Alkaline phosphatase treatment improves renal function in severesepsis or septic shock patients*

Suzanne Heemskerk, PhD; Rosalinde Masereeuw, PhD; Olof Moesker; Martijn P. W. J. M. Bouw;Johannes G. van der Hoeven, MD, PhD; Wilbert H. M. Peters, PhD; Frans G. M. Russel, PhD;Peter Pickkers, MD, PhD; on behalf of the APSEP Study Group

Septic shock is the most com-mon cause of death in patientsrequiring intensive care, withan overall mortality rate of ap-

proximately 50% to 60% (1). The kidney

is one of the first organs to fail in approx-imately 20% of patients, associated with70% mortality in septic acute kidney in-jury (AKI), highlighting the urgent needfor interventions to protect the kidneys(2, 3). Despite important advances in un-derstanding its pathophysiology, till nowno clinical studies are available that showan improvement of renal function in pa-tients with severe sepsis or septic shock.

Among various other effects, excessiveproduction of nitric oxide (NO) duringsepsis, by activation of inducible NO syn-thase (iNOS), causes systemic vasodila-tion, which in turn leads to renal vaso-constriction with sodium and waterretention (2). In addition and in agree-ment with animal models, we previouslydemonstrated that the induction of renaliNOS, which is constitutively expressedin the kidney (4), is associated with prox-imal tubule injury during systemic in-flammation in humans (5).

Alkaline phosphatase (AP) functions as ahost defense molecule at physiologic pHwhen endotoxin is used as a substrate (6),

and is present in many cells and organs andabundant in kidney tissue (7, 8). In severalanimal studies, exogenous administrationof AP resulted in attenuation of the inflam-matory response induced by Gram-negativebacteria and lipopolysaccharide (LPS) andreduced mortality (9–13). It is suggestedthat upregulation of endogenous glomeru-lar AP reflects a local host defense principleagainst LPS in the kidney (14), and that APcan be inactivated by reactive oxygen spe-cies and peroxynitrite generated duringsepsis. Additionally, the expression of iNOSprotein levels correlates negatively with theactivity of AP in duodenum of rats exposedto LPS (11). In this exploratory study, wetested the hypothesis that administration ofexogenous AP attenuates the expression ofrenal iNOS and subsequent NO production,thereby reducing kidney damage in pa-tients with severe sepsis or septic shock.

MATERIALS AND METHODS

Patients. The present study was approvedby the Ethics Committee, and written in-formed consent was obtained from each pa-

*See also p. 740.From the Department of Pharmacology and Toxi-

cology (SH, RM, FGMR), Nijmegen Centre for MolecularLife Sciences; and Department of Intensive Care Med-icine (OM, MPWJMB, JGH, PP), Nijmegen UniversityCentre for Infectious Diseases (JGH), Department ofGastroenterology (WHMP), Radboud University Nijme-gen Medical Centre, Nijmegen, The Netherlands.

AM-Pharma BV funded the conduct of the studyand the main statistical analysis (by SGS Biopharma;Wavre, Belgium). S. Heemskerk was supported by agrant from the Netherlands Organisation for ScientificResearch (NWO); P. Pickkers is a recipient of a ClinicalFellowship grant of the Netherlands Organisation forScientific Research (NWO-ZonMw), which supportedthe evaluations and analyses of renal markers.

A supplementary table is available online at http://www.ccmjournal.org.

For information regarding this article, E-mail:p.pickkers@ic.umcn.nl

Copyright © 2009 by the Society of Critical CareMedicine and Lippincott Williams & Wilkins

DOI: 10.1097/CCM.0b013e31819598af

Objective: Alkaline phosphatase (AP) attenuates inflammatoryresponses by lipopolysaccharide detoxification and may preventorgan damage during sepsis. To investigate the effect of AP inpatients with severe sepsis or septic shock on acute kidneyinjury.

Design and Setting: A multicenter double-blind, randomized,placebo-controlled phase IIa study (2:1 ratio).

Patients: Thirty-six intensive care unit patients (20 men/16women, mean age 58 � 3 years) with a proven or suspectedGram-negative bacterial infection, >2 systemic inflammatory re-sponse syndrome criteria (<24 hours), and <12 hours end-organdysfunction onset were included.

Intervention: An initial bolus intravenous injection (67.5 U/kgbody weight) over 10 minutes of AP or placebo, followed bycontinuous infusion (132.5 U/kg) over the following 23 hours and50 minutes.

Measurements and Main Results: Median plasma creatininelevels declined significantly from 91 (73–138) to 70 (60–92)�mol/L only after AP treatment. Pathophysiology of nitric oxide

(NO) production and subsequent renal damage were assessed ina subgroup of 15 patients. A 42-fold induction (vs. healthy sub-jects) in renal inducible NO synthase expression was reduced by80% � 5% after AP treatment. In AP-treated patients, the in-crease in cumulative urinary NO metabolite excretion was atten-uated, whereas the opposite occurred after placebo. Reducedexcretion of NO metabolites correlated with the proximal tubuleinjury marker glutathione S-transferase A1-1 in urine, whichdecreased by 70 (50–80)% in AP-treated patients compared withan increase by 200 (45–525)% in placebo-treated patients.

Conclusions: In severe sepsis and septic shock, infusion of APinhibits the upregulation of renal inducible NO synthase, leadingto subsequent reduced NO metabolite production, and attenuatedtubular enzymuria. This mechanism may account for the observedimprovement in renal function. (Crit Care Med 2009; 37:417–423)

KEY WORDS: inducible nitric oxide synthase expression; acutekidney injury; inflammation-tubular enzymuria; glutathione S-trans-ferase; creatinine

417Crit Care Med 2009 Vol. 37, No. 2

tient or their legal representative. This de-scribes the results of AP on plasma creatinineand renal replacement therapy (RRT) fre-quency of all recruited (n � 36) patients in aprospective phase IIa study, and a subgroupanalysis on the effects of AP on renal damageand function in 15 patients included at theRadboud University Nijmegen Medical Centre.

Patients admitted to the intensive care unitwith sepsis (proven or suspected Gram-negative bacterial infection and at least twoSystemic Inflammatory Response Syndromecriteria existing for less than 24 hours), whofulfilled the inclusion criteria, were recruitedin a double-blind, randomized (2:1), placebo-controlled design. Patients were eligible forinclusion if aged between 18 and 80 years andif acute onset of end-organ dysfunction, in-cluding sustained hypotension requiring vaso-pressor therapy, was present for not longerthan 12 hours (15). Exclusion criteria wereprior therapy with AP, a known allergy to cowmilk, participation in another investigationalstudy within 90 days before the start of thisstudy, expected death within 24 hours, chronicrenal failure requiring hemodialysis or perito-neal dialysis, acute pancreatitis with no estab-lished source of infection, human immunodefi-ciency virus infection, pregnancy or lactation,confirmed Gram-positive or fungal sepsis, andtreatment with immunosuppressants includinghigh doses of glucocorticosteroids (equivalentto prednisone �1 mg/kg/day). Throughout thestudy, all patients received standard conven-tional therapy. During the study period, non-steroidal anti-inflammatory drugs, aminoglyco-sides activated protein C, and diuretics were notadministered, and no radiologic examinationswith potentially nephrotoxic contrast agentswere performed.

Clinical parameters, including body tem-perature, blood pressure, heart rate, and re-spiratory rate were determined before, during,and after the intervention. Both Sepsis-Related Organ Failure Assessment and AcutePhysiology and Chronic Health Evaluation II(16) were recorded at inclusion of the patients.Serum creatinine levels were determined bythe Jaffe method at different time points dur-ing the 28-day follow-up period. AKI was de-fined as a serum creatinine �150 �mol/L orpatient already on RRT at baseline. Becausethe sample size was too small, more gradualdistribution (e.g., Risk, Injury, and FaiLurE cri-teria) was not possible. The general dialysis in-tervention policy was one of the following:oliguria�serum urea �20 mmol/L oranuria�serum urea �15 mmol/L, or serum po-tassium �5.5 mmol/L despite pharmacologicmeasures, or serum pH �7.25, or serum phos-phate �3.5 mmol/L, or treatment-resistant fluidoverload, but differences between study centersand physicians cannot be excluded.

In the subgroup analysis, arterial bloodand urine from an indwelling catheter werecollected at several time points between 0 and24 hours after inclusion. Urine volumes wererecorded and samples for the determination of

NO metabolites were frozen at �80°C untilassayed as described (5). The collected urinesamples for the determination of glutathioneS-transferase A1-1 (GSTA1-1) and GSTP1-1were supplemented with 10% (v/v) of a storagebuffer, to ascertain a stable pH and to preventbacterial growth, and frozen at �20°C (17).C-reactive protein, leukocyte counts, sodium,and protein were determined by routine clin-ical chemistry.

Intervention. AP of bovine origin (AM-Pharma, Bunnik, The Netherlands) was de-rived from the intestinal mucosa of calves �6month. Eligible patients received either AP ormatching placebo (2.5 mM Tris-HCL, 2.5 mMmagnesium chloride, 0.05 mM zinc chloride,pH 7.3, with 40% glycerol as stabilizer) intra-venously for 24 hours in a 2:1 ratio (Table 1).AP treatment consisted of initial bolus intra-venous injection (67.5 U/kg body weight) over10 minutes, followed by continuous infusion(132.5 U/kg) over the following 23 hours and50 minutes. AP dose was based on the phar-macokinetics of AP in piglets (10), and previ-ous safety data in healthy volunteers given APalone and in combination with a 4-ng/kg bodyweight LPS dose (see demographic character-istics in Supplementary Table A available on-line; Vd � 0.68 � 0.20 L/kg, T1/2 � 2.6 � 0.7hours, CL � 0.19 � 0.06 L/hr/kg unpublisheddata, AM-Pharma).

Study medication was packaged accordingto a randomization list in blocks of three,labeled with subject number and additionaladministration information. The responsiblepharmacist calculated the exact volume of APor placebo to be administered to each patientbased on body weight and treatment code. Themaster randomisation list was held by SGSBiopharma (Wavre, Belgium). Individualsealed envelopes with treatment code for eachpatient were filed in a secure location at eachhospital and at SGS Biopharma.

Chemical Assays. The total amount of thestable NO metabolites (NOx), nitrate and ni-trite, were determined as a measure of theproduction of NO radicals, using the Griessreaction, according to Moshage et al (18).Heparinized plasma and urine samples werediluted 4- and 40-fold with distilled water,respectively. The amounts of GSTA1-1 andGSTP1-1 in urine were determined to differ-entiate between proximal and distal tubularcell injury (17) and were assayed in triplicateby enzyme-linked immunosorbent assay as de-scribed previously (19, 20).

Determination of iNOS mRNA Expression.Quantitative real-time polymerase chain reac-tion was used to determine the levels of iNOSmessenger RNA (mRNA) in cell pellets thatwere isolated from urine samples at baselineand at more than three separate time points inthe first 24 hours after start of the infusion ofAP or placebo. All experiments were per-formed in triplicate. Sample quantities werenormalized to the expression of the house-keeping gene glyceraldehyde 3-phosphate de-hydrogenase. The relative expression of iNOS in

control healthy volunteers (n � 4, data not shown)was normalized for the mean cycle threshold (CT)value of glyceraldehyde 3-phosphate dehydroge-nase (CT � 23.6 � 0.3, delta CT � 12.1 � 0.1),and set to 1 as described earlier (5).

Statistical Analysis. Values are given asmean � SE or as median (range 25% to 75%range), as appropriate. The projected samplesize of 15 patients with urinary marker dataensured at least 80% power to detect a rele-vant assumed difference in the end point pa-rameter of at least 100 �g/10 mmol creatininein cumulative urinary GSTA1-1 excretion (5)at 24 hours between placebo- and AP-treatedpatients (1:2 ratio) with a two-sided p value�0.05. For transparency, the sample size ofthe total study group was based on the ac-quirement of safety and kinetics data in whichrenal function was one of the parameters.Fisher’s exact test was used for mortality andpatients on RRT, and Table 1. Differences be-tween experimental groups were tested byanalysis of variances for repeated measures.For iNOS mRNA expression and Table 1, aMann-Whitney U test was used.

RESULTS

Patients and Outcomes of the TotalGroup. In total, 36 patients (20 men/16women, mean age 58 � 3 years) wererecruited. Patients in the two treatmentgroups had similar demographic andclinical characteristics (Table 1). Drug-related serious adverse events were notobserved. The 28-day overall mortality af-ter inclusion in the AP-treated group was24%, compared with 36% in the placebo-treated group (p � 0.45). Renal failureoccurred in 16 of the 36 patients and themortality rate in patients with AKI (base-line serum creatinine �150 �mol/L orpatient already on RRT at baseline)tended to be lower in the AP group rela-tive to the placebo group (AP � 27%;placebo � 60%; p � 0.21).

Four patients on placebo (36%) andsix on AP (24%) required RRT during the28-day follow-up period (p � 0.45). Inpatients with AKI, serum creatininetended to decrease over the 2 days afterstart of the AP intervention (p � 0.12),whereas in placebo-treated patients anonsignificant increase was observed(Fig. 1B; p � 0.49).

There was no difference in medianmean arterial pressure at baseline (Table1) or during the first 24 hours betweenpatients treated with placebo or AP (datanot shown). Vasopressor therapy wasneeded in 80% of the patients in bothtreatment groups, and there was no dif-ference in the percentage of patients re-ceiving catecholamines before study

418 Crit Care Med 2009 Vol. 37, No. 2

Table 1. Patient recruitment and baseline characteristics

Alkaline Phosphatase (n � 25) Placebo (n � 11) p

All patients (n � 36)Demographics

M/F 15/10 5/6 0.48Age (yrs) 59 � 3 56 � 5 0.71Body mass index (kg/m2) 26 � 1 27 � 1 0.13Acute Physiology and Chronic Health Evaluation II 19 � 2 19 � 2 0.97Sequential Organ Failure Assessment 9 � 1 10 � 1 0.80

Systemic inflammatory response syndromeTemperature (°C)

�36.5 36.2 (36.0–6.4) 36.2 (35.6–36.4) 0.85�38.5 38.8 (38.7–39.2) 39.3 (39.3–39.4) 0.21

Leukocytes (�109/L)�4 3.2 (2.3–3.4) 3.2 (2.3–3.3) 1.00�12 17.3 (14.3–22.6) 17.5 (16.1–23.5) 0.79

Heart rate (bpm) 111 (93–130) 107 (94–136) 0.90Mean arterial pressure (mm Hg) 71 (64–77) 78 (72–83) 0.45Patients on ventilator (%) 23 (92) 10 (91) 1.00

Alkaline Phosphatase (n � 10) Placebo (n � 5) p

Patients in renal markers analysis (n � 15)Demographics

M/F 7/3 2/3 0.33Age (yrs) 55 � 6 55 � 9 0.73Body mass index (kg/m2) 25 � 2 27 � 1 0.60Acute Physiology and Chronic Health Evaluation II 18 � 2 20 � 2 0.44Sequential Organ Failure Assessment 9 � 1 8 � 2 0.61

Systemic inflammatory response syndromeTemperature (°C)

�36.5 36.4 (36.0–36.4) 36.4 (36.4–36.4) 0.86�38.5 39.7 (39.6–39.7) 39.2 (39.1–39.4) 1.00

Leukocytes (�109/L)�4 2.9 (2.1–3.7) — —�12 20.6 (17.4–22.0) 19.3 (16.8–21.9) 1.00

Heart rate (bpm) 95 (90–120) 104 (92–110) 0.83Mean arterial pressure (mm Hg) 69 (65–79) 77 (77–81) 0.83Norepinephrine infusion rate (�g/min/kg) 0.10 (0.04–0.32) 0.21 (0.17–0.27) 0.33Respiratory rate (breaths/min) 19 (16–23) 17 (16–24) 0.48Patients on ventilator 10a 4b 0.33

RRT, renal replacement therapy; iNOS, inducible NO synthase; NO, nitric oxide; M, male; F, female.Data are expressed as mean � SE or as median (range 25% to 75%). Demographics were assessed at inclusion of the patients and systemic inflammatory

response syndrome criteria at the start of the intervention (�24 hrs).aPatients were ventilated with a FIO2 of 57% � 8% by pressure control (n � 5), pressure support (n � 4), or noninvasive positive pressure ventilation

(n � 1), Ppeak 30 � 3 cm H2O, positive end-expiratory pressure 14 � 1 cm H2O, leading to tidal volumes of 8.2 � 2.0 mL/kg ideal body weight for femalesand 7.5 � 0.8 mL/kg ideal body weight for males.

bPatients were ventilated with a FIO2 of 53% � 5% by pressure control (n � 4), Ppeak 26 � 2 cm H2O, positive end-expiratory pressure 9 � 1 cm H2O,leading to tidal volumes of 8.5 � 0.2 mL/kg ideal body weight for females and 8.3 � 1.1 mL/kg ideal body weight for males. No significant differencesbetween groups.

419Crit Care Med 2009 Vol. 37, No. 2

treatment and 24 hours afterward be-tween both groups (data not shown). An-tibiotic treatment included broad-spec-trum penicillins (95%), cephalosporins(45%), fluoroquinolones (40%), and oth-ers (20%), and the use of these antibioticswas evenly distributed between bothgroups. Of the total group of patients,30% remained culture negative. Of thepatients with positive blood cultures,70% had Gram-negative infections (71%placebo vs. 72% AP) and 30% had Gram-positive infections (29% placebo vs. 28%AP).

Patients and Outcomes of the Sub-group Analysis. To gain insight into thepathophysiology of NO-mediated renaldamage during systemic inflammation, asubgroup of 15 patients (AP, n � 10;placebo, n � 5), consisting of seven pa-tients with severe sepsis and eight pa-tients with septic shock, was investigatedin more detail at one of the hospitals(Radboud University Nijmegen MedicalCentre, The Netherlands). Patients in thissubgroup had similar demographic andclinical characteristics compared withthe total group (Table 1).

AP Attenuates Renal iNOS Induction.The expression of iNOS in cells isolatedfrom urine samples, the origin of whichwe have recently demonstrated to be pre-dominantly renal proximal tubule (5),was 42-fold induced in sepsis patientscompared with controls (healthy sub-jects, data not shown). AP administrationreduced this induction by 80% � 5%during the first 24 hours after start of theintervention (Fig. 2A; p � 0.05). In con-trast, placebo-treated patients had a fur-ther increase in iNOS levels during thesame period (840% � 80%; p � 0.05),compared with baseline levels. There wasno difference in baseline iNOS expressionbetween AP- and placebo-treated patientsand no correlation between norepineph-rine infusion rate and iNOS expressionwas found (data not shown).

During the first 24 hours after start ofthe intervention, NOx levels in blood werenot significantly different between AP andplacebo-treated patients (Fig. 2B). Theurinary excretion of NOx did not differsignificantly between the AP group andthe placebo group at baseline. However,the urinary excretion of NOx decreased

from 230 (165–530) at baseline to 41(28–84) �mol/10 mmol creatinine (p �0.05) after 24 hours AP administration,whereas after placebo it increased from81 (64–419) to 630 (65–1480) �mol/10mmol creatinine (p � 0.05). In addition,the cumulative urinary NOx excretionwas significantly lower in the AP-treatedpatients compared with the placebogroup (Fig. 2C; p � 0.05).

AP Attenuates Kidney Damage. Al-though all patients in this subgroupshowed some degree of AKI with mildproteinuria (Table 2), none of these pa-tients required RRT during the 28-dayfollow-up period. In the first 24 hours,the creatinine clearance improved by45% (30%–180%) in patients on AP ther-apy (p � 0.05) and there was a trendtoward deterioration by 25% (15%–35%)on placebo (Table 2; p � 0.13). Further-more, sepsis patients with a fractionalsodium excretion of more than 2% exhib-ited an improvement in renal functionafter AP (p � 0.05), whereas patients witha fractional sodium excretion of less than1% showed no improvement (p � 0.36),suggesting a protective role of AP in im-

Figure 1. Follow-up serum creatinine in all patients (A), in patients with (B) and without (C) acute kidney injury and subgroup patients renal markers (D).Creatinine levels in serum were measured at various times for placebo (, n � 5–11) and alkaline phosphatase (AP)-treated patients (f, n � 10–25). Acutekidney injury is defined by baseline serum creatinine �150 �mol/L or already on renal replacement therapy at entry. No renal replacement therapy wasused during these measurements in the subgroup. Data are expressed as median with 25% range for AP and 75% range for placebo and analyzed by analysisof variance with repeated measures over the complete curve (##; significantly different compared with baseline levels; p � 0.01).

420 Crit Care Med 2009 Vol. 37, No. 2

minent tubular damage (Table 2). In thesubgroup, the median plasma creatininelevels declined significantly in AP-treatedpatients from 91 (73–138) �mol/L atbaseline to 70 (60–92) �mol/L 7 daysafter the intervention (Fig. 1D; p � 0.01).No significant change in placebo-treatedpatients was observed (114 [99 –117]�mol/L at baseline vs. 106 [73–141]�mol/L at 7 days, p � 0.48).

The urinary excretion of bothGSTA1-1 and GSTP1-1 was elevated in all

patients, indicating both proximal anddistal tubule damage, respectively. Dur-ing the first 24 hours, the amount ofGSTA1-1 in urine of AP-treated patientsdecreased from 33 (12–131) to 6.5 (5.4–16) �g/10 mmol creatinine (p � 0.05),compared with an increase from 27 (15–33) to 39 (33–206) �g/10 mmol creati-nine in the placebo group, p � 0.05.Cumulative urinary GSTA1-1A excretionwas also significantly lower in AP-treatedpatients, compared with the placebo

group (Fig. 3A; p � 0.05). Furthermore,there was a trend toward an decrease inurinary GSTP1-1 excretion (from 23 [14–41] at baseline to 12 [8–83] �g/10 mmolcreatinine after 24 hours, p � 0.07) inAP-treated patients, which was absent in

Figure 2. Inducible nitric oxide (NO) synthase (iNOS) induction. (A) iNOS messenger RNA expressionis illustrated for placebo-treated (open bars, n � 4) and alkaline phosphatase (AP)-treated (closed bars,n � 8) sepsis patients. The relative expression of iNOS mRNA in control healthy volunteers (data notshown) was normalized for the mean cycle threshold (CT) value of the housekeeping gene GAPDH (CT �23.6 � 0.3, delta CT � 12.1 � 0.1), and set to 1. (B) NO metabolites in blood were not significantly differentbetween placebo (open bars, n � 5) and AP-treated (closed bars; n � 10) patients. (C) NOx levels in urinewere measured at various times after the intervention in placebo (, n � 5) and AP-treated patients (f; n �10). The urinary excretion of NOx was corrected for creatinine excretion and analyzed by analysis ofvariance with repeated measures over the complete curve. (A) Data are expressed as mean � SE and (B�C)as median with 25% range for AP and 75% range for placebo. (#; significantly different compared withbaseline levels; p � 0.05 and *; significantly different compared with the placebo group, p � 0.05).

Table 2. Renal function parameters for patients in the subgroup analysis (n � 15)

Kidney Function Time AP (n � 10) Placebo (n � 5)

Total urine volume (mL) 0–24 hrs 1880 (940–2230) 1470 (1120–2780)Protein excretion (mg/day) 0–24 hrs 454 (323–533) 447 (414–769)Creatinine clearance (mL/min) Baseline 54 (24–84) 80 (77–91)

24 hrs 76 (25–101)a 59 (45–59)Fractional sodium excretion (%)

�2% Baseline 4.5 (3.3–5.7) 4.1 (3.8–4.3)24 hrs 1.2 (1.1–1.3)a 5.2 (4.4–6.1)

�1% Baseline 0.5 (0.1–0.5) 0.2 (0.1–0.2)24 hrs 0.1 (0.1–0.4) 0.1 (0.1–0.2)

AP, alkaline phosphatase.Data are expressed as median (range 25% to 75%).aP � 0.05, significantly different compared with the placebo group.

Figure 3. Urinary excretion of markers of renalinjury. Glutathione S-transferase (GSTA1–1) (A)and GSTP1-1 (B) levels in urine were measured atvarious times after the intervention in placebo(, n � 5) and alkaline phosphatase-treated pa-tients (f, n � 10). The urinary excretion of GSTswere corrected for creatinine excretion and ana-lyzed by analysis of variance with repeated mea-sures over the complete curve. Data are expressedas median with 25% range for alkaline phospha-tase and 75% range for placebo. (*; significantlydifferent compared with the placebo group, p �0.05). (C) Correlation between the cumulative uri-nary excretion of NOx and GSTA1-1, 24 hours afterinclusion (rs � 0.5, p � 0.01, n � 14). One patientis missing, because he died within 24 hours afterthe start of the intervention.

421Crit Care Med 2009 Vol. 37, No. 2

the placebo group. However, there wereno differences in the cumulative urinaryGSTP1-1 excretion between the AP andplacebo-treated groups (p � 0.23) andduring the first 24 hours of the treatmentin patients who received placebo (Fig. 3B;p � 0.29). Importantly, the cumulativeurinary excretion of NOx correlated withGSTA1-1 excretion during the first 24hours (rs � 0.5, p � 0.01; Fig. 3C).

DISCUSSION

This is the first controlled exploratorystudy with AP in sepsis patients with de-tailed monitoring of renal function. Ourresults show that infusion of this enzymeattenuates iNOS induction and subse-quent renal NO metabolite productionand prevents further renal damage in pa-tients with severe sepsis or septic shock.Furthermore, the present data are thefirst to show that a pharmacologic inter-vention is able to reduce the expression ofiNOS in septic humans in vivo. To deter-mine the effect of the attenuated NO me-tabolite production on renal damage, weexamined the urinary excretion of GSTsas tubular injury markers and found thatAP decreased iNOS activity and preservedthe integrity of the proximal tubule andthat the renal release of NO metabolitescorrelated with the tubular enzymuria.These findings indicate that septic AKI ismediated through the NO pathway. Al-though we have monitored carefully thekey clinical parameters, outcomes andkey marker (serum creatinine), to framethe research-oriented renal markerswithin usual management, we appreciatethat the detailed and comprehensive na-ture of the investigation of the subgroupmay not be feasible in a large-scale inter-vention study. Nevertheless, despite thesmall size and heterogeneity of the stud-ied population, we were able to demon-strate beneficial effects of AP and ourfindings are in keeping with animal datapreviously reported (9–13). These benefi-cial effects of AP could be explained by itsdual mode of action. First, it binds to and,subsequently, dephosphorylates LPS.Second, the enzymatic reaction product,monophosphoryl-LPS, is a nontoxic sub-stance for mammals which acts as a par-tial antagonist on the LPS receptor com-plex (8, 21). However, we cannot excludethat other mechanisms, independent ofLPS detoxification, could have played arole in the protective effect of AP in thekidney.

In sepsis patients, we demonstrate anincreased iNOS mRNA expression in cellsisolated from urine. The fact that iNOS isup-regulated in a remote organ not pri-marily involved in the infection is a novelfinding in humans. In a previous report,Annane et al (22) found an up-regulationof iNOS in patients with septic shockcaused by cellulitis only at the site ofinfection, suggesting a compartmental-ized induction of iNOS. The absence ofiNOS expression in skin, muscle, fat, andartery samples that were not involved inthe primary infection may be explainedby the fact that these tissues do not ex-press iNOS under physiologic circum-stances. In contrast, iNOS is constitu-tively expressed in the kidney (4), whichmay facilitate a rapid and more pro-nounced response during systemic in-flammation. Treatment with AP resultedin an attenuation of the increased expres-sion of iNOS mRNA, and in a reducedurinary excretion of NO metabolites. Theattenuated iNOS expression is likely to becaused by AP inhibitory effects on LPSand LPS-mediated inflammatory cytokineresponse, as there are no reports indicat-ing that AP exerts a direct effect on iNOSexpression. Therefore, pharmacologic in-terventions that interfere with mediatorsin the sepsis cascade through NO couldprotect the kidneys during sepsis.

Endogenous AP activity in the kidneyhas been found along the proximal tubu-lar brush border membrane (9), whichmay be inactivated by the reactive oxygenspecies and peroxynitrite generated dur-ing sepsis (11). Therefore, we hypothe-sized that administration of exogenousAP during sepsis may attenuate septicAKI through detoxification of LPS. In-deed, we found an improvement in kid-ney function and significantly attenuatedincrease in urinary GSTA1-1 excretion inpatients treated with AP, indicating prox-imal tubule protection. Interestingly, re-nal iNOS is predominantly expressed inthe proximal tubule (4), where excessiveNO release induces direct toxic effects(23). We report here that the cumulativeamount of urinary excreted NO metabo-lites correlates with GSTA1-1 excretion inurine, suggesting that increased renal NOmetabolite production accounts for theproximal tubule damage. Thus, treatmentwith AP may result in a reduced induc-tion of renal iNOS expression, whichleads to a diminished production of NOmetabolites and less proximal tubulardamage.

We realize that this study is under-powered to demonstrate beneficial long-term effects on hard end points, such asRRT frequency, length of stay in the in-tensive care unit, or mortality. However,our preliminary results suggest that theclinical benefit by AP in patients withrapidly progressive sepsis is possibly de-rived from its potential protective effectson renal tissue. Therefore, these resultswarrant confirmation in a larger trial thatincludes only sepsis-induced AKI pa-tients. Pharmacologic interventions,other than AP, aimed to inhibit LPS ormediators early in the septic cascade haveoften exerted positive effects in sepsismodels in animals (24), but efforts totranslate this to the treatment of sepsispatients have yielded disappointing re-sults (25–28). Among several explana-tions (24), timing of the pharmacologicintervention seems to be of major impor-tance. Apart from one animal study inwhich the AP was administrated 2 hoursafter the induction of fecal peritonitis(13), our data show a beneficial effectafter AP treatment in patients with Sys-temic Inflammatory Response Syndromecriteria existing less than 24 hours.

In conclusion, in severe sepsis andseptic shock, infusion of AP inhibits theupregulation of renal iNOS, leading tosubsequent reduced NO metabolite pro-duction, and attenuated tubular en-zymuria. This mechanism may accountfor the observed improvement in renalfunction. Because morbidity, mortality,and management costs are very high insepsis patients with AKI, this findingforms the basis for a large clinical trial.

ACKNOWLEDGMENTS

We are grateful to the nursing andmedical staff of all the intensive careunits for their help with the patients anddata collection.

APSEP Study Group (Principal Inves-tigators in participating hospitals, alpha-betically): J. Bakker, MD, PhD, Depart-ment of Intensive Care, Erasmus MedicalCentre, Rotterdam; P. Jorens, MD, PhD,Department of Intensive Care, UniversityHospital Antwerp; J. Meulenbelt, MD,PhD, Division of Intensive Care Centre,University Medical Centre Utrecht; P. Ro-giers, MD, PhD, Department of IntensiveCare, Hospital Network Antwerp; F.Snellen, MD, Department of Anesthesiol-ogy, Isala Klinieken, Zwolle; H. Spapen,MD, PhD, Department of Intensive Care,University Hospital, Vrije Universiteit

422 Crit Care Med 2009 Vol. 37, No. 2

Brussels; J. E. Tulleken, MD, PhD, De-partment of Internal Medicine, IntensiveCare and Respiratory Care Unit, Univer-sity Medical Centre Groningen; J. G. vander Hoeven, MD, PhD, Department of In-tensive Care, Radboud University Nijme-gen Medical Centre.

REFERENCES

1. Alberti C, Brun-Buisson C, Goodman SV, etal: Influence of systemic inflammatory re-sponse syndrome and sepsis on outcome ofcritically ill infected patients. Am J RespirCrit Care Med 2003; 168:77–84

2. Schrier RW, Wang W: Acute renal failure andsepsis. N Engl J Med 2004; 351:159–169

3. Yegenaga I, Hoste E, Van BW, et al: Clinicalcharacteristics of patients developing ARFdue to sepsis/systemic inflammatory re-sponse syndrome: Results of a prospectivestudy. Am J Kidney Dis 2004; 43:817–824

4. Ortiz PA, Garvin JL: Cardiovascular and renalcontrol in NOS-deficient mouse models.Am J Physiol Regul Integr Comp Physiol2003; 284:R628–R638

5. Heemskerk S, Pickkers P, Bouw MP, et al:Up-regulation of renal iNOS during humanendotoxemia and sepsis is associated withproximal tubule injury. Clin J Am Soc Neph-rol 2006; 1:853–862

6. Poelstra K, Bakker WW, Klok PA, et al: De-phosphorylation of endotoxin by alkalinephosphatase in vivo. Am J Pathol 1997; 151:1163–1169

7. Harris H: The human alkaline phosphatases:What we know and what we don’t know. ClinChim Acta 1990; 186:133–150

8. Poelstra K, Bakker WW, Klok PA, et al: Aphysiologic function for alkaline phospha-tase: Endotoxin detoxification. Lab Invest1997; 76:319–327

9. Bentala H, Verweij WR, Huizinga-Van derVlag A, et al: Removal of phosphate from lipidA as a strategy to detoxify lipopolysaccharide.Shock 2002; 18:561–566

10. Beumer C, Wulferink M, Raaben W, et al: Calfintestinal alkaline phosphatase, a novel ther-apeutic drug for lipopolysaccharide (LPS)-mediated diseases, attenuates LPS toxicity inmice and piglets. J Pharmacol Exp Ther2003; 307:737–744

11. Koyama I, Matsunaga T, Harada T, et al:Alkaline phosphatases reduce toxicity of lipo-polysaccharides in vivo and in vitro throughdephosphorylation. Clin Biochem 2002; 35:455–461

12. van Veen SQ, van Vliet AK, Wulferink M, etal: Bovine intestinal alkaline phosphatase at-tenuates the inflammatory response in sec-ondary peritonitis in mice. Infect Immun2005; 73:4309–4314

13. Su F, Brands R, Wang Z, et al: Beneficialeffects of alkaline phosphatase in septicshock. Crit Care Med 2006; 34:2182–2187

14. Kapojos JJ, Poelstra K, Borghuis T, et al:Induction of glomerular alkaline phospha-tase after challenge with lipopolysaccharide.Int J Exp Pathol 2003; 84:135–144

15. Bone RC, Balk RA, Cerra FB, et al: Defini-tions for sepsis and organ failure and guide-lines for the use of innovative therapies insepsis. The ACCP/SCCM Consensus Confer-ence Committee. American College of ChestPhysicians/Society of Critical Care Medicine.Chest 1992; 101:1644–1655

16. Knaus WA, Draper EA, Wagner DP, et al:APACHE II: A severity of disease classifica-tion system. Crit Care Med 1985; 13:818–829

17. Branten AJ, Mulder TP, Peters WH, et al:Urinary excretion of glutathione S trans-ferases alpha and pi in patients with protein-uria: Reflection of the site of tubular injury.Nephron 2000; 85:120–126

18. Moshage H, Kok B, Huizenga JR, et al: Ni-trite and nitrate determinations in plasma: Acritical evaluation. Clin Chem 1995; 41:892–896

19. Mulder TP, Peters WH, Court DA, et al: Sand-wich ELISA for glutathione S-transferase al-pha 1-1: Plasma concentrations in controls

and in patients with gastrointestinal disor-ders. Clin Chem 1996; 42:416–419

20. Mulder TP, Peters WH, Wobbes T, et al: Mea-surement of glutathione S-transferase P1-1 inplasma: Pitfalls and significance of screeningand follow-up of patients with gastrointestinalcarcinoma. Cancer 1997; 80:873–880

21. Gustafson GL, Rhodes MJ: A rationale for theprophylactic use of monophosphoryl lipid Ain sepsis and septic shock. Biochem BiophysRes Commun 1992; 182:269–275

22. Annane D, Sanquer S, Sebille V, et al: Com-partmentalised inducible nitric-oxide syn-thase activity in septic shock. Lancet 2000;355:1143–1148

23. Yu L, Gengaro PE, Niederberger M, et al:Nitric oxide: A mediator in rat tubular hyp-oxia/reoxygenation injury. Proc Natl AcadSci U S A 1994; 91:1691–1695

24. Natanson C, Hoffman WD, Suffredini AF, etal: Selected treatment strategies for septicshock based on proposed mechanisms ofpathogenesis. Ann Intern Med 1994; 120:771–783

25. Abraham E, Wunderink R, Silverman H, etal: Efficacy and safety of monoclonal anti-body to human tumor necrosis factor alphain patients with sepsis syndrome. A random-ized, controlled, double-blind, multicenterclinical trial. TNF-alpha MAb Sepsis StudyGroup. JAMA 1995; 273:934–941

26. Angus DC, Birmingham MC, Balk RA, et al:E5 murine monoclonal antiendotoxin anti-body in gram-negative sepsis: A randomizedcontrolled trial. E5 Study Investigators.JAMA 2000; 283:1723–1730

27. Petros A, Lamb G, Leone A, et al: Effects of anitric oxide synthase inhibitor in humanswith septic shock. Cardiovasc Res 1994; 28:34–39

28. Ziegler EJ, Fisher CJ Jr, Sprung CL, et al:Treatment of gram-negative bacteremia andseptic shock with HA-1A human monoclonalantibody against endotoxin. A randomized,double-blind, placebo-controlled trial. TheHA-1A Sepsis Study Group. N Engl J Med1991; 324:429–436

423Crit Care Med 2009 Vol. 37, No. 2