# Novel Hypotheses: AD Polygenic Risk Scores and Rare Variant Burden
---
## Hypothesis 1: Synergistic Epistasis Between Rare Variants and Polygenic Risk
**Description:** Rare functional variants in AD-risk genes (TREM2, ABCA7, PLCG2) do not merely add to polygenic risk but exhibit epistatic interaction with PRS through multiplicative enhancement of effect sizes. Specifically, rare variant carriers with high PRS show disproportionately elevated risk beyond what additive models predict, driven by convergence on microglial pathways amplifying amyloid pathology and neuroinflammation.
**Target genes/proteins:** TREM2, ABCA7, PLCG2, microglial signaling network
**Confidence score:** 0.72
**Evidence basis:** TREM2 R47H carriers show ~3-fold increased AD risk (Guerreiro et al., 2013); PLCG2 rare variants show protective effects (Sims et al., 2017); ABCA7 loss-of-function variants enriched in AD cases (Wollmer et al., 2003). These converge on microglial activation states that could synergize with polygenic inflammatory burden.
---
## Hypothesis 2: Rare Variant Burden in Synaptic Pathways Explains PRS Variance in Early-Onset AD
**Description:** Current PRS models underperform for early-onset AD (EOAD <65 years) due to enrichment of rare, highly penetrant variants in synaptic genes (SNAP25, SYT1, Complexin family) that bypass polygenic load calculations. Measuring rare variant burden in synaptic transmission pathways will capture this variance, improving prediction specifically for EOAD where common variant burden is relatively less deterministic.
**Target genes/proteins:** Synaptic vesicle release machinery, postsynaptic density proteins
**Confidence score:** 0.64
**Evidence basis:** EOAD cases show higher rates of monogenic causes (PSEN1, PSEN2, APP); synaptic dysfunction is established downstream of amyloid (Shankar et al., 2008); heritability of EOAD exceeds late-onset AD (Mosconi et al., 2004).
---
## Hypothesis 3: Ancestral Differential Rare Variant Architecture Underlies PRS Performance Disparity
**Description:** The well-documented reduction in PRS predictive accuracy in non-European ancestry cohorts is partially explained by population-specific rare variant burdens in AD-relevant genes that are not captured in European-ancestry GWAS. Specifically, African ancestry populations carry rare variants in AD genes with distinct allele frequencies and effect sizes, creating genetic risk architecture unaccounted for by PRS built on European-ancestry summary statistics.
**Target genes/proteins:** ABCA7 (shows population-specific variants), APOE region complexity, CLU
**Confidence score:** 0.68
**Evidence basis:** ABCA7 null variants show higher frequency in African ancestry (Genin et al., 2018); APOE ε4 has differential effect sizes across ancestries; PRS portability is consistently reduced in non-European cohorts (Martin et al., 2019).
---
## Hypothesis 4: Rare Variant Burden in Myeloid Genes Identifies PRS Non-Responders
**Description:** A subset of individuals with high PRS remain cognitively healthy into late life ("PRS non-responders"), while others with low PRS develop AD. Rare variant burden in genes regulating myeloid cell function (TREM2, PLCG2, SPI1) identifies these subgroups: non-responders carry protective rare variants that enhance microglial amyloid clearance capacity, counteracting polygenic inflammatory risk.
**Target genes/proteins:** TREM2, PLCG2, SPI1 (PU.1 transcription factor), TYROBP/DAP12
**Confidence score:** 0.75
**Evidence basis:** TREM2 haploinsufficiency impairs microglial clustering around amyloid plaques (Wang et al., 2016); PLCG2 P522R variant shows protective effect (Sims et al., 2017); microglial states determine amyloid clearance efficiency (Parhizkar et al., 2019).
---
## Hypothesis 5: Temporal Threshold Model: Rare Variants Accelerate Age-Dependent PRS Effects
**Description:** The relationship between PRS and AD risk is not linear but follows an age-dependent threshold model where rare variants shift the inflection point of risk acceleration. High PRS individuals without rare variants show gradual risk increase after age 65, while rare variant carriers in the same PRS stratum demonstrate steeper risk curves with earlier onset, explaining why PRS predictive accuracy peaks in specific age ranges.
**Target genes/proteins:** Any pathogenic rare variant combined with PRS architecture
**Confidence score:** 0.61
**Evidence basis:** APOE ε4 shows age-dependent effects (Michels et al., 2021); rare variant carriers demonstrate earlier onset in familial AD genes; age-stratified PRS analyses show variable performance.
---
## Hypothesis 6: Multi-Ancestry PRS Plus Rare Variant Burden Creates Clinically Actionable Thresholds
**Description:** Integrating PRS from multiple ancestries with rare variant burden in lipid metabolism genes (APOB, LDLR, ABCA7, APOE) will identify individuals with significantly elevated 5-year conversion risk from MCI to AD who would be missed by single-ancestry PRS. This composite score will achieve AUC >0.85, exceeding current clinical prediction benchmarks.
**Target genes/proteins:** ABCA7, APOE, LDLR family, lipid metabolism genes
**Confidence score:** 0.58
**Evidence basis:** Lipid dysregulation is central to AD pathogenesis; APOE ε4 affects lipid transport (Luttinger et al.); combination genetic scores show improved prediction in cardiovascular disease (Nagai et al., 2020).
---
## Hypothesis 7: Endophenotype-Specific Prediction: Rare Variants Differentially Modify PRS for Biomarker vs. Clinical Outcomes
**Description:** Rare variant burden modifies PRS prediction accuracy differentially depending on the outcome measured. Rare variants in TREM2/ABCA7 enhance PRS prediction of amyloid PET positivity but not cognitive decline, while rare variants in synaptic genes (NLGN3, SHANK3) enhance prediction of cognitive trajectories but not biomarker changes. This explains inconsistent PRS-biomarker associations and suggests pathway-specific prediction models.
**Target genes/proteins:** TREM2 (microglial/amyloid), NLGN3/NLGN4 (synaptic/cognition), CNTNAP2 (neural development)
**Confidence score:** 0.67
**Evidence basis:** TREM2 variants correlate with amyloid burden in PET studies (Leyns et al., 2019); synaptic gene variants affect cognition independent of amyloid (Scheefhals et al., 2019); dissociation between biomarkers and cognition documented in AD literature.
---
## Summary Table
| Hypothesis | Focus | Confidence |
|------------|-------|------------|
| 1 | Epistatic synergy rare variants + PRS | 0.72 |
| 2 | Early-onset AD synaptic rare burden | 0.64 |
| 3 | Ancestral rare variant architecture | 0.68 |
| 4 | Protective variants in PRS non-responders | 0.75 |
| 5 | Temporal threshold acceleration model | 0.61 |
| 6 | Multi-ancestry lipid metabolism integration | 0.58 |
| 7 | Endophenotype-specific modification | 0.67 |
**Predicted testing strategies:** Stratified survival analysis by rare variant carrier status; interaction term modeling in Cox proportional hazards; pathway burden score construction using burden tests (SKAT-O, VT); ancestral population-specific validation cohorts; longitudinal biomarker × genotype correlation studies.