**4. Conclusions and Perspectives**

Thanks to the introduction of exome arrays technologies, great efforts have been conducted to extend association analyses to rare and coding variants. Recently, the joint work of large consortia allowed the interrogation of hundreds of thousands of SNVs in up to 475,000 individuals [28–38]; (Table 1). Some new low-frequency and rare variants have been reported that are consistently associated with BP traits, with size effects higher than 1.5 mmHg, and that should undergo deep functional testing. Considering the single variant analyses described here, the largest effect, to date, was observed for a rare missense SNV in the *KLH3* gene in relation to SBP (8.2 mmHg with se = 4.1) [37]. Despite the large sample size (up to 422,604 subjects for the exome analysis), the study from Warren and coauthors was still underpowered to identify rare variants with statistical significance. When considering the joint impact of 107 mostly common variants, a 9.3 mmHg higher SBP was reported for subjects >50 years and carrying the highest genetic risk score [37]. This finding has potential implications concerning early lifestyle interventions in high-risk individuals. In summary, although several complex networks of interacting pathways controlling BP have been established (e.g., RAAS and ENaC-related pathways), the current efforts on rare variants analysis have not yet provided a clear answer on where the missing heritability lies.

The advent of NGS provided the opportunity to detect, in a high-throughput way, the entire spectrum of genomic variation ranging from rare to common variants and from SNVs to insertions, deletions, and copy number variants. Despite the undeniable advantages, few studies have been conducted so far using NGS technologies in relation to hypertension and/or BP [61–67]; (Table 3). WES and, more so, WGS costs are still too high to analyse the large sample size required to identify rare variants. Target re-sequencing allows the cutting of laboratory costs and increases the statistical power by reducing multiple signals testing, therefore, this approach could be useful to detect causative variants underlying the trait by deeply analysing BP-associated loci described by GWAS. However, the

studies reported here failed to identify new rare variants, likely because of the reduced sample size compared to GWAS [63,64]. The joint effort of large consortia with available sequencing data would be helpful to meet the need of a larger sample size.

Novel statistical approaches have been developed to overcome the limit imposed by the extremely low frequencies. Also, these tests attempt to take into account the high heterogeneity of the genetic regions in which both common and rare as well as causative and non-causative variants are more likely to occur [99]. However, detecting the few true causative variants among the large number of non-coding variants arising from NGS still represents a big challenge, and additional improvements to better annotate and filter the variants are required.

Another main limitation of rare variants analysis is the study of gene-gene and gene-environment interactions at a population level, which can be investigated only in terms of burden and collapsing tests, with environmental factors playing, anyway, an important role in systemic hypertension. Functional in vitro and in vivo models should further support the statistical interactions.

Rodent models represent an attractive genetic resource to functionally evaluate previously identified rare variants overlapped with human loci. Several rat and mouse strains have been developed for complex phenotypes, including hypertension, and exploited to perform QTL analysis and genome sequencing [100–104]. Here, we reported the study of Vecchione et al. [27], in which, thanks to experimental models, the authors clarified how a rare variant within the *BPIFB4* gene, a possible genetic risk factor for high BP, was implicated in the BP homeostasis by altering eNOS signaling.

It should also be considered that, as hypertension is an age-related condition, additional longitudinal studies incorporating repeated measures of BP would be advantageous. Lastly, most findings should be treated as trait-specific (SBP, DBP, PP, MAP, or hypertension) and population-specific. The majority of studies reported findings deriving from European populations. Allele frequencies and hypertension risk may differ among different geographic regions because of a selective pressure that occurred during the Out-of-Africa Expansion [105].

In conclusion, as sequencing costs will sufficiently decrease to ensure the proper sample size, and novel bioinformatic and biostatistical tools will be available for appropriate analyses, the identification of functional rare and low-frequency variants could really contribute to solving the high complexity of the genetics of hypertension and to elucidate whether new genes might improve patients care and the stratification of patients to distinguish those who will respond best to antihypertensive treatments.

**Acknowledgments:** This work was supported by Compagnia di San Paolo-IIGM (to Giuseppe Matullo) and by Ministero dell'Istruzione, dell'Università e della Ricerca—MIUR project "Dipartimenti di Eccellenza 2018–2022" to Department of Medical Sciences (Giuseppe Matullo), University of Torino.

**Author Contributions:** Alessia Russo: defined the structure of the review, performed literature paper selection, contributed to writing and critical reading; Cornelia Di Gaetano: contributed to literature paper selection, writing, and critical reading; Giovanni Cugliari: contributed to literature paper selection, writing, and critical reading; Giuseppe Matullo: defined the structure of the review, supervised the paper selection, contributed to writing and critical reading.

**Conflicts of Interest:** The authors declare no conflicts of interest.

## **Abbreviations**


