Part 1 · The disease
Genes and inheritance
Most Parkinson's is not inherited in a simple way. About 15% of people with Parkinson's have a family history, and 5–10% have a single-gene form. 80 A few gene changes raise the chance a lot, many common ones raise it a little, and most people who carry even the strongest of them never develop the disease.
How much of Parkinson's is genetic?
Press play to meet the seven genes now counted as causes, see how many people with Parkinson's carry one, then the common variants that only shift the odds, and finally how many carriers actually become ill.
Odds ratios are on a log scale. Grid squares are rounded to whole people; each grid shows several studies' estimates layered on top of each other.
Seven genes, three patterns
Rare variants in more than 20 genes have been reported to cause Parkinson's, most of them highly penetrant and often with early or atypical onset. 32 Seven have been formally adjudicated as causal contributors. 67 In genetic forms, Lewy bodies may be absent even when the illness looks typical, so they are not required for the diagnosis. 318 See Alpha-synuclein and Lewy bodies.
LRRK2 (PARK8)
Found in 2004 by two groups; one named the protein "dardarin", from the Basque word for tremor. 271,388 Six mutations meet the consortium's criteria for proven pathogenicity; G2019S is by far the commonest (391 unrelated carriers). 140 G2019S and R1441C raise the enzyme's kinase activity, which points to a gain of function. 375
The illness was "more benign" than idiopathic Parkinson's in one comparison: dyskinesia after 5 years in 11% versus 25%, cognitive impairment in 23% versus 70%. 140 Lewy bodies were found in 17 of 37 autopsied cases. 168
GBA1
GBA1 makes glucocerebrosidase, a lysosomal enzyme. Two disease-causing variants cause Gaucher disease; one variant raises the risk of Parkinson's with reduced penetrance. 312,333 Newer numbering adds 39 amino acids: N409S was N370S, L483P was L444P, E365K was E326K. 312
"Severe" variants carried more risk than "mild" ones in Ashkenazi Parkinson's (odds ratio 10.3 versus 2.2), and earlier onset (mean 53.1 versus 58.1 years). 119 Carriers had a higher risk of dementia (hazard ratio 3.16) and death (1.85). 62
SNCA
The first Parkinson's gene: A53T, found in 1997 in an Italian kindred and three Greek families. 287 A30P followed in 1998, and E46K in a Spanish family with dementia and hallucinations. 181,385 Extra whole copies of the gene also cause it, and more copies mean earlier onset (see below). 36,53,339
PRKN (parkin)
Identified in 1998 as the cause of autosomal recessive juvenile parkinsonism. 173 In 2000 parkin mutations were found in 49% of early-onset families with unaffected parents, in 77% of isolated cases starting at 20 or younger but only 3% starting after 30. 201 Carriers started younger (32 versus 42 years), more often had dystonia at onset, responded well to levodopa and developed dyskinesia. 201
PINK1 (PARK6)
Homozygous mutations in this mitochondrial kinase were found in three consanguineous families in 2004: the first direct genetic link between mitochondria and Parkinson's. 356 Across more than 1,100 people with two PRKN, PINK1 or DJ-1 variants, median onset was about 30, the illness was clinically typical with an excellent treatment response, and cognitive decline was relatively uncommon. 169
DJ-1 (PARK7)
Linked in 2003 to autosomal recessive early-onset parkinsonism. DJ-1 is implicated in the response to oxidative stress, and its loss leads to neurodegeneration. 35 It is very rare: one positive result among 2,699 Israeli patients tested. 17
VPS35
D620N was found by exome sequencing in a Swiss family with tremor-predominant, levodopa-responsive Parkinson's (mean onset 50.6), then in three more families, and not in 3,309 controls. 363 An Austrian family with 16 affected members (mean onset 53) carried the same variant. VPS35 is part of the retromer complex. 389
SNCA: the dose of one gene
Most people have two copies of SNCA, one on each chromosome 4. A duplication on one chromosome makes three copies; a triplication, or a duplication on both, makes four. Pick a dose.
Mean ± 1 standard deviation from a meta-analysis of 59 families. The chromosome is schematic.
Dominant and recessive, in a Punnett square
LRRK2 shows the dominant pattern: one changed copy is enough. Parkin (PRKN) shows the recessive pattern: both copies must be changed. Choose a family, play it, then turn on the penetrance layer to see why "inheriting the variant" and "getting Parkinson's" are not the same thing.
Each box is an equally likely outcome for every pregnancy; the fractions are counted from the square itself. The penetrance shading uses published estimates for LRRK2 G2019S.
Penetrance: how many carriers develop Parkinson's, by age
Penetrance is the share of carriers who develop the disease by a given age. Drag the age, switch genes, and show the disagreement: the estimates depend heavily on how the families were found.
Only published points are plotted; lines join a study's own points. The panel shows each study's latest estimate at or below the chosen age.
Why the estimates disagree. Published G2019S penetrance ranges from 24% to 100%, because studies differ in ascertainment (families with many affected members versus unselected relatives), age, sex, ancestry and modifiers. 140,187,209 GBA1 estimates differ about threefold for the same reason. 10,16,23,303
What they agree on. Risk rises with age, and penetrance is incomplete for both genes: LRRK2 about 25–80% by 79–80, GBA1 about 8–30% by 80. 10,16,140,187,209 In the Ashkenazi kin-cohort, women carriers had a higher estimate than men (29% versus 22%). 209
Who carries what: founder populations
Some variants are far more common in particular ancestries, because they trace back to a shared founder. Pick a population to see the share of people with Parkinson's who carry LRRK2 or GBA1 variants.
Shares are of people with Parkinson's, not of the whole population. Studies used different tests and counted differently.
Odds ratio is not lifetime risk
In a 16-centre study, people carrying any GBA1 variant had about five times the odds of Parkinson's (odds ratio 5.43). 333 That sounds close to destiny, but it is a ratio. In Ashkenazi Gaucher families, Parkinson's by 80 was 2.1% in non-carriers and 7.7% in heterozygous parents. 10 Switch the carrier estimate to see the spread.
The non-carrier grid comes from the Gaucher-family study; the other two carrier estimates come from different studies, so those comparisons cross studies.
Common variants and polygenic scores
GWAS: 90 signals
The 2019 meta-analysis tested 7.8 million variants in 37,688 cases, 18,618 "proxy cases" (people with an affected first-degree relative) and 1.4 million controls, all of European ancestry. It found 90 independent risk signals in 78 genomic regions, 38 of them new. 237
Heritability
Common variation accounts for about 22% of Parkinson's liability (0.18–0.26), slightly below an earlier estimate of 0.27. The 90 variants explain 16–36% of the heritable risk, depending on the assumed prevalence. 32,237
Polygenic scores
People in the top quarter of a polygenic score had higher odds than those in the bottom quarter: 3.74 in one cohort and 6.25 in a validation cohort. Prediction reached an area under the curve near 0.69; reviews summarise 0.65–0.69, a modest ability to tell people apart. 32,237
Family history
In a meta-analysis, a first-degree relative with Parkinson's gave an odds ratio of 3.23 (2.65–3.93) and any affected relative 4.45 (3.39–5.83), the strongest risk factors examined. 247 See Risk and protection.
Common variants in known genes
A common SNCA promoter allele (REP1, 263 bp) raised the odds 1.43-fold without changing age at onset. 208 GBA1 E326K (E365K) raised them 1.99-fold in 15 studies. 156 The association of T408M is inconsistent across studies; no pooled figure was checked for this guide. 312
Mostly European data
Most risk variants were found in people of European ancestry, and relatively little is known about Parkinson's genetics in other populations. 32,237 LRRK2 G2019S, for example, was found in fewer than 0.1% of 4,172 Asian patients. 140
What a genetic test means
What testing finds
ROPAD found a Parkinson's-relevant result in 14.8% of 12,580 patients from 16 countries; PD GENEration in 13% of 8,301 North American participants. Most positives were GBA1 risk variants. 67,376 Counting matters: PD GENEration reported GBA1 in 7.7% as "reportable" but 10.1% as "disease-relevant". 67
Who tests positive
In ROPAD the yield was 19.9% with onset at 50 or younger, 19.5% with a family history and 26.9% with both; it fell 3% for each later year of onset, and women were 22% more likely to test positive. 376 In PD GENEration it was 23.9% with Ashkenazi Jewish, Basque or North African Berber ancestry and 9.1% with no risk factor. 67
Criteria miss most carriers
A pooled analysis of 25,063 patients found that an onset threshold of 50 or younger captured only 23–32% of carriers in the population cohorts, and adding family history barely helped. Its authors, and the ROPAD authors, argue for broader testing in the gene-targeted trial era. 22,376
Testing has not changed standard care
A 2023 MDS-published review notes "no proven gene-targeted therapies, but clinical trials are underway". 273 In 2026 the largest LRRK2 kinase-inhibitor trial and a GBA1 enzyme-activator trial reported no slowing. 229,239 See Toward a cure.
How results were given
In PD GENEration every participant received results in a genetic-counselling session with a neurologist or certified genetic counsellor, and people with two GBA1 variants were referred to Gaucher centres. 67 Early consortium and Israeli authors urged caution in counselling and questioned screening people without symptoms until penetrance is better defined and prevention exists. 140,259
No final guideline yet
No final MDS consensus on clinical genetic testing was found. The 2023 review lists open questions: testing without medical actionability, whether testing should vary by ethnicity, long-term outcomes of testing people without symptoms, and cost. 273 Many people with Parkinson's do not know their genetic status, which hampers trial enrolment. 66,67