Improvement in Sleep Apnea during Nocturnal PeritonealDialysis Is Associated with Reduced Airway Congestion andBetter Uremic Clearance

Sydney C. W. Tang,* Bing Lam,† Andrew S. H. Lai,‡ Clara B. Y. Pang,‡ Wai Kuen Tso,‡

Pek Lan Khong,‡ Mary S. M. Ip,† and Kar Neng Lai**Division of Nephrology and †Division of Respiratory Medicine, Department of Medicine, The University of Hong Kongand Queen Mary Hospital; and ‡Department of Diagnostic Radiology, Queen Mary Hospital, Hong Kong, China

Background and objectives: Among peritoneal dialysis (PD) patients, nocturnal PD (NPD) is known to improve sleep apneacompared with continuous ambulatory peritoneal dialysis (CAPD), but the contributing factors are unclear.

Design, setting, participants, and measurements: Thirty-eight incident ESRD patients underwent overnight polysomnog-raphy (PSG) during NPD and CAPD. Bioelectrical impedance analysis, magnetic resonance imaging of the upper airway, andurea kinetics (Kt/V) during sleep were measured on both occasions.

Results: The prevalence of severe sleep apnea (apnea-hypopnea index, AHI > 15/h) was 21.1% during NPD, and 42.1%during CAPD. Mean AHI increased from 9.6 � 2.7/h during NPD to 21.5 � 4.2/h during CAPD. Both obstructive and central apneaworsened after conversion to CAPD. NPD achieved greater reductions in total body water, hydration fraction, and net ultrafiltrationthan CAPD during sleep. Overnight peritoneal Kt/V and creatinine clearance were lower after conversion. Both peritoneal Kt/V andperitoneal creatinine clearance correlated with AHI, as did their changes after conversion. Volumetric magnetic resonance imagingrevealed reduced pharyngeal volumes and cross-sectional area, and tongue enlargement after conversion.

Conclusions: Improvement in sleep apnea during NPD versus CAPD is associated with better fluid and uremic clearanceand reduced upper airway congestion during sleep.

Clin J Am Soc Nephrol 4: 410–418, 2009. doi: 10.2215/CJN.03520708

S leep apnea has been reported in up to 50% to 70% ofpatients with end-stage renal disease (ESRD) (1), a value atleast 10 times higher than the prevalence reported in the

general population (2–4). The pathogenesis of sleep apnea inpatients with ESRD remains unclear. Unruh et al. (5) recentlyreported that patients on hemodialysis had a fourfold increase inprevalence of sleep-disordered breathing and nocturnal hypox-emia even after adjusting for cardiovascular morbidity and dia-betic status, compared with participants from the Sleep HeartHealth Study matched for age, gender, body mass index, and race,indicating that the pathophysiology of sleep apnea is uniquelyassociated with the development of chronic renal failure. Previousinvestigators have observed features of both central and obstruc-tive sleep apnea (OSA) in patients with ESRD (6–10), which sug-gest that its pathogenesis is related both to destabilization ofcentral respiratory control and upper airway occlusion. Moreover,nocturnal hypoxemia is a strong predictor for incident cardiovas-cular complications in the dialysis population (11).

Although sleep apnea is not corrected by conventional he-modialysis or peritoneal dialysis (9,10), it has been reversed

both by nocturnal hemodialysis (7) and kidney transplantation(6,8). Recently, we also reported improvement of sleep apneaby nocturnal cycler-assisted peritoneal dialysis (NPD) com-pared with conventional continuous ambulatory peritoneal di-alysis (CAPD) (12) through better fluid clearance during sleep.However, a direct proof of whether a reduction in total bodywater (TBW) content would alleviate airway occlusion is stilllacking. Furthermore, improved fluid removal per se remains aninadequate explanation of why central sleep apnea is also im-proved by NPD. In this study, we hypothesized that NPD, incomparison to CAPD, promotes better fluid and solute clear-ance during sleep and this may be associated with improve-ments of both the obstructive and central components of sleepapnea, respectively.

Materials and MethodsStudy Design

We performed a modified cross-over study. We made use of the fact thatin Hong Kong, all incident patients who required CAPD had to undergo,after Tenckhoff catheter placement, a temporary period of mandatorycycler-assisted NPD for approximately 8 wk while awaiting their turn forCAPD training. Such a design circumvents the logistic difficulties in per-forming cross-over studies in PD subjects caused by the inherent complex-ity in establishing a patient on a particular system of NPD or CAPD, aseach system has its own hardware, dialysis solution, and connectivemethodology, together with the possible training needed for the patient,family member, helper to use the particular system. Consenting adultsubjects underwent one PSG study toward the end of their 6 to 8 wk of

Received July 16, 2008. Accepted October 29, 2008.

Published online ahead of print. Publication date available at www.cjasn.org.

Correspondence: Prof. Kar Neng Lai, Department of Medicine, Queen MaryHospital, 102 Pokfulam Road, Hong Kong SAR, China. Phone: 852-2855-4250; Fax:852-2816-2863; E-mail: knlai@hku.hk

Copyright © 2009 by the American Society of Nephrology ISSN: 1555-9041/402–0410

NPD treatment. They then underwent training for CAPD, and a secondPSG study was performed as soon as they had been established on stableCAPD. All subjects had to have been deemed clinically euvolemic, withserum sodium between 135 and 145 mmol/L on study entry and beforePSG. The study protocol was reviewed and approved by the InstitutionalReview Board and Clinical Research Ethics Committee of the University ofHong Kong, and all patients gave written informed consent to participatein the study.

PatientsThe PSG data in 24 incident PD subjects and body water composition

data in 15 of these 24 subjects during NPD and CAPD were reportedpreviously (12). Here, we extended these data by recruiting another 22incident patients to undergo PSG and magnetic resonance imaging(MRI) of the upper airway during NPD and CAPD. Among them, 8completed the first PSG and MRI study but declined to participate inthe second measurement, and their data were not included in theanalyses. The final number of patients with evaluable PSG data was 38,of whom 14 were new. The final numbers of patients with evaluablebody water composition and MRI data were 29 and 14, respectively.

The underlying renal disease of the 38 patients (19 men; mean age,54.2 � 13.3 y) were diabetes mellitus in 14, IgA nephropathy in four,membranous nephropathy in one, lupus nephritis in one, focal segmentalglomerulosclerosis in two, pauci-immune crescentic glomerulonephritis intwo, anti-GBM disease in one, polycystic kidney disease in two, andunknown in 11. The average interval between the two sets of PSG record-ings was 4.12 � 0.63 mo, and there was no appreciable change in residualrenal function, neck circumference, neck:height ratio, body mass index,and other biophysical parameters over this period (Table 1).

PSGComprehensive overnight PSG was performed in hospital with the

Alice 3 or Alice 5 apparatus (Healthdyne, Atlanta, GA). All PSGs werescored manually according to standard criteria by an independentexpert in sleep medicine who was unaware of the mode of dialysis (13).The average number of episodes of apnea and hypopnea per hour ofsleep (apnea-hypopnea index [AHI]) was calculated as the summarymeasurement of sleep-disordered breathing.

Significant sleep apnea was defined as an AHI of � 15 events perhour of sleep. Apnea were classified as central if there was no chest andabdominal movement, or as obstructive if chest and abdomen movedparadoxically, and as mixed if an initial absence of ventilatory effort

was followed by an obstructive apnea pattern on resumption of effort.For practical purposes, all hypopneic or mixed events were classified asobstructive events.

Dialysis ProtocolDuring NPD, patients performed overnight exchanges of 10 to 12 L of

peritoneal dialysis fluid (PDF) at 2.0 to 2.2 L/cycle for 5 to 6 cycles bymeans of an automated cycler (HomeChoice, Baxter Healthcare, Mc-Gaw Park, IL). For CAPD, patients performed 3 to 4 daily exchanges of2 L PDF using either the UltraBag (Baxter Healthcare, Guangzhou,China), StaySafe or AndyDisk (Fresenius Medical Care, Bad Homburg,Germany), or Gambrosol Trio (Gambro Lundia AB, Lund, Sweden)system. The number of exchanges was clinically determined to achieveeuvolemia.

Residual Renal Function, Overnight Kt/V, and CreatinineClearance (CrCl) Assessment

Residual GFR was calculated according to the four-variable MDRDequation at the point of Tenckhoff catheter placement (14). The ade-quacy of dialysis indices were assessed by measuring total, peritoneal,and renal Kt/V and CrCl. The peritoneal component (nightly peritonealKt/V [pKt/V] and peritoneal CrCl [pCrCl]) was estimated solely fromthe overnight effluent dialysate on the night of PSG examination,whereas the renal component was estimated from 24-h urine collection.

For nightly pKt/V and pCrCl during NPD, all spent dialysatedrained from the PD cycler was collected in a bucket. The volume of thedialysate was accurately measured, and the dialysate was stirred vig-orously before a representative aliquot was obtained for calculation ofurea and creatinine concentrations. Because NPD patients did notundergo dialysis during the day, the nightly pKt/V and pCrCl was alsothe 24-h pKt/V and pCrCl. Renal clearance was derived from 24-hurine urea and creatinine divided by plasma urea and creatinine con-centrations, respectively. The nightly renal component of Kt/V andCrCl was estimated from the 24-h urine result normalized for time inbed (24-h result � time in bed in h/24 h).

For nightly pKt/V and pCrCl during CAPD, the spent dialysate fromthe overnight dwell of PDF was used for calculation of nightly clear-ance in a fashion similar to that for NPD. For 24-h clearance, all spentdialysate from day and night exchanges were used for calculation.Renal clearance (24-h and nightly result) was estimated as stated above.The crude body weight was used to calculate V according to themethod of Watson (15).

Table 1. Biophysical and clinical parameters while on NPD or CAPD (n � 38)

Parameter On NPD On CAPD

Body mass index (kg/m2, abdomen emptied) 23.2 � 3.6 23.9 � 4.2Neck circumference (cm) 37.1 � 3.4 36.6 � 3.1Neck:height ratio 0.219 � 0.024 0.222 � 0.022Systolic blood pressure (mmHg) 133 � 16 131 � 17Diastolic blood pressure (mmHg) 83 � 10 86 � 12No. of antihypertensive drugs 2.4 � 0.68 2.3 � 0.54Hemoglobin (g/dl) 9.5 � 1.3 10.1 � 1.9Serum albumin (g/L) 34.8 � 4.9 34.5 � 3.7Plasma bicarbonate (mmol/L) 24.0 � 3.2 26.4 � 3.6Residual urine output (L/24 h) 0.88 � 0.13 0.84 � 0.10Residual GFR (ml/min/1.73 m2 BSA) 8.77 � 2.2 8.69 � 1.9

Data are presented as mean (� SD). P � 0.05 for all comparisons during NPD or CAPD. GFR, glomerular filtration rate;BSA, body surface area.

Clin J Am Soc Nephrol 4: 410–418, 2009 NPD Improves SA via Fluid and Urea Removal 411

Bioelectrical Impedance AnalysisBioelectrical impedance analysis (BIA) to assess body water compo-

sition using Nutrigard-M Bioelectrical Impedance Analyzer (Data InputGmbH, Darmstadt, Germany) was performed on the night of PSGbefore starting NPD or CAPD night dwell, and in the morning aftercompleting NPD or the CAPD night dwell, with the peritoneum emp-tied. All measurements were made with the patient in the euvolemicstate, i.e., after the patients’ condition had become stabilized while theywere undergoing a particular mode of PD. Formulas for deriving bodywater compositions were described in our recent report (12).

Magnetic Resonance Imaging (MRI)Each patient underwent an MRI scan of the upper airway while

awake, within 6 h after completing NPD or the night dwell of CAPD.The protocol was modified from that reported by Ryan et al. (16) Briefly,the patient’s head was placed in a holding frame, and scans wereperformed during tidal breathing using a GE MRI 3T scanner. Sagittaland axial scans were acquired from T1-weighted images.

All MRI images were recorded digitally and interpreted by twoindependent observers who were unaware of the patient’s mode ofdialysis. Five anatomical sites were measured (Figure 1) and weredefined as follows:

(1) Nasopharynx was airway bordered by the soft palate anteriorly,the nasal turbinates anteriorly and superiorly, and the adenoids poste-riorly and superiorly. The inferior border was at the level of the inferiortip of the uvula.

(2) Oropharynx was the aerated space bordered by the hard palatesuperiorly, the tongue inferiorly, and the soft palate posteriorly.

(3) Hypopharynx was the aerated space bordered by the posterioraspect of the tongue anteriorly, the posterior pharyngeal wall posteri-orly, and the inferior aspect of the soft palate superiorly. The inferiorborder was defined by the inferior extent of the base of the tongue.

(4) Minimal pharyngeal cross-sectional area was taken from the axialslice at which the pharyngeal area was the smallest.

(5) Volume of the tongue measured on sagittal images with axialcorrelation.

Volume measurements were then calculated on the basis of thesagittal areas of a series of contiguous slices and on the thickness of thescan slices.

Statistical AnalysesData were expressed as means � SD unless otherwise specified.

Statistical analyses were performed using SPSS for Windows softwareversion 14.0 (Statistical Package for the Social Sciences Inc., Chicago,IL). Comparisons between groups were performed by �2 test for cate-gorical data, and Mann-Whitney U test for continuous data. Because ofindividual variations in the volumetric analyses of the upper airwayparameters as shown on MRI, all numeric changes were expressed aspercentage variation with respect to values obtained during NPD.Nonparametric paired-sample Wilcoxon signed rank test was used todetermine changes in sleep disturbance parameters, nightly and 24-hKt/V and creatinine clearances, and volumetric measurements on MRIstudies during NPD and CAPD. Comparison of within-group differ-ences in prevalence of sleep apnea during NPD or CAPD was per-formed using the McNemar test. Relationships between absolute ordifferential nightly pKt/V or pCrCl and AHI were analyzed usingSpearman rank correlation. A P value of � 0.05 was considered signif-icant. All probabilities were two-tailed.

ResultsThe prevalence of sleep apnea (AHI � 15/h) was 21.1% (n �

8) during NPD and 42.1% (n � 16) during CAPD (Table 2).

Figure 1. Anatomic demarcations of the predefined landmarks(see text for definition) of the upper airway illustrated by MRI.(A) Representative sagittal image showing the anatomic bound-aries for measuring the areas of the oropharynx (1), tongue (2),nasopharynx (3), and hypopharynx (4). The corresponding vol-umes were then calculated on the basis of the sagittal areas ofa series of contiguous slices and on the thickness of the scanslices. (B) Representative axial image showing the slice at whichthe pharyngeal area was the smallest (1) among contiguousaxial slices of the pharynx.

412 Clinical Journal of the American Society of Nephrology Clin J Am Soc Nephrol 4: 410–418, 2009

When the AHI cutoff was lowered to 10/h, the prevalence was26.3% (n � 10) during NPD, and 50% (n � 19) during CAPD. Asa group, the absolute AHI values increased from 9.6 � 2.7/hduring NPD to 21.5 � 4.2/h during CAPD (P � 0.001 forcomparison before and after transfer to CAPD). Both obstruc-tive and central components of sleep apnea became signifi-cantly more severe, and the individual changes in AHI valuesduring the two different modes of PD are shown in Figure 2.The overall sleep pattern during NPD or CAPD was summa-rized in Table 2. Although there was no significant difference inthe total sleep time during NPD or CAPD, the duration of sleepwith hypoxia (oxygen saturation � 90%) was substantiallylonger during CAPD (P � 0.025). In addition, there were sig-nificantly more frequent arousals after conversion to CAPD(P � 0.001).

Multifrequency BIA revealed no difference in absolute bodywater content before NPD or the night dwell of CAPD (Figure3), alleviating concerns that there may be significant change inbody water composition in the brief interval between NPD andCAPD. Applying BIA just before and after NPD or the nightdwell of CAPD to calculate the change in body fluid composi-tion during sleep, NPD achieved a significantly larger volumeof fluid removal, as reflected by greater reductions in absolutebody water contents (Figure 4A). Compared with CAPD, NPDled to a 2.2-fold (P � 0.003), 1.9-fold (P � 0.005), and 1.6-fold(P � 0.003) more intense reduction in total body water, extra-cellular body water, and intracellular body water, respectively,during sleep. When total body water was expressed as hydra-tion fraction (percentage of body weight due to water only),there was a sixfold difference in reduction achieved by NPDversus CAPD (�3.77 � 0.49 versus �0.62 � 0.41%, P � 0.001).Similarly, the percentage change in hydration before and afterNPD or CAPD was also significantly higher in the former (P �

0.001) (Figure 4B).Volumetric MRI measurements showed that there were sig-

nificant reductions in the nasopharyngeal (�24.5 � 9.6%) and

oropharyngeal (�34.0 � 11.6%) volumes, minimal pharyngealcross-sectional area (�16.8 � 4.7%), and increase in tonguevolume (�8.22 � 2.4%) after conversion from NPD to CAPD(Figure 5). There were also numerical reductions in the hypo-pharyngeal volume (�21.6 � 11.9%), albeit not reaching statis-tical significance. When the contiguous volumes of the naso-,oro-, and hypopharynxes were examined in aggregate, therewas a 30.8 � 9.2% reduction (P � 0.004) after conversion. Thereduction in this aggregate volume also correlated with thechange in the indices of obstructive sleep apnea (r � �0.565,P � 0.035).

Urea kinetics analyses of the overnight spent dialysate dur-ing the night of PSG examination revealed that nightly pKt/Vand pCrCl were both significantly higher during NPD thanCAPD (pKt/V: 0.289 � 0.070 versus 0.065 � 0.023 per night, P �

0.001; pCrCl: 4.87 � 1.22 versus 1.84 � 0.67 L per night, P �

0.001) (Figure 6). The difference in nocturnal clearance betweenNPD and CAPD remained significant (total Kt/V: 0.338 � 0.078versus 0.118 � 0.051 per night, P � 0.001; total CrCl: 6.42 � 1.50versus 3.22 � 1.20 L per night, P � 0.001) even after taking intoaccount the estimated endogenous component of Kt/V andCrCl that was derived from 24-h urine and normalized againstthe time in bed for each patient (Table 3). To further substan-tiate the impact of peritoneal clearance on sleep apnea, weexamined the relationship between overnight pKt/V or pCrCland AHI in all subjects during NPD and CAPD. There weresignificant negative correlations between pKt/V and AHI (r �

�0.332, P � 0.003), and between pCrCl and AHI (r � �0.319,P � 0.005). The difference in overnight pKt/V or pCrCl be-tween each mode of PD was significantly correlated with thedifference in AHI for each individual patient (Figure 7).

DiscussionWe have extended our fixed intervention study cohort to

investigate the mechanisms through which NPD improvessleep apnea compared with CAPD. Our results reaffirmed a

Table 2. Polysomnographic data while on NPD or CAPD (n � 38)

On NPD On CAPD P

Total sleep time (h) 5.0 � 1.60 5.55 � 1.41 NSSleep efficiencya 57.4 � 14.4 62.8 � 12.7 NSStage of sleep (% of total sleep time)

rapid eye movement 18.2 � 10.5 16.7 � 8.6 NSstage 1, 2 78.8 � 22.1 79.3 � 20.0 NSslow wave (stage 3, 4) 1.9 � 3.2 4.0 � 3.6 NS

AHI (no./h) 9.6 � 2.74 21.5 � 4.15 �0.001subjects with AHI � 15 (n, %) 8 (21.1%) 16 (42.1%) 0.008subjects with AHI � 10 (n, %) 10 (26.3%) 19 (50%) 0.004subjects with AHI � 5 (n, %) 11 (28.9%) 22 (57.9%) 0.001

Duration with oxygen saturation � 90% (min) 64.3 � 15.8 94.8 � 20.5 0.025Arousals (no./h) 16.6 � 9.8 26.6 � 16.4 �0.001Periodic leg movement (no./h) 0.5 � 0.21 1.4 � 0.35 0.007

aRatio of total sleep time to total time in bed. Data are presented as mean (� SD). AHI, apnea-hypopnea index (i.e., averagenumber of episodes of apnea and hypopnea per hour of sleep).

Clin J Am Soc Nephrol 4: 410–418, 2009 NPD Improves SA via Fluid and Urea Removal 413

significant worsening of sleep apnea after conversion fromNPD to CAPD, as reflected by substantial increases in AHI,

increased prevalence of sleep apnea, prolongation of the dura-tion of hypoxia, and more frequent arousals during sleep. On amechanistic basis, we showed that a major component thatalleviated the severity of sleep apnea during NPD in compar-ison to CAPD was more intensive ultrafiltration, which led tolarger reductions in total body water and hydration fractionduring sleep, as indicated by BIA measurements. This maytranslate indirectly to reduced airway edema and therefore lesssevere OSA.

Here, we performed MRI of the upper airway to detect anysubtle changes in the airway during each of the two modes ofPD. Volumetric MRI is a powerful tool for detecting changes inthese structures (17). A practical disadvantage of MRI is pa-tient-perceived discomfort and claustrophobia. A considerableproportion (up to 36%) of patients in this study, having under-gone the first MRI, declined to undergo a second.

Among patients who underwent MRI during NPD andCAPD, there were reduced aggregate pharyngeal volumes and

Figure 2. Individual changes in overall apnea-hypopnea index (A), obstructive apnea index (B), and central apnea index (C) in 38subjects during NPD and after conversion to CAPD. Horizontal bars represent mean � SEM values during each mode of PD. †P �0.001 compared with values while on NPD.

Figure 3. Absolute body water composition before NPD (open bars)or the overnight dwell of CAPD (solid bars) using bioelectrical im-pedance analyses (n � 29). Error bars are mean � SD. TBW, totalbody water; ECW, extracellular water; ICW, intracellular water.

414 Clinical Journal of the American Society of Nephrology Clin J Am Soc Nephrol 4: 410–418, 2009

minimal pharyngeal cross-sectional area, together with con-comitant increase in tongue volume after switching from NPDto CAPD. Tongue enlargement may also represent fluid accu-mulation in other soft tissues bordering the airway such as thenuchal and peripharyngeal areas, which are difficult to assessin an objective manner because of the lack of clear-cut anatom-ical demarcations. It is conceivable that such anatomic alter-ations favoring narrowing of the upper airway may give rise toOSA during sleep in predisposed patients when there is areduced tone of the supporting airway musculature. This isevidenced by the correlation between the reduction in theaggregate pharyngeal volume and the worsening of the ob-structive component of sleep apnea after conversion to CAPD.Furthermore, it is envisaged that fluid accumulation in the

upper airway will increase airflow resistance of the pharynx, asreflected by the decrease in minimal pharyngeal cross-sectionalarea on switching to CAPD. In support of this notion, Chiu et al.(18) recently applied lower body positive pressure using anti-shock trousers to displace fluid from the legs to the upper body,and found that the maneuver resulted in increased pharyngealairflow resistance. Beecroft et al. (19) recently showed thatpharyngeal narrowing contributes to the pathogenesis of OSAin dialysis-dependent patients. Furthermore, in patients con-verted from conventional to nocturnal hemodialysis, there wasan increase in pharyngeal size together with improvement insleep apnea (20). This study is the first to show a reduction inairway diameter on conversion from NPD to CAPD.

Apart from improving OSA, our data showed that NPD alsoimproved the central component of sleep apnea. It remains un-clear why a modality switch that principally affects fluid balance

Figure 4. Changes in body water composition during sleep afterNPD (open bars) or the overnight dwell of CAPD (solid bars)using bioelectrical impedance analyses (n � 29). (A) Absolutechanges in total body water (TBW), extracellular water (ECW),and intracellular water (ICW), calculated by the difference in theabsolute body water content before and after each mode of PDwhich reflects the net fluid loss during sleep. (B) Changes inhydration fraction (TBW [in liters]/body weight [in kilograms] �100%) and percentage changes in hydration fraction (postdialyticminus predialytic hydration fraction/predialytic hydration frac-tion) before and after each mode of PD, which reflects the net fluidshifts during sleep. Error bars are mean � SEM.

Figure 5. Percentage volumetric change in the various anatomicsites of the upper airway after conversion to CAPD (n � 14).*P � 0.004, †P � 0.04, ‡P � 0.02 versus values obtained duringNPD. NP, nasopharynx; OP, oropharynx; HP, hypopharynx;MPXA, minimal pharyngeal cross-sectional area. Error bars aremean � SEM.

Figure 6. Absolute peritoneal solute clearance during sleep.Kt/V and CrCl that was derived from the overnight effluentduring NPD or CAPD for each patient was computed. Stippledand hatched bars represent overnight pKt/V and pCrCl, re-spectively. pKt/V, peritoneal Kt/V; pCrCl, peritoneal creati-nine clearance.

Clin J Am Soc Nephrol 4: 410–418, 2009 NPD Improves SA via Fluid and Urea Removal 415

and airway patency, and hence resistance, should affect centralsleep apnea. One possible explanation, learned from cross-overstudies in hemodialysis subjects (7), is through correction of met-abolic acidosis and hypocapnia using nocturnal hemodialysis. Theattenuation of hypocapnia may then disinhibit the central respi-ratory center that became suppressed in the face of hypocapnia.However, this is not a likely explanation for NPD, which is abiochemically modest procedure compared with HD in terms ofacidosis correction, as reflected by similar serum bicarbonate lev-els during NPD or CAPD. An alternative mechanism that wecontemplated was a difference in the rate of removal of uremicwaste products. Because residual renal function remained un-changed during NPD or CAPD, we focused on the peritonealcomponent of solute clearance during sleep and calculated thenightly Kt/V and CrCl contributed solely by PD during sleep. Theonly demonstrated difference between NPD and CAPD, apartfrom fluid status, was the much higher nightly pKt/V and pCrClduring NPD, which could represent a circumstantial evidence ofthe effect of uremic toxins on sleep apnea. Even after taking into

account the estimated contribution of endogenous clearance dur-ing the night, the total nighttime clearance was still significantlyhigher during NPD than during CAPD. Although it could beargued that the daytime component of CAPD that has to besignificantly higher than that of NPD could lead to a better statusat the beginning of sleep, several lines of evidence suggest thatnocturnal clearance plays a more important role: (1) the negativecorrelation between absolute pKt/V and pCrCl with AHI values;(2) the correlation between the decrease in pKt/V and pCrCl withthe rise in AHI on conversion from NPD to CAPD; (3) patientswho underwent conventional hemodialysis that had to confer asignificantly higher daytime clearance than that of nocturnal he-modialysis still had significantly worse sleep apnea during day-time versus nocturnal hemodialysis (7); and (4) the decrease inchemoreflex responsiveness to carbon dioxide after conversionfrom conventional to nocturnal hemodialysis, which is associatedwith increased uremic clearance during sleep (21).

There are several limitations to this study. First, we did notmeasure airway pressure and pharyngeal resistance, because it

Table 3. Dialysis-related parameters while on NPD or CAPD (n � 38)

Parameter On NPD On CAPD P

24-h Kt/Vtotal 0.427 � 0.101 0.358 � 0.097 �0.001peritoneala 0.289 � 0.070 0.207 � 0.071 �0.001renalb 0.138 � 0.062 0.151 � 0.121 0.908

24-h CrCl (L)total 9.20 � 2.64 8.32 � 2.01 0.024peritoneala 4.87 � 1.22 4.36 � 1.16 0.069renalb 4.33 � 2.38 3.96 � 2.51 0.233

Nightly Kt/V (time in bed), Ltotal 0.338 � 0.078 0.118 � 0.051 �0.001peritonealc 0.289 � 0.070 0.065 � 0.023 �0.001renald 0.049 � 0.023 0.053 � 0.043 0.523

Nightly CrCl (time in bed), Ltotal 6.42 � 1.50 3.22 � 1.20 �0.001peritonealc 4.87 � 1.22 1.84 � 0.67 �0.001renald 1.55 � 0.89 1.38 � 0.91 0.179

Net ultrafiltration (ml per night) 1,729 � 799 (range, 170 to 3,053) 394 � 315 (range, �100 to 1,050) �0.001Body weight (kg)

night before PSG 56.2 � 9.8 56.8 � 10.8 0.084morning after PSG 54.1 � 9.7 56.0 � 10.6 �0.001% change �3.7 � 1.7% �1.4 � 1.1% �0.001

Peritoneal transport statuse

overall D/PCr at 4 h (n, %) N/A 0.66 � 0.093low 0low average 21 (55.2%)high average 13 (34.2%)high 4 (10.5%)

Data are presented as mean (� SD).aBased on overnight peritoneal effluent for NPD and 24-h effluent for CAPD.bBased on 24-h urine collection.cBased on overnight peritoneal effluent.dEstimation based on 24-h result � time in bed (h) /24 h.eAccording to the method of peritoneal equilibration test and classification of transport status by Twardowski (31).

416 Clinical Journal of the American Society of Nephrology Clin J Am Soc Nephrol 4: 410–418, 2009

induces too much discomfort in a patient undergoing CAPDand PSG monitoring to place another two catheter transducersat the naso/hypopharynxes and a tightly fitting face mask forflow recording during the same night (18). Second, the numberof subjects is small. It was difficult to recruit subjects to un-dergo PSG and MRI in succession on two occasions, as reflectedby the high drop-out rate among consenting subjects. Third,our results suggest, but do not directly prove, that better noc-turnal clearance improves apnea. To achieve this goal willrequire adding day exchanges to the NPD phase to match thedaytime clearance during the subsequent CAPD phase, whichwas not feasible in our institutional setting. Likewise, reducedairway congestion on NPD, as shown on MRI, provides onlycircumstantial evidence that this is linked to improved sleepapnea. Nevertheless, this contention is supported by the closeassociation between airway edema and sleep apnea (22–25).Finally, we have not studied the changes in intraperitonealpressure and the correlation with oxygen saturation. This arisesmainly from our concern with introducing a potential source of

contamination during PD caused by extra manipulation andconnection to a pressure transducer for continuous pressurerecording (26). Nevertheless, time-averaged intraperitonealpressure may be higher during CAPD than during NPD be-cause of the permanent load of peritoneal dialysate during thenight in CAPD. Theoretically, this may have two effects: (1)changes in respiratory mechanics, such as reduction in vitalcapacity (27), although recent studies showed that sleep apneais unrelated to pulmonary function measurements [12,28] andthat flow-volume curve indices have no value in predictingsleep apnea (29); (2) more pressure effect on the diaphragm. Tomaintain the same tidal volume, a more negative intrathoracicpressure has to be generated, which may cause arousal andairway collapsibility, which could in turn lead to sleep apnea.Determining whether the difference in intraperitoneal pressureduring CAPD and NPD causes a large enough change in in-trathoracic pressure to bring about worsening of sleep apnea isbeyond the scope of the present study.

In conclusion, our results suggest that while many factorsincluding body habitus, comorbidities, and psychosocial cir-cumstances (30) contribute to the development sleep apnea inpatients with chronic renal failure, fluid status and solute clear-ance rates may be associated with the development of sleepapnea in PD patients (Figure 8).

AcknowledgmentsThis study is supported by the Seed Funding Programme for Basic

Research of the University of Hong Kong. The authors are grateful toAgnes Lai and Barbara Law (Sleep Laboratory), Sandra Luen (Divisionof Nephrology), Helena Leung and all nursing staff (K18N Dialysis

Figure 7. Correlation between the difference in overnight pKt/V(A) or pCrCl (B) and the change in AHI following conversionfrom NPD to CAPD. pKt/V, peritoneal Kt/V; pCrCl, peritonealcreatinine clearance; AHI, apnea-hypopnea index.

Figure 8. Postulated pathophysiologic pathways leading tosleep apnea in chronic renal failure. Volume overloaded statuscauses redistribution of body water during the recumbent pos-ture and gives rise to upper airway congestion as evidenced byreduced pharyngeal luminal volumes, pharyngeal narrowing,and tongue enlargement, leading predominantly to obstructivesleep apnea. Accumulation of uremic toxins induces metabolicacidosis and hypocapnia and increases chemosensitivity to car-bon dioxide, leading predominantly to central sleep apnea.Other comorbid conditions and body habitus also play anadjuvant role. Nocturnal dialysis or renal transplantation mayalleviate volume and uremic burdens and hence improve sleepapnea. Solid arrows, enhancing effect; interrupted arrows, in-hibitory effect. NPD, nocturnal peritoneal dialysis; NHD, noc-turnal hemodialysis; RTx, renal transplantation

Clin J Am Soc Nephrol 4: 410–418, 2009 NPD Improves SA via Fluid and Urea Removal 417

Unit) for coordinating the PSG and MRI studies; Jack Lam (SleepLaboratory) for scoring all PSGs manually; Kan Ming Lo (Sleep Labo-ratory) for performing all BIA measurements; and Suki Ho (DialysisUnit) for performing all urea kinetics computation.

DisclosuresNone.

References1. Kraus MA, Hamburger RJ: Sleep apnea in renal failure.

Adv Perit Dial 13: 88–92, 19972. Young T, Blustein J, Finn L, Palta M: Sleep-disordered

breathing and motor vehicle accidents in a population-based sample of employed adults. Sleep 20: 608–613, 1997

3. Ip MS, Lam B, Lauder IJ, Tsang KW, Chung KF, Mok YW,Lam WK: A community study of sleep-disordered breath-ing in middle-aged Chinese men in Hong Kong. Chest 119:62–69, 2001

4. Ip MS, Lam B, Tang LC, Lauder IJ, Ip TY, Lam WK: Acommunity study of sleep-disordered breathing in middle-aged Chinese women in Hong Kong: Prevalence and gen-der differences. Chest 125: 127–134, 2004

5. Unruh ML, Sanders MH, Redline S, Piraino BM, UmansJG, Hammond TC, Sharief I, Punjabi NM, Newman AB:Sleep apnea in patients on conventional thrice-weeklyhemodialysis: Comparison with matched controls fromthe Sleep Heart Health Study. J Am Soc Nephrol 17:3503–3509, 2006

6. Auckley DH, Schmidt-Nowara W, Brown LK: Reversal ofsleep apnea hypopnea syndrome in end-stage renal dis-ease after kidney transplantation. Am J Kidney Dis 34: 739–744, 1999

7. Hanly PJ, Pierratos A: Improvement of sleep apnea inpatients with chronic renal failure who undergo nocturnalhemodialysis. N Engl J Med 344: 102–107, 2001

8. Langevin B, Fouque D, Leger P, Robert D: Sleep apneasyndrome and end-stage renal disease. Cure after renaltransplantation. Chest 103: 1330–1335, 1993

9. Mendelson WB, Wadhwa NK, Greenberg HE, Gujavarty K,Bergofsky E: Effects of hemodialysis on sleep apnea syn-drome in end-stage renal disease. Clin Nephrol 33: 247–251,1990

10. Wadhwa NK, Seliger M, Greenberg HE, Bergofsky E, Men-delson WB: Sleep related respiratory disorders in end-stage renal disease patients on peritoneal dialysis. PeritDial Int 12: 51–56, 1992

11. Zoccali C, Mallamaci F, Tripepi G: Nocturnal hypoxemiapredicts incident cardiovascular complications in dialysispatients. J Am Soc Nephrol 13: 729–733, 2002

12. Tang SC, Lam B, Ku PP, Leung WS, Chu CM, Ho YW, IpMS, Lai KN: Alleviation of sleep apnea in patients withchronic renal failure by nocturnal cycler-assisted perito-neal dialysis compared with conventional continuous am-bulatory peritoneal dialysis. J Am Soc Nephrol 17: 2607–2616, 2006

13. Meoli AL, Casey KR, Clark RW, Coleman JA, Jr., Fayle RW,Troell RJ, Iber C: Hypopnea in sleep-disordered breathingin adults. Sleep 24: 469–470, 2001

14. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D:A more accurate method to estimate glomerular filtrationrate from serum creatinine: A new prediction equation.

Modification of Diet in Renal Disease Study Group. AnnIntern Med 130: 461–470, 1999

15. Watson PE, Watson ID, Batt RD: Total body water vol-umes for adult males and females estimated from simpleanthropometric measurements. Am J Clin Nutr 33: 27–39,1980

16. Ryan CF, Lowe AA, Li D, Fleetham JA: Magnetic reso-nance imaging of the upper airway in obstructive sleepapnea before and after chronic nasal continuous positiveairway pressure therapy. Am Rev Respir Dis 144: 939–944,1991

17. Welch KC, Foster GD, Ritter CT, Wadden TA, Arens R,Maislin G, Schwab RJ: A novel volumetric magnetic reso-nance imaging paradigm to study upper airway anatomy.Sleep 25: 532–542, 2002

18. Chiu KL, Ryan CM, Shiota S, Ruttanaumpawan P, Arzt M,Haight JS, Chan CT, Floras JS, Bradley TD: Fluid shift bylower body positive pressure increases pharyngeal resis-tance in healthy subjects. Am J Respir Crit Care Med 174:1378–1383, 2006

19. Beecroft JM, Hoffstein V, Pierratos A, Chan CT, McFarlanePA, Hanly PJ: Pharyngeal narrowing in end-stage renaldisease: Implications for obstructive sleep apnoea. Eur Re-spir J 30: 965–971, 2007

20. Beecroft JM, Hoffstein V, Pierratos A, Chan CT, McFarlaneP, Hanly PJ: Nocturnal haemodialysis increases pharyn-geal size in patients with sleep apnoea and end-stage renaldisease. Nephrol Dial Transplant 23: 673–679, 2008

21. Beecroft JM, Duffin J, Pierratos A, Chan CT, McFarlane P,Hanly PJ. Decreased chemosensitivity and improvement ofsleep apnea by nocturnal hemodialysis. Sleep Med 2008 Jan26 [epub ahead of print]

22. Anastassov GE, Trieger N: Edema in the upper airway inpatients with obstructive sleep apnea syndrome. Oral SurgOral Med Oral Pathol Oral Radiol Endod 86: 644–647, 1998

23. Blankfield RP, Ahmed M, Zyzanski SJ: Effect of nasal con-tinuous positive airway pressure on edema in patientswith obstructive sleep apnea. Sleep Med 5: 589–592, 2004

24. Blankfield RP, Ahmed M, Zyzanski SJ: Idiopathic edema isassociated with obstructive sleep apnea in women. SleepMed 5: 583–587, 2004

25. Fleetham JA: Upper airway imaging in relation to obstruc-tive sleep apnea. Clin Chest Med 13: 399–416, 1992

26. Lai KN, Pun CO, Leung JW: Study of hepatic venouswedge and intraperitoneal pressures in cirrhotic patientswith refractory ascites treated by dialytic ultrafiltration. JGastroenterol Hepatol 4: 325–330, 1989

27. Durand PY, Chanliau J, Gamberoni J, Hestin D, Kessler M:Measurement of hydrostatic intraperitoneal pressure: Anecessary routine test in peritoneal dialysis. Perit Dial Int16[Suppl 1]: S84–S87, 1996

28. Hoffstein V, Oliver Z: Pulmonary function and sleep ap-nea. Sleep Breath 7: 159–165, 2003

29. Campbell AH, Guy PA, Rochford PD, Worsnop CJ, PierceRJ: Flow-volume curve changes in patients with obstruc-tive sleep apnoea and brief upper airway dysfunction.Respirology 5: 11–18, 2000

30. Parker KP: Sleep disturbances in dialysis patients. SleepMed Rev 7: 131–143, 2003

31. Twardowski ZJ. PE: T–a simpler approach for determiningprescriptions for adequate dialysis therapy. Adv Perit Dial6: 186–191, 1990

418 Clinical Journal of the American Society of Nephrology Clin J Am Soc Nephrol 4: 410–418, 2009