Variants in the GBA1 gene are strong genetic risk factors for developing sporadic Parkinson’s disease (PD). Some of these variants (p.N370S, p.L444P, and p.E326K) have high allelic frequencies in patients with PD and increased risk for developing the disease1,2. The prevalence of GBA1 variants in Ashkenazi Jewish populations with PD is the highest (18%) compared to other ethnic backgrounds1. In non-Ashkenazi patients with PD, the prevalence of GBA1 variants is close to 10%, but it may vary according to the country and screening methodology.

For GBA1 analysis, most studies use genotyping methods targeting specific common pathogenic variants or sequencing3. GBA1 is a large gene located in a complex genomic region, with a pseudogene with 96% of homology in the coding sequence, which increases the risk of recombination and poses challenges to PCR and sequencing. Regarding GBA1 sequencing, while some studies opted for sequencing and analyzing all exons (full sequencing), other groups are sequencing only specific exons, like exons 9 and 10, where common variants are found (targeted sequencing)3. Recently, long-read sequencing methods have been employed to study better the region4. The full sequencing of GBA1 warrants the identification of rare or population-specific variants, shedding light on the pathophysiology of effects of GBA1 in PD and helping to recruit more eligible patients for GBA1-targeted trials5.

In Latin America, a region with a population of approximately 660 million, few studies on PD associated with GBA1 variants (PD-GBA) have been published. Latinos, referring to those individuals from Latin America, are genetically very heterogeneous due to a complex three-way admixture (Native American, European and African), with large differences in the amounts of each ancestry not only between but also within countries. Furthermore, unfavorable socioeconomic conditions among its countries may hamper the capacity to perform the full sequencing of GBA1, which may negatively influence the results of these studies. To explore the prevalence and odds ratio of PD-GBA in Latin America, we conducted a systematic literature review and meta-analysis.

A total of 11 clinic-based studies (one study was conducted in two different countries, comprising 12 cohorts) were included in the analysis (Supplementary Fig. 1). A total of 1,719 patients with PD and 1,444 controls were analyzed, and an average prevalence of 5.4% of GBA1 carriers was found (Table 1). No participant reported Ashkenazi Jewish ancestry. There are some methodological differences among studies: four studies recruited only early-onset patients with PD (age at onset varying from less than 45 to less than 55 years), and two studies did not include healthy controls. Only four cohorts performed full sequencing of GBA1 (n patients = 735; n controls = 445). Despite Brazil being the most populous country in Latin America and with more patients with PD screened for GBA1 variants, surprisingly no study from this country used full sequencing of the gene.

Table 1 Main characteristics of studies and frequencies of GBA1 pathogenic variants (plus p.E326K) in patients with PD and controls included in the meta-analysis.
Full size table

We performed a meta-analysis to estimate the odds ratio (OR) of developing PD for carriers of all pathogenic GBA1 variants plus p.E326K, and specifically for carriers of p.L444P and p.N370 variants. We used a fixed-effect model with a continuity correction of 0.5 for studies with zero GBA1 carriers in the PD or control group. For calculating the OR, we used the Cochran–Mantel–Haenszel test and the Tarone test to examine heterogeneity6. Statistical tests were performed using the R software version 4.0.4, with the package metafor.

In the meta-analysis to estimate the OR of all pathogenic GBA1 variants (p.L444P, p.N370S, p.K198E, IVS2 + 1 G > A, Rec1) plus p.E326K, we used only studies which performed full sequencing, avoiding the underestimation of other methods of genetic screening. Carrying any pathogenic GBA1 variant plus p.E326K was associated with an increased risk of PD (OR = 3.51, 95% CI = 1.6–7.4), but heterogeneity was significant (Tarone p = 0.02). After excluding the cohort with extreme OR (with two carriers of variants in controls, but none in patients with PD), the heterogeneity was removed (OR = 4.63, 95% CI = 1.9–10.7; Tarone p = 0.96; Fig. 1a). For p.L444P (combining full and targeted sequencing studies), the average allelic frequency was 0.028 in patients and 7.7 × 10−4 in controls; the variant increased substantially the risk of PD (OR = 20.2, 95% CI = 3.4–118.9, Fig. 1b), and heterogeneity was not significant (Tarone p = 0.39). For p.N370S (combining full and targeted sequencing studies), the average allelic frequency was 0.012 in patients and 0.001 in controls; the variant increased the risk of PD (OR = 4.9, 95% CI = 1.2–19.8, Fig. 1c); however, the heterogeneity was significant (Tarone p = 0.02). We must highlight that the p.L444P variant causes a more severe form of Gaucher’s disease than p.N370S6. Only one study (including individuals from Peru and Colombia) explored the effect of pathogenic variants of GBA1 on clinical phenotype (age at onset), and carriers had motor symptoms approximately eight years earlier than non-carriers7.

Fig. 1: Meta-analyses of GBA1 variants in patients with Parkinson’s disease from Latin America.
figure 1

a Meta-analysis of combined GBA1 pathogenic variants plus p.E326K from studies which performed full sequencing of the gene. b Meta-analysis of the GBA1 L444P variant. c Meta-analysis of the GBA1 N370S variant.

Full size image

Until now, approximately 10,000 patients with PD have been full-sequenced for GBA1 in Europe5. The low proportion of variants in controls could be reduced with the full sequencing and larger numbers of participants. The OR values calculated by our meta-analysis for combined pathogenic GBA1 variants plus p.E326K and p.N370S were similar to previous studies in non-Ashkenazi populations (Supplementary Table 1). However, the overall OR for p.L444P in Latin America was one of the highest values in non-Ashkenazi populations (Supplementary Table 1), mainly due to studies from Brazil8,9,10,11,12,13 and Mexico14. We detected heterogeneity between studies, probably due to low numbers and distinct inclusion criteria (studies recruiting any patients with sporadic PD, focusing on patients with early-onset PD or positive family history of PD). For example, we excluded the study from Costa Rica from the meta-analysis to reduce heterogeneity. The study found no GBA1 variants in patients with PD, probably due to low sample size, the method of sequencing (molecular inversion probes), and a geographical enrollment bias (participants from a specific metropolitan area, in the detriment of metropolitan areas and coastal zones of the country)15. Full sequencing of GBA1 in Latinos has already discovered the pathogenic p.K198E variant in Colombia, with a prevalence of 6% in patients with PD, increasing the disease risk six-fold7, showing that some variants of clinical relevance may only be reported after sequencing the whole gene.

Thus, considering the results of our review, it remains unclear how much the full sequencing of GBA1 would increase the number of PD-associated variants found compared to targeted approaches in Latin America. The number of studies that performed full sequencing of GBA1 is very low (four cohorts) and included a limited number of patients compared to other populations. More large-scale studies in PD-GBA using the full sequencing in Latin America are needed and may elucidate this issue. Promising ongoing clinical trials on PD-GBA may bring novel therapies for these patients; providing more GBA1 full-sequencing opportunities for Latin American populations would diminish health disparities for underrepresented communities.

Methods

We performed a search of PubMed/MEDLINE and EMBASE from inception until October 2021. We created search strings for each database using “Parkinson’s disease,” “GBA,” and the countries in Latin America (Supplementary Table 2). Two rounds of study selection were performed. In the first round, we included original studies describing patients with PD carrying GBA1 variants from all countries of Latin America. Reviews, meta-analyses, and studies with animal models were excluded. In the second round, full texts were evaluated, and we selected articles that reported GBA1 genotyping (pathogenic variants plus p.E326K variant) or GBA1 sequencing on cohorts of patients with PD. Two reviewers performed selection rounds independently, and disagreements were resolved by consensus. After, data were collected through an online spreadsheet.