Pharmacogenomic architecture of antihypertensive switching implicates neurotensin-NTSR1 signaling in ACE inhibitor-induced cough

Technical Summary: Pharmacogenomic Architecture of Antihypertensive Switching

Problem Statement
Early failure of antihypertensive medication treatment affects approximately one in four patients, yet the biological mechanisms driving this variation remain poorly understood. While genetic variability shows promise for personalizing hypertension care, previous pharmacogenetic studies have been underpowered for genome-wide discovery. Furthermore, conventional laboratory-derived response phenotypes (e.g., plasma concentration changes) often fail to translate to real-world treatment trajectories such as medication switching or discontinuation. There is a critical need to identify genetic determinants of short-term treatment failure using large-scale, longitudinal data.

Methodology
The authors conducted a two-stage study integrating genome-wide profiles with longitudinal medication data from over 400,000 individuals across three biobanks: FinnGen, the UK Biobank (UKB), and the Estonian Biobank (EstBB).

• Phenotype Definition: Medication use trajectories within the first year were classified into three categories: Continue (maintained index medication), Switch (changed to a different antihypertensive class), and Discontinue (stopped treatment). This classification was applied to five major antihypertensive classes: thiazide diuretics (THZ), selective β-blockers (sBB), dihydropyridine calcium channel blockers (dCCB), angiotensin-converting enzyme inhibitors (ACEI), and angiotensin 2 receptor blockers (ARB).
• Genetic Analysis: Genome-wide association studies (GWAS) were performed in FinnGen (discovery) comparing Switch cases against Continue controls, followed by replication in UKB and EstBB. The study utilized imputed genotype data and whole-exome sequencing where available.
• Downstream Analyses:
  • Fine-mapping: SuSiE was used to identify credible sets of causal variants.
  • Coding Variant Meta-analysis: A meta-analysis across all three cohorts prioritized protein-coding variants.
  • Medical Code Enrichment: Logistic regression was used to link switching phenotypes to specific adverse events (ICD-10/ICPC-2 codes).
  • Drug-Gene Interaction: A self-controlled risk interval design with conditional logistic regression was employed to estimate interactions between genetic variants and medication exposure regarding adverse events.
  • Heritability and Correlation: LD score regression (LDSC) was used to estimate SNP-based heritability and genetic correlations between Switch and Discontinue phenotypes.

Key Contributions and Results

1. Identification of 14 Genome-Wide Significant Loci:
   The study identified 14 significant loci associated with switching, specifically for ACEI and dCCB classes. No genome-wide significant loci were found for sBB, THZ, or ARB switching.

   • ACEI Switching: 13 loci were identified, including four novel loci (OSBPL10, SYBU, CACNA1H, SHISAL1).
   • dCCB Switching: A single locus at CYP3A43 was identified.

2. Neurotensin-NTSR1 Signaling in ACEI-Induced Cough:
   Multiple independent signals converged on the neurotensin-NTSR1 pathway for ACEI switching:

   • NTSR1 Missense Variant: A 320-fold Finnish-enriched protective missense variant, p.Gly301Arg (rs148569146) in the neurotensin receptor gene NTSR1, was identified (OR=0.49, P=3.3×10⁻⁴³). This variant localizes to the intracellular allosteric site of the receptor.
   • Neurotensin Gene (NTS): The strongest GWAS signal (rs181941187 near RASSF9) tagged a cluster of variants near the NTS gene, the endogenous ligand for NTSR1.
   • Mechanistic Link: In drug-gene interaction analyses, the NTSR1 G301R variant was associated with a significantly reduced risk of ACEI-induced cough (OR=0.39, P=8.1×10⁻⁴). The authors propose that this variant acts as a constitutive biased modulator of NTSR1, potentially altering signaling downstream of neurotensin accumulation (which occurs during ACE inhibition), thereby mitigating cough.

3. CYP3A4*22 as a Predictor of dCCB Intolerance:
   The dCCB switching locus at CYP3A43 was found to be in near-complete linkage disequilibrium (r²=0.99) with the functional CYP3A422* allele (rs35599367). This allele is known to reduce CYP3A4 enzyme activity. The study confirms that carriers of this allele have an increased risk of dCCB switching (OR=1.23), likely due to elevated serum drug concentrations and subsequent side effects like peripheral edema.

4. Distinction Between Switching and Discontinuing:
   The study observed contrasting genetic architectures between Switch and Discontinue phenotypes. Discontinue phenotypes showed high genetic correlation across drug classes, suggesting a shared basis potentially driven by behavioral factors or poor persistence unrelated to drug response. In contrast, Switch phenotypes were more specific to drug class, and genetic effect sizes for Switch were consistently attenuated in Discontinue analyses, supporting the use of "switching" as a more specific phenotype for pharmacogenetic discovery.

Significance and Claims
The paper claims to extend the understanding of antihypertensive treatment failure by:

• Reinterpreting ACEI Cough Biology: The findings implicate the neurotensin-NTSR1 signaling pathway as a key mechanism in ACEI-induced cough, offering a genetic and mechanistic expansion beyond the traditional bradykinin-substance P hypothesis.
• Validating Pharmacogenetic Predictors: The study provides genome-wide, cross-biobank evidence supporting CYP3A422* as a mechanistically supported predictor of dCCB intolerance.
• Methodological Advancement: The authors demonstrate that longitudinal medication trajectories (specifically switching) offer a rich and specific framework for pharmacogenetic discovery, potentially improving statistical power over traditional discontinuation phenotypes or laboratory-based endpoints.

The authors conclude that these results highlight the potential for genotype-informed antihypertensive medication selection, while acknowledging limitations regarding ancestry generalizability (focused on Finnish and European populations) and the inability to directly observe the specific clinical reasons for switching.