PSEN1 variation should map directly onto neurodegeneration. These data argue that the pathogenic mechanism is not a clean gene knockout. Instead, disease appears to require a mutant protein product with altered function, or at minimum a transcript capable of escaping full degradation and leaving behind an abnormal presenilin-1 species.
A rare stop-gain variant may not have been a true null allele
The most important challenge to the earlier loss-of-function model comes from re-analysis of the reported stop-gain variant p.K109*. On closer inspection, that variant was associated with a cryptic splice site predicted to generate an in-frame deletion of five amino acids, p.K109_L113del, rather than a pure null state. In practical terms, the variant may not abolish the protein at all; it may instead reshape it just enough to alter downstream biology in a way that can drive Alzheimer’s pathology.
One sentence matters here: a true null allele did not separate from the phenotype seen with the splice-altered variant.
That distinction is consequential for variant interpretation in clinical genetics. If a stop-gain annotation is assumed to equal complete loss of function, clinicians and laboratorians may overcall disease causality in families carrying variants that actually preserve an abnormal presenilin-1 product. The study’s central message is narrower and more actionable than the headline suggests: PSEN1 variants must produce an altered protein product to cause early-onset Alzheimer’s disease, while true null alleles do not appear to do so.
Population databases push against a simple gene-dose model
The strength of the analysis lies in its use of multiple large datasets: the Alzheimer’s Disease Sequencing Project, UK Biobank, and BioVU. Those resources make it possible to look beyond a single family or a single published case and ask what happens when predicted PSEN1 null alleles occur in the broader population. The absence of recorded early-onset dementia across 45 carriers is hard to dismiss, particularly when the cohort included older individuals who had reached their seventh decade without documented Alzheimer’s disease.
This does not prove that every null allele is benign in every context. It does, however, make a strong case that the disease mechanism is not simply haploinsufficiency. For a gene long linked to autosomal dominant early-onset Alzheimer’s disease, that is a meaningful biological correction. A monoallelic knockout model would usually be expected to show at least some signal of age-related neurodegeneration if presenilin dosage alone were enough to tip the system. The lack of such a signal suggests that pathogenicity depends on qualitative changes in the protein, not merely quantitative reduction.
Why presenilin biology points toward altered function rather than absence
Presenilin-1 sits at the core of a proteolytic machinery that shapes amyloid precursor protein cleavage. A complete absence of the protein would be expected to disrupt gamma-secretase activity broadly, but the human phenotype described here does not look like that route to disease. The data fit better with a model in which specific amino-acid substitutions, truncations, or in-frame deletions generate a dysfunctional protein that perturbs processing in a dominant manner.
In Alzheimer’s genetics, this distinction has immediate bedside consequences. A variant labeled as predicted loss of function cannot automatically be treated as causal for familial early-onset disease. Variant curation now has to consider whether the transcript is rescued by splicing, whether the protein is still produced, and whether the molecular consequence is truly null or instead structurally altered. In a gene where a single misannotated variant can distort counseling, that nuance has real clinical weight.
Genetic counseling will need a stricter reading of PSEN1 annotations
Families carrying rare PSEN1 variants often face intense uncertainty, especially when the variant appears disruptive on paper. This study argues for a more disciplined interpretation of stop-gain calls and other predicted null alleles. A variant that predicts nonsense-mediated decay may not actually behave that way in vivo, and the biological consequence may be very different from what sequence annotation alone suggests.
For counseling, the practical message is restraint. A predicted null allele in PSEN1 should not be assumed to explain a family history of early-onset Alzheimer’s disease without additional molecular evidence. Functional follow-up, transcript analysis, and careful review of splicing effects become essential when the stakes include risk prediction, cascade testing, and reproductive planning. The finding also narrows the search for pathogenic mechanisms in this gene: the most credible disease-causing variants are those that alter presenilin-1 protein structure or function, not those that merely eliminate the protein product.
The preprint strengthens one conclusion while leaving others open
Because this is a preprint, the analysis still awaits peer review. The underlying datasets are powerful, but the study remains observational, and phenotype capture in large biobanks can miss undiagnosed or incompletely documented dementia. Even so, the direction of the evidence is consistent across resources and supported by the re-interpretation of the p.K109* variant.
What remains unresolved is the full boundary between benign null alleles and pathogenic altered-function alleles across the PSEN1 spectrum. The study does not settle every mechanistic question, but it does force a revised reading of the gene-disease relationship. Early-onset Alzheimer’s disease associated with PSEN1 does not appear to be a simple consequence of losing the gene entirely; the pathogenic event seems to require a protein product that is still there, but no longer normal.