Genes, Aquaporins, and Peritoneal Dialysis

More information
AQP1 VA for NephJC.png
The AQP1 Visual Abstract
Genes, Aquaporins, and Peritoneal Dialysis
Genes, Aquaporins, and Peritoneal Dialysis

This week, we discuss the discovery that peritoneal membrane water transport, and PD outcomes, are influenced by genetic variations in the Aquaporin 1 channel.

Introduction

The ancient Egyptians were the first to describe the peritoneal cavity as early as 3000 BCE in the Ebers Papyrus, a lengthy text describing medical knowledge at the time (Loukas et al, Clinical Anatomy 2011) . In 1877 Georg Wegner discovered that a concentrated sugar solution would increase in volume when left in the peritoneal cavity. In 1923, Georg Ganter performed the first peritoneal dialysis on a guinea pig (Boen S. T., History of Peritoneal Dialysis 1985). He observed that the urea nitrogen concentration of the peritoneal dialysate fluid approached that of the serum and after several exchanges, and the animal improved.

The art and science of peritoneal dialysis have come a long way since then, with advancements in peritoneal cavity access and improvements to the types of dialysate available, which now include a variety of  both crystalloid and colloid-based solutions. While hemodialysis is the most commonly used dialysis, PD is the leading form of home-based dialysis due to many advantages over center-based hemodialysis, such as ease of use and accessibility in remote areas (Mehrotra et al, JASN 2016, Kam-Tao Li et al Nature Reviews Nephrology 2017).

However, due to the nature of PD, there is large inter-patient variability in how efficient the procedure is at ultrafiltration (​​Brimble et al, JASN 2006). The specific factors that influence these differences however, are poorly defined. Recently, a genome-wide association study found a polygenic association and 17% heritability in peritoneal small-solute transport rate, supporting the idea that there is a genetic influence on this key variable (Mehrotra et al, Kidney International 2021). 

Aquaporin-1 has been implicated as a potential contributor to this difference. First identified in red blood cells, Aquaporin-1 is expressed on endothelial cells that form the peritoneum’s diffusion membrane . In studies with Aqp1-knockout mice (Yang et al, American Journal of Physiology 1999, Ni et al, Kidney International 2006 ), aquaporin-1 was identified as the primary mediator of water transport and accounts for up to half of the ultrafiltration that occurs during peritoneal dialysis. These studies suggest that variations in the expression of AQP1 may influence volume management in PD patients.

The Study

Methods

This was a retrospective cohort study. Patients were recruited from seven multinational cohorts of PD patients including patients from Belgium, the Netherlands, Spain, the UK and China. Patients had to have available DNA samples and information about peritoneal water transport, outcomes or both. Patients were followed until death, transfer to HD, kidney transplant, withdrawal from dialysis, or until the end of the follow up period. Most information was obtained via a chart review.

Genetics
Variants of the Aquaporin-1 gene were genotyped, specifically those with a minor allele frequency greater than 10%. Four variants that covered the major linkage disequilibrium blocks of AQP1 were identified (see figure below). Genomic DNA was extracted from peripheral-blood leukocytes (Puregene, Gentra Systems) and analyzed with a competitive allele-specific polymerase-chain-reaction assay (LGC Group).

Figure S1 from Morelle et al, NEJM 2021

Peritoneal Transport Assessment
The authors tested each variant for an association with two surrogate measures of free-water transport: net ultrafiltration and sodium sieving.

Net ultrafiltration was defined as the difference between volume drained at 4 hours and volume infused at the start of the standard peritoneal equilibration test (PET). Data from the 4.25% dextrose (3.86% glucose) solution tests were used in the analysis.

Sodium sieving was defined as the ratio of the difference between sodium level in dialysate at the beginning of the dwell and the level at 1 hour to the sodium level in plasma with a correction for sodium diffusion. These data were collected in the context of a patient’s baseline peritoneal solute transfer rate, which was defined as the ratio of the creatinine level in dialysate to the creatinine level in plasma at 4 hours.

In one cohort, the association between one of the variants, rs2075574 and UF induced by 4.25% dextrose was compared to UF induced by 7.5% icodextrin (colloid).

Analysis
The groups (based on genotypes) were compared for net ultrafiltration and the clinical outcomes (death, technique failure and combination). Multivariate analyses were conducted, adjusting for these prespecified covariates:
Net Ultrafiltration Outcome

  • peritoneal solute transfer rate

  • diabetes status ;

Daily Net Ultrafiltration Outcome

  • peritoneal solute transfer rate,

  • residual urine volume,

  • and duration of dialysis treatment

Survival Analysis

  • age,

  • sex,

  • cardiovascular disease status,

  • diabetes status, and

  • peritoneal solute transfer rate (in the survival analysis).

Results

Investigation of the association between the AQP1 variants and ultrafiltration occurred in two phases, a discovery cohort of 433 patients and a validation phase that included 985 participants.

Figure S2, Supplemental Appendix, Morelle et al NEJM 2021

The former was primarily to identify which of the variants had an association with differences in ultrafiltration, if at all. This was determined to be the rs2075574 variant, located within the AQP1 promoter. The purpose of the validation phase was to confirm this association. Diverse patient populations were represented in this international study. The discovery phase included patients of  Belgian, Dutch and Spanish descent, while the validation phase utilized British and Chinese cohorts.

Fig. 1B, Morelle et al NEJM 2021

The baseline characteristics of the patients are shown below. The average age of participants was 54 years and about 75% of participants were from European descent (notably, very few people of African descent participated). Medical comorbidities were common, with 21% of patients having cardiovascular disease and 24% with diabetes. The authors delineated patients by their AQP1 genotype at rs2075574, with:

  • 41% of patients having the CC genotype (wild-type)

  • 45% having the CT genotype (heterozygous for promoter variant)

  • 14% having the TT genotype. (homozygous for promoter variant)

 Table 1, Morelle et al NEJM 2021

In the discovery phase, those with the TT genotype (homozygous for rs2075574) had significantly lower net ultrafiltration than those with the CC wild-type genotype (506ml +/- 237ml vs 626ml +/- 283ml, P=0.007).  There was a similar peritoneal solute transfer rate and similar clinical characteristics. This association was independent of other covariates.

Fig. 1C, Morelle et al NEJM 2021

Next, the validation phase confirmed the independent association between the AQP1 genotypes and peritoneal ultrafiltration. Daily net ultrafiltration level was lower in those with the TT genotype compared to the CC genotype (368ml +/- 603ml vs 563ml +/- 641ml, P= 0.003).

The authors also investigated further the specific role of this risk variant. The T allele of rs2075574 was associated with significantly lower AQP1 promoter activity compared to the C allele, and this was reflected by the 27% lower AQP-1 mRNA level and 37% lower AQP-1 protein level in the peritoneum of patients with the TT genotype. Given this difference in ultrafiltration and prior knowledge of the importance of the permeability of the peritoneal membrane in dialysis efficacy, the authors sought to investigate this specific variant’s effect on outcomes, namely mortality and technique failure.

In total, 139 patients (15%) out of 898 for whom data was available died during the 31 month follow up period. Additionally, 280 (31%) transferred to HD, 410 (46%) underwent transplantation. Patients with the TT genotype had a higher risk of the composite of death or technique failure than patients with CC (58% vs 42%, P=0.01) as well as higher risk of death from any cause (24% vs 15%, P=0.03). In univariate analyses, the TT genotype had a significantly higher risk of the composite outcome, with a hazard ratio of 1.51 (95% CI 1.13-2.02; P=0.005). Accounting for competing risks such as receiving transplantation, the TT genotype again was associated with a higher risk of the composite outcomes, with a hazard ratio of 1.67 (95% CI 1.24-2.25; P=0.001). Additionally, adjusting for other risk factors such as age, sex, CV disease status, diabetes status, and peritoneal solute transfer rate at baseline strengthened all associations, with a hazard ratio of 1.70 (95% CI 1.24-2.33; P=0.001) and a subdistribution hazard ratio of 1.89 (95% CI 1.40-2.56; P<0.001). Of note, the trend towards worse outcomes in patients with the TT genotype was seen in all subgroups and cohorts, with women being less likely than men to have poor outcomes.

Table 2, Morelle et al NEJM 2021

Finally, the authors describe a subgroup analysis to determine if those with the TT genotype had different ultrafiltration levels based on the type of dialysate used -- glucose-based crystalloid solution versus icodextrin-based colloid solution. They found a cohort of patients who had undergone PD with both types of dialysate who had a comparable frequency of the TT genotype (11%). These patients also had significantly lower ultrafiltration compared with the CC genotype, as expected and previously described above, but no association was found between osmosis induced by icodextrin and the presence of the risk variant, suggesting a possible role for use of colloid solution in these patients.

Discussion

This study identifies an important and previously unrecognized role of genetics in the efficacy of peritoneal dialysis. A relatively common genotype  is associated with  significant morbidity in patients undergoing PD, likely  due lower ultrafiltration leading to volume overload. A major limitation of this claim however is the fact that direct assessment of the patients’ volume status was not possible due to the nature of the study, so this assumption is made based solely on the roughly 150 ml difference in ultrafiltration in the absence of clinical exam findings.

In investigating the role of Aqp1, the authors found that mice with defective gene expression (Aqp1 +/- knockout mice) had lower net ultrafiltration than those with the wild-type genotype when glucose was used as the osmotic agent in PD. Using the colloid icodextrin eliminated this difference, likely because colloids induce water transport in an aquaporin-independent method. This held true for the human subjects in the subgroup analysis. In this study however, the change to a different dialysate was not done intentionally in patients with the mutation when the data was initially collected, so it is difficult to make assumptions on the primary outcomes without further trials.

Should we be screening for Aquaporin-1 variants in all patients initiating PD? Or would it be more prudent to selectively screen those with poor ultrafiltration? Better yet, would it be more cost-effective to empirically switch to colloid solutions in those with poor ultrafiltration without any genetic analysis? This study certainly provides a spark to light the conversation regarding the utility and practicality of precision medicine in PD, and potentially nephrology as a whole.

Summary prepared by Michael Turk, DO
Internal Medicine resident,
Allegheny General Hospital, Pittsburgh, PA
NSMC Intern, Class of 2021