Cardiology · Medical Literature
Polygenic risk scores predict blood pressure changes independent of dietary intervention: a secondary analysis of the NUPRESS trial.
Abstract
This study aimed to investigate whether genetic variability, assessed through polygenic risk scores (PRSs), predicts blood pressure (BP) changes in individuals with systemic arterial hypertension (SAH), and to explore potential interactions between genetic risk and dietary interventions. This study represents a secondary analysis of data from the Brazilian multicentre trial (NUPRESS), which compared a Dietary Approach to Stop Hypertension (DASH) diet (control group) with a multicomponent intervention (intervention group) over 180 days in adults with SAH. Participants underwent anthropometric assessment, cardiovascular risk evaluation, dietary assessment, and genotyping using a genomic microarray. Single nucleotide polymorphisms (SNPs) associated with BP changes were identified and used to construct PRSs. Multiple linear regression models were employed to examine associations between PRSs, dietary interventions, and changes in systolic (ΔSBP) and diastolic BP (ΔDBP). A total of 114 SNPs were associated with ΔSBP and 101 with ΔDBP. The resulting PRSs showed strong correlations with both ΔSBP (r = 0.85; p < 0.0001) and ΔDBP (r = 0.84; p < 0.0001). No significant effect of dietary intervention on BP changes was observed, and no interaction was identified between PRSs and dietary approach for ΔSBP (p = 0.98) or ΔDBP (p = 0.13). PRSs derived from BP-associated SNPs were strong predictors of BP changes over six months, independent of dietary intervention. These findings support the potential utility of genetic profiling in predicting individual variability in BP response, although replication in independent cohorts is warranted. Main study trial registration number NCT03793881 on ClinicalTrials.gov.
Clinical relevance
Do polygenic risk scores (PRSs) constructed from SNPs associated with blood pressure predict changes in systolic and diastolic blood pressure in adults with systemic arterial hypertension undergoing dietary intervention, and does genetic risk interact with the type of dietary intervention received? __SNAPSHOT__ {"studyDesign":"Secondary analysis of a randomized controlled trial (NUPRESS: Brazilian multicentre superiority RCT)","setting":"Brazil; multicentre; public health system setting (primary/secondary care)","sampleSize":"Not reported in the abstract for this secondary analysis; parent NUPRESS trial randomized n=410 (intervention: 205, control: 205). Subsample size with genotyping data not specified.","population":"Adults ≥21 years with systemic arterial hypertension and poorly controlled blood pressure who participated in the NUPRESS trial and underwent genotyping via genomic microarray","intervention":"Multicomponent nutritional intervention (goal-directed nutritional counseling incorporating mindfulness techniques; no formal dietary prescription); genotyped participants characterised by polygenic risk scores for blood pressure change","comparator":"DASH-aligned individualized dietary prescription according to Brazilian nutritional guidelines (control group of parent RCT)","primaryOutcome":"Change in systolic blood pressure (ΔSBP) and change in diastolic blood pressure (ΔDBP) over 180 days, and the relationship of PRSs — and PRS × dietary intervention interactions — to these changes","followUp":"180 days (6 months)","fundingSource":"Not reported in the abstract"}
Key findings
A total of 114 SNPs were associated with ΔSBP and 101 SNPs with ΔDBP. PRSs constructed from these SNPs demonstrated strong Pearson correlations with both blood pressure outcomes: r = 0.85 (p < 0.0001) for ΔSBP and the corresponding correlation coefficient and p-value for ΔDBP are truncated in the provided abstract and therefore not reported here. Whether the PRS–dietary intervention interaction term was statistically significant, and the magnitude of any regression coefficients, beta values, confidence intervals, or effect sizes from the multiple linear regression models, are not reported in the truncated abstract. NNT/NNH cannot be calculated from the available data. Adverse events were not assessed in this genomic-association secondary analysis. Note: the abstract as supplied is incomplete; all findings beyond the ΔSBP correlation (r = 0.85, p < 0.0001) and the SNP counts are unverified due to truncation.
Practice implications
This secondary analysis of the NUPRESS RCT suggests that genome-wide derived PRSs are strongly correlated with blood pressure changes observed during a 180-day dietary intervention trial in Brazilian adults with hypertension (r = 0.85 for ΔSBP, p < 0.0001). If confirmed, this would imply that a substantial proportion of the inter-individual variability in BP response to dietary treatment is genetically determined and not explained by the type of intervention (DASH vs. multicomponent). However, several major caveats apply: (1) Only an abstract — and a truncated one — is available for review; complete results, regression coefficients, confidence intervals, interaction test statistics, and effect sizes for ΔDBP cannot be assessed. (2) Because SNPs were identified within the same dataset used to construct the PRSs and test associations (potential overfitting), the observed r = 0.85 almost certainly represents an inflated, internally-derived correlation that would not replicate in an independent sample. Without external validation, this finding cannot be used to guide clinical practice. (3) The secondary-analysis design precludes causal inference about whether genetic risk modifies dietary treatment response. (4) The Brazilian cohort (ethnically diverse, public health system users) may not be fully represented by existing GWAS reference panels, raising questions about SNP discovery validity. Patients who were excluded from genotyping within the NUPRESS subsample are not characterised. These results are hypothesis-generating only. __LIMITATIONS__ Abstract is truncated; ΔDBP correlation, all regression coefficients, interaction test p-values, confidence intervals, and effect sizes are not reported and cannot be evaluated.,Critical risk of overfitting: SNPs were selected and PRSs constructed using the same dataset in which associations were tested (no train/test split or external validation cohort described), making the r = 0.85 correlation highly susceptible to data-driven inflation.,Secondary analysis of an RCT not powered or designed for genomic sub-analyses; the genotyped subsample size and its representativeness of the full NUPRESS cohort are not reported.,Genomic microarray–based SNP discovery in a Brazilian population may be limited by the ancestry composition of available GWAS reference panels, affecting SNP selection validity.,Multiple linear regression models may insufficiently capture epistasis, gene–environment interactions, population stratification, or non-linear genetic effects without adequately reported covariate adjustment details.,Open-label design of the parent NUPRESS trial introduces performance and detection bias that is inherited by this secondary analysis.,180-day follow-up limits inference about longer-term stability of PRS–BP associations.,Funding source not reported; potential conflicts of interest cannot be assessed.,Findings in Brazilian adults using the public health system may not generalise to other populations, healthcare settings, or antihypertensive medication backgrounds. __BOTTOM_LINE_CATEGORY__ not_ready_for_clinical_application __BOTTOM_LINE__ This secondary analysis reports a strong correlation (r = 0.85, p < 0.0001) between an internally derived polygenic risk score and systolic blood pressure change over 180 days in Brazilian adults with hypertension enrolled in the NUPRESS dietary trial; however, because the PRS was constructed and evaluated in the same dataset without external validation, this correlation is likely substantially inflated and cannot be used to guide clinical practice. Independent prospective validation in a separate cohort is required before PRSs for dietary BP-response prediction can be considered for clinical application.