Water removal during automated peritoneal dialysis assessed by remote patient monitoring and modelling of peritoneal tissue hydration

Water removal which is a key treatment goal of automated peritoneal dialysis (APD) can be assessed cycle-by-cycle using remote patient monitoring (RPM). We analysed ultrafiltration patterns during night APD following a dry day (APDDD; no daytime fluid exchange) or wet day (APDWD; daytime exchange). Ultrafiltration for each APD exchange were recorded for 16 days using RPM in 14 patients. The distributed model of fluid and solute transport was applied to simulate APD and to explore the impact of changes in peritoneal tissue hydration on ultrafiltration. We found lower ultrafiltration (mL, median [first quartile, third quartile]) during first and second vs. consecutive exchanges in APDDD (−61 [−148, 27], 170 [78, 228] vs. 213 [126, 275] mL; p < 0.001), but not in APDWD (81 [−8, 176], 81 [−4, 192] vs. 115 [4, 219] mL; NS). Simulations in a virtual patient showed that lower ultrafiltration (by 114 mL) was related to increased peritoneal tissue hydration caused by inflow of 187 mL of water during the first APDDD exchange. The observed phenomenon of lower ultrafiltration during initial exchanges of dialysis fluid in patients undergoing APDDD appears to be due to water inflow into the peritoneal tissue, re-establishing a state of increased hydration typical for peritoneal dialysis.


Peritoneal transport model
The spatially distributed model of peritoneal transport 1,2 was used to simulate changes -as a function of distance from the peritoneal cavity -of peritoneal tissue characteristics such as interstitial pressure, tissue hydration (formally interstitial fluid void volume ratio), and solute interstitial concentration, based in part on reported changes of intraperitoneal volume and solute concentrations occurring during a peritoneal dialysis dwell 3,4 . We explored intraperitoneal volume and pressure equilibration and associated changes of pressure and hydration of the peritoneal tissue close to its surface in contact with dialysate during the whole peritoneal dialysis exchange. The accumulation of water in the peritoneal tissue was calculated as the aggregated (over distance from the peritoneal cavity) difference between volume profiles of tissue hydration, i.e., for the first APD cycle (C1) as the difference between the end of the first dwell of APD exchange vs. end of day exchange, and for the second APD cycle (C2) as the difference between volume profiles obtained at the end of second APD dwell minus the volume at the end of C1. To obtain total water accumulation during the investigated periods, the values (estimated by the model per unit surface area) were multiplied by the corresponding effective peritoneal surface area that remained in contact with dialysate. The peritoneal transport system was modelled with blood and lymph capillaries spatially distributed within the peritoneal tissue space. The three-pore model with ultra-small (aquaporins), small and large pores was used to describe transport across the blood capillary wall 5,6 ; a similar model was proposed to describe transport across the whole peritoneal barrier by Rippe et al 7 . In this approach, water and solute peritoneal transport through the tissue depends not only on the local pressures, tissue hydration and concentrations that drive each transport component but also on the local properties of the tissue that are changing due to physiological responses to the ongoing treatment. The variability of the effective peritoneal surface area (EPSA, i.e., the surface that remains in contact with the dialysate) caused by the intraperitoneal volume changes was taken into account using a function previously described by Keshaviah et al 8,9 . The peritoneal tissue adapts to the increased intraperitoneal pressure and volume caused by the start of APD session by increased hydration of the peritoneal tissue that remains in contact with dialysate within a thin layer, close to the peritoneal cavity. However, in case of the APDDD regime, the previously overhydrated peritoneal tissue loses contact with dialysate during the day exchange. This results in local leakage of fluid into the peritoneal cavity from parts of the peritoneal tissue that remain without contact with dialysate leading towards a new physiological equilibrium with less hydration. The above-mentioned process occurring during the daytime exchange was taken into account, assuming free outflow of fluid from the parts of the tissue that were no longer in contact with dialysate.
To avoid overestimation of the peritoneal absorption (especially during the infusion and drainage procedures), a more detailed description was applied. Namely, two types of the overall peritoneal fluid absorption was considered: direct absorption by diaphragmatic lymphatics that are open to the peritoneal cavity (assumed to account for 30% of total peritoneal absorption and not present if the intraperitoneal volume is lower than 500 mL) and absorption to the peritoneal tissue, that accounts for 70% of total peritoneal absorption and is proportional to the EPSA 10 . The transport of solutes (glucose, sodium, urea, and creatinine) was also taken into account, as previously proposed 1 ; data not presented here.
Computer simulations were performed for a typical patient undergoing standard APDDD regime with 6 APD cycles for 90 min each with infused volume of 2 L of glucose 1.36%, followed by dry day albeit with infusion of 100 mL of glucose 1.36% to avoid pain associated with completely empty abdominal cavity. The peritoneal membrane characteristics was taken for a typical patient undergoing CAPD with 2.27% glucose solution based on results of fitting parameters for clinical data published by Heimbürger et al. 11 , cited in 2 . The residual volume was assumed to be 300 mL and the maximal rate of peritoneal absorption equal to 1 mL/min. Simulations of three days of dialysis exchanges were performed to assure complete adaptation of the peritoneal tissue to the treatment. Results are presented in Figures   2 and 3.

Additional numerical simulations
The discrepancy observed in APDDD between first and remaining APD cycles would conceivably become less apparent if a larger volume of dialysis fluid was infused during the daytime because then a larger fraction of the peritoneal tissue would remain in contact also during the daytime exchange resulting in less decline of peritoneal tissue hydration. Nevertheless, numerical simulations performed for the same, typical patient using instead of APDDD a wet day APDWD regime with 2 L of infused volume of glucose 1.36%, revealed that inter-cycle changes of tissue hydration, although less pronounced, might be still present in APDWD, see Figures S1. However, the obtained results cannot be directly compared with the actual clinical data of the studied APDWD group, which differed from the APDDD group and all except one of the APDWD patients received icodextrin-based dialysis fluid for the daytime exchange.
One may speculate, that for patients using icodextrin for the daytime exchange, higher UF associated with usage of icodextrin [12][13][14][15][16] would result in the even smaller decrease of tissue hydration and therefore lower discrepancy in water removal between the first vs. consecutive APD cycles than obtained in our simulations with 1.36% glucose for the wet day.
Since water removal depends on the glucose tonicity, higher glucose concentration used for the daytime exchange would result in higher ultrafiltration, larger peritoneal surface area remaining in contact with dialysis fluid during the daytime exchange, lower leakage of water from the tissue that loses its contact with peritoneal fluid, and, in consequence, less pronounced decrease of peritoneal tissue hydration towards a physiological state. However, due to the low infused volume used for the daytime exchange in APDDD the impact of fluid tonicity would remain within measurements error.