The functional integrity of the peritoneal membrane is essential for a successful treatment with peritoneal dialysis (PD). The evolutionary monitoring of peritoneal transport is part of the current routine of most PD units, and is usually performed based on the classic peritoneal equilibration test (PET).1 Although initially it was very focused on the transport of small solutes, over the last twenty years the standardized evaluation of ultrafiltration (UF) capacity and its fractions has gained progressive relevance. This interest has led to the progressive use of PET with 3.86/4.25% glucose solutions and to the emergence of variants of the procedure that facilitate the analysis.2 Of the various parameters examined, the transport of small solutes (creatinine and glucose) and the standardized UF capacity at 240' during PET, most commonly with 3.86/4.25% glucose solution,3–8 but also classic PET at 2.27/2.5%,8,9 currently constitute the core of the periodic evaluation of peritoneal functionalism.

Standardized and serial estimation of UF capacity by PET demands a sufficient degree of precision. However, it is not uncommon to observe significant variations between different studies in the same patient. When this occurs in successive estimates, the effects on the membrane of PD exposure and evolutionary incidences (especially peritoneal infections) usually justify the changes. Despite this, other factors may interfere with a correct assessment of this parameter. The need for good peritoneal drainage mechanics in the two changes associated with PET (the previous to and during the test itself) is a potential distorting element. The estimation of the residual volume (RV) at the beginning and end of PET can provide guidance on the level of imprecision introduced by this factor, and allow indirect estimation of the real UF during PET. However, measurement of RV is not part of the recommended routines for performing this test.8 To analyze the impact of this correction parameter, we conducted an observational and prospective study, whose main objective was to quantify RV in a large sample of incident and prevalent patients on PD, exploring the demographic and clinical profile of RV magnitude, and analyzed how it influences the categorization of UF capacity during PET with 3.86/4.25% glucose solution.

Table 1 shows the demographic and clinical characteristics and PD prescription pattern of the population studied. Table 2 shows the basic results of PET. Regarding estimated RV, it was higher if calculated from urea than from creatinine (p = 0.001), although both values showed a close correlation (r = 0.78 pre-PET samples, r = 0.82 post-PET samples, p < 0.0005).

The RVpost was significantly greater than RVpre (Table 2). As a logical consequence, UF for PET corrected for RV (494 ± 362 mL) was significantly higher than the standard estimate (449 ± 395 mL) (p < 0.0005). In addition, RVpre and RVpost showed a significant correlation, but of limited magnitude (r = 0.34, p = 0.001) (Table 3). Univariate analysis (Table 4) confirmed the sense of inconsistency, showing that none of the scrutinized parameter showed association with the double estimate of RV. This analysis was performed separately for RVpre and RVpost, given the relatively poor correlation between the two estimates (Table 4). We also did not observe an apparent effect of RV on small solute transport parameters or fall of sodium at 60 min (Table 4).

Patients with UF failure had lower RVpre (118.6 ± 66.4 mL) than those without UF failure (165.6 ± 108.0 mL)(p = 0.007), with no differences in RVpost (p = 0.73). Overall, 15 patients (12.9%) had a clinically significant variation (>200 mL) in UF estimation during PET when taking into account RV. In 12 of them standard UF underestimated actual UF during PET, and only in three cases it was overestimated. The standard estimate of UF indicated UF failure in 57 patients (49.1%), versus 45 (40.9%) if corrected UF was used (p < 0.0005). In other words, 21.1% of the patients who presented UF failure with standard estimation did not present such failure using corrected UF. Only one patient presented UF failure with corrected estimation, without UF failure by standard estimation.

In 27 patients, the study was repeated in a second PET, which confirmed once again the inconsistency of RV. Thus, the correlation between RVpre in first and second PET was 0.20 (p = 0.35), and the correlation for RVpost was 0.29 (p = 0.19).

Although the classical view of peritoneal transport assumed a sharp inverse relationship between small solute transport and UF capacity, the magnitude of that relationship is known to be limited.15 This apparent mismatch is primarily because UF capacity depends not only on the active surface area of ultrapores and small pores, but also on other factors, including the variable character of the rate of lymphatic reabsorption16 and acquired alterations in membrane function, associated with interstitial fibrosis and vasculopathy, that modify the ability of osmotic agents to generate and maintain water transfer (osmotic conductance).8 Awareness of these issues was the driving force behind the interest in standardizing UF during PET as an independent parameter of clear prognostic relevance.9 In addition, other factors may contribute to the discordance between solute transport and UF during PET, as well as to the not infrequent incongruence between the UF capacity observed in the clinical setting and that obtained in a standardized manner during PET. Thus, daily UF is highly conditioned by the prescription pattern (PD modality, glucose load, duration and volume of changes, icodextrin use); conversely, UF during PET represents a point estimate and is therefore subject to the specific variations that may occur during a single PD refill. Standardization of the test is an attempt to minimize these factors (for example, by separating it in time from episodes of peritoneal infection or catheter dysfunction, or by meticulously protocolizing the procedure), but the mechanics of the exchange cannot be predicted or modulated and, during a PET, these mechanics must be adequate in two changes (the previous one and that of the PET itself), for the test to be valid. One way to control for this factor is to estimate the RV (RVpre and RVpost).11 RV can be calculated quite accurately using dextran or albumin as markers,10,16,17 but this methodology has no application in clinical practice. The use of small solutes is less precise,11,12,18 but allows affordable estimates in clinical routine.11 Despite this, RV is not routinely estimated in most PD units, since the most common tendency is to simplify the development of the test.

The results of our study provide some interesting data. The quantification of RV obtained was consistent with the majority of previous studies11,16,18,19 although lower than other estimates.17 Of note is the inconsistency of the parameter, which varies markedly between patients and, above all, in each individual patient (RVpre vs. RVpost and in different PET). This circumstance suggests a significant component of randomness in the mechanics of peritoneal shift possibly linked, at least in part, to the mobility of the intraperitoneal portion of the catheter. Furthermore, the epidemiological analysis was unable to detect minimally consistent predictors of increased RV (Table 4). In particular, recent PD initiation, obesity, dialysate volume used in PET, PIP, presence of polycystic kidneys or history of abdominal surgery(s) did not show consistent relationship with RV. These results suggest that only estimation of RV in each PET would allow adequate control of this factor. Conversely, RV did not significantly influence, on average, the results regarding solute transport,16 the sodium fall at 60 min or UF during PET, nor did it change the categorization of UF capacity in most patients. However, 21.1% of patients who had UF failure based on standard criteria did not fall into this category if the UF was adjusted for RV. In general, the PET results are slightly affected on average by RV, but may influence a relatively low, but significant percentage of tests. This factor may help explain inconsistencies in successive estimates of the UF capacity of patients on PD.

The present study has significant limitations. The most important is the inexactitude inherent in the use of small solutes as markers for estimating RV. However, it should be emphasized that higher precision methods are not within the reach of clinical routine in most centers. Although good exchange mechanics is a prerequisite for a correct assessment of PET, correct positioning of the peritoneal catheter tip may influence the RV; this factor was not analyzed in our study. The strengths of the study include a prospective and rigorous design, a sufficient number of cases for analysis, and the inclusion of the main potential predictive variables for RV.

In conclusion, the consideration of RV affects little, on average, the estimation of UF capacity during PET at 3.86/4.25%, but it may have a significant impact in some cases, changing the categorization of UF failure in more than one-fifth of the patients so defined from the standard criteria. The amount of RV shows notable variations in each patient, and is not associated with a particular demographic or clinical profile. The data suggest the convenience of estimating this parameter in patients in whom PET has to be repeated due to unexpected inconsistencies in the estimation of UF capacity. In addition, although the estimation of RV is relatively simple, it increases the laboriousness of the PET protocol so that, according to our results, systematic monitoring of this parameter does not seem necessary.

The authors declare the absence of specific funding for this research.

The authors declare that they have no conflicts of interest.