Tumor necrosis factor-alpha during continuous high-flux hemodialysis in sepsis with acute renal failure

Suppressed ex vivo endotoxin (ET)-induced production of the proinflammatory cytokine, tumor necrosis factor-alpha (TNF-alpha), in isolated mononuclear cells (PBMCs) is associated with fatal outcome in severe sepsis. PBMCs from surviving patients, but not those from nonsurviving patients, recover their capacity to produce normal amounts of TNF-alpha. We tested the influence of two modalities of continuous renal replacement therapy (CRRT) on ex vivo-induced whole-blood production of TNF-alpha and inhibitory TNF-soluble receptor type I (TNFsRI) in 12 patients with acute renal failure and sepsis (APACHE II score 22 to 30).

METHODS

Standard continuous venovenous hemofiltration (CVVH; 36 liters of bicarbonate substitution fluid per day) was performed in 7 patients using polyamid hemofilters (FH66; Gambro). In an additional five patients, we performed daily 18 hours of high-flux hemodialysis (CHFD) using polysulfon F60S dialyzers (Fresenius) and 75 liters of bicarbonate dialysate using the GENIUS single-pass batch dialysis system. Samples were separated from the blood circuit as well as from the ultrafiltrate/spent dialysate lines at the start, during, and end of treatment. Whole-blood samples were incubated with 1 ng/ml of ET for three hours at 37 degrees C. Ultrafiltrate or dialysate samples were incubated with donor whole blood in the presence of ET to measure suppressing activity in ultrafiltrate and spent dialysate.

RESULTS

At the start of CRRT, ET-induced whole-blood TNF-alpha production was suppressed to approximately 10% of that in normal controls. During CVVH, median ET-induced TNF-alpha production increased from 0.35 ng/ml at the start to 1.2 ng/ml at three hours, but decreased to pre-CVVH levels at the end of a 24-hour period. In contrast, in patients on CHFD, the median ET-induced TNF-alpha production was 0.5 ng/ml at the start, 1.1 ng/ml at 3 hours, 1.6 ng/ml at six hours, and 1.5 ng/ml at the end of 18 hours of treatment. The ultrafiltrate obtained after three hours of CVVH did not contain suppressing activity. In CHFD, the spent dialysate as compared with fresh dialysate suppressed ET-induced TNF-alpha production in donor blood by 33% throughout the 18 hours of treatment. Whole-blood production of TNFsRI did not change significantly at any time point during CVVH or CHFD.

CONCLUSION

These data suggest that high-volume CHFD is superior to standard CVVH in removing a suppressing factor of proinflammatory cytokine production. As CVVH only transiently improves TNF-alpha production, it is most likely that the putative suppressing factor is removed because of saturable membrane adsorption in CVVH. In CHFD, there is a combination of adsorption and detectable diffusion into the dialysate. It remains to be shown whether a further increase in the volume of dialysate per day is able to not only improve but normalize the cytokine response and improve outcome in septic patients with acute renal failure.

Use of adsorptive mechanisms in continuous renal replacement therapies in the critically ill

The pathophysiology of sepsis is becoming a more complicated scenario. In sepsis, endotoxin or other gram-positive derived products induce a complex and dynamic cellular response giving rise to several mediators known to be relevant in the pathogenesis of septic shock, such as specific mediators. substances responsible for up- or down-regulation of cytokine receptors and cytokine antagonists, inactivators of nuclear factor-kappaB or signal transduction pathways, and precursor molecules. In this article, we delve into some new concepts stemming from the use of sorbents in continuous plasma filtration. The rationale is based on the assumption that the nonspecific removal of several mediators of the inflammatory cascade and cytokine network may improve outcome in a rabbit model of septic shock and hemodynamics in a pilot clinical study. The importance of looking for innovative treatments specifically targeted for the special needs of the critically ill patients rather than using concepts and technology applied to the treatment of chronic renal failure is underlined.

Plasma C-reactive protein is linked to backfiltration associated interleukin-6 production

Bacterial contamination of dialysate may enhance cytokine production in hemodialysis. The authors tested the hypothesis that C-reactive protein and interleukin-6 (IL-6) may be linked in a large group of patients exposed to backfiltration of dialysate over a long period of observation. Plasmas stored in a recently published multicenter study were reevaluated. Plasma C-reactive protein and IL-6 concentrations in patients with chronic uremia undergoing hemodiafiltration, which is known to be associated with backfiltration (Group II, 12 patients), were compared with those found in patients treated with a modified hemodiafiltration modality without backfiltration (Group I, 16 patients), and in patients shifted from one modality to the other (Group III, 27 patients), and in 10 patients on hemodialysis (Group IV) in a 1 year multicenter study. Plasma C-reactive and IL-6 both increased significantly (p < 0.002), but slowly (after 8 months) in Group II compared with I, and during the 4 month period in hemodiafiltration with backfiltration in Group III. Backfiltration of dialysate with a moderate to low degree of contamination may enhance synthesis of cytokine and C-reactive protein in the long term. Thus, the relevance for dialytic strategies aiming at improving dialysate quality or at reducing backfiltration is highlighted.

Removal of cytokines and activated complement components in an experimental model of continuous plasma filtration coupled with sorbent adsorption

BACKGROUND

Sepsis is associated with enhanced cytokine production. Here, we examined the in vitro removal of plasma cytokines during continuous plasmafiltration coupled with sorbent adsorption.

METHODS

Proinflammatory (tumour necrosis factor-alpha, interleukins-1, -8) and anti-inflammatory (interleukin 1 receptor antagonist, soluble tumour necrosis factor receptor type I and II) cytokines in whole blood spiked with Escherichia coli endotoxin were determined during 2-h recirculation in the ultrafiltrate (condition A), plasma filtrate (condition B), before and after different sorbents (of the Amberlite-, Amberchrome- Ambersorb -type and charcoal). We studied the maximal adsorbing capacity, the 1% leakage test for cytokines and C3a des Arg and the adsorption of complement-dependent leukocyte chemiluminescence. Plasma proteins eluted from the resins were examined by sodium dodecyl sulphate polyacrylamide gel electrophoresis and immunoblotting with an anti-human alpha2-macroglobulin.

RESULTS

In condition B, we observed a 40- and 121-fold % increase (vs condition A) in the removed mass and clearance of tumour necrosis factor-alpha. For all other cytokines, the removed mass and the clearance increased from 2.3- up to 6-fold. The Amberchrome but not the Amberlite or Ambersorb resins could remove the highest amount of cytokines and could reduce complement-dependent chemiluminescence. Two protein bands of approximately 400,000 D and 200,000 D were eluted only from Amberchrome resins and immunoprecipitated by anti-human alpha2-macroglobulin and anti-human C3c antibodies, respectively.

CONCLUSIONS

These studies suggest an efficient removal of cytokines in continuous plasmafiltration with sorbent adsorption. The binding of alpha2-macroglobulin, a carrier of cytokines in plasma, might be a additional mechanism in the removal of cytokines from plasma.

Transfer of endogenous pyrogens across artificial membranes?

Synthetic high-flux dialyzer membranes used in continuous veno-venous hemofiltration are permeable to middle molecular size endogenous pyrogens, the pro-inflammatory cytokines IL-1 beta and TNF-alpha. The quantities removed by sieving are, however, negligible in vitro as well as in vivo. Adsorption of cytokines to the membrane polymer is the major mechanism of pyrogen removal. Adsorption seems to be semispecific for pro-inflammatory cytokines because levels of anti-inflammatory mediators were not changed or even increased during CVVH. Thus, CVVH may change cytokine profiles in septic patients supporting the predominance of anti-inflammatory over pro-inflammatory activity in plasma. It remains to be demonstrated whether modifications of extracorporeal blood purification systems (high-volume CVVH, plasma separation + adsorption) are able to amplify the change in cytokine profiles and whether this change influences outcome of septic patients.

Clinical, hematologic, and immunologic effects of interleukin-10 in humans

We conducted a double-blind, placebo-controlled study to investigate the safety, pharmacokinetics, and immunological properties of interleukin-10 (IL-10) administration in healthy humans. Volunteers received a single intravenous bolus injection of recombinant human IL-10 (1, 10, or 25 micrograms/kg) or placebo. Cytokine production in whole blood and peripheral blood mononuclear cells (PBMC) was assessed before and 3, 6, 24, and 48 hr after the injection. Peak serum concentrations of IL-10 (15 +/- 1.1, 208 +/- 20.1, and 505 +/- 22.3 ng/ml) occurred after 2-5 min for 1, 10, and 25 micrograms/kg IL-10, respectively. The terminal-phase half-life was 3.18 hr. A transient leukocytosis (24-63% above baseline) was observed 6 hr after injection, which coincided with a dose-dependent decrease (12-24%) in neutrophil superoxide generation. There was a marked inhibition (60-95%) of endotoxin-induced IL-6 production from whole blood in each group receiving IL-10. Production of IL-8 in endotoxin-stimulated blood was reduced in the 10 micrograms/kg group. In PBMC stimulated with phytohemagglutinin and phorbol ester, there was a decrease (72-87%) in interferon-gamma (IFN gamma) production 6 hr after IL-10 with a return to pre-IL-10 levels after 24 hr. This reduction was only partially associated with a decrease in the number of CD2-bearing cells. We conclude that IL-10 administration into humans is without significant side effects, and a single injection reduces ex vivo production of IL-6, IL-8, and IFN gamma.

Bacterial challenge of NISSHO ultrafilter ETF 609: results of in vitro testing

In hemodialysis, a certain degree of bacterial contamination on the dialysate side is a regular finding. Concern has been growing that this contamination may lead to a chronic inflammatory response in the patient. Ultrafiltration of dialysate can be used to reduce bacterial content and levels of cytokine-inducing substances upstream of the patient's dialyzer. The aim of this study was to test in vitro the rejection capacity of a polysulfone hollow-fiber ultrafilter (ETF 609, NISSHO Co., Osaka, Japan) challenged with bacterial filtrates derived from Pseudomonas aeruginosa PA103. Results showed a reduction of interleukin-1 beta-inducing activity (measured on peripheral blood mononuclear cells) from 5,035 +/- 394 pg/ml prefilter to nondetectable levels postfilter and endotoxin levels (limulus amebocyte lysate assay) of 4,167 +/- 1,079 versus 12 +/- 2 pg/ml, respectively. In conclusion, ultrafiltration of dialysate with the polysulfone ultrafilter ETF 609 leads to a potent reduction of cytokine-inducing activity.

Induction of interleukin-1 and interleukin-1 receptor antagonist during contaminated in-vitro dialysis with whole blood

BACKGROUND

Previous studies on the permeability of cellulosic and synthetic dialysers for bacterial-derived cytokine-inducing substances gave conflicting results. We tried to study this issue as close to the in-vivo situation as possible.

METHODS

An in-vitro dialysis circuit with whole human blood present in the blood compartment of cuprophane (Cup), polysulphone (PS), and polyamide (PA) dialysers was employed; sterile filtrates derived from Pseudomonas aeruginosa cultures were added to the dialysate. We studied the induction of interleukin-1 beta (IL-1 beta) by plasma samples taken from the blood compartment as well as the induction of IL-1 beta and interleukin-1 receptor antagonist (IL-1Ra) in mononuclear cells separated from whole blood after circulation by radioimmunoassay and polymerase chain reaction.

RESULTS

Plasma samples from the blood side of all dialysers induced IL-1 beta from non-circulated mononuclear cells after addition of pseudomonas filtrates to the dialysate; the maximal amount of IL-1 beta induced by samples from the blood compartment was 4.8 +/- 1.2 ng/ml for Cup, 1.9 +/- 0.5 ng/ml for PS, and 2.0 +/- 0.6 ng/ml for PA. Mononuclear cells separated after contaminated dialysis will all types of dialysers expressed increased mRNA levels for IL-1 beta and IL-1Ra. Production of IL-1Ra by cells separated after contaminated dialysis was determined after Cup and PS dialysis; there was increased production of IL-1Ra by these cells (Cup, 10.3 +/- 4.2; PS, 7.3 +/- 2.5 ng/ml) compared to cells separated after sterile dialysis (Cup, 5.6 +/- 2.1, P < 0.05; PS, 4.5 +/- 1.1 ng/ml, n.s.) or from non-circulated blood (Cup experiments, 4.7 +/- 1.5, P < 0.05; PS experiments, 4.1 +/- 1.2 ng/ml, n.s.).

CONCLUSIONS

These data suggest penetration of cytokine-inducing substances through both cellulosic and synthetic dialysers. Differences between dialysers may exist regarding extent and time course of penetration. The detection of cytokine mRNA as well as the measurement of IL-1Ra synthesis is a more sensitive marker for the transfer of cytokine-inducing substances through dialyser membranes than the measurement of IL-1 beta protein synthesis.

A randomized, controlled trial of IL-10 in humans. Inhibition of inflammatory cytokine production and immune responses

In vitro, IL-10 inhibits T cell proliferation and LPS-induced monocyte production of IL-1, TNF-alpha, IL-6, and IL-8. We studied the safety and immunomodulatory effects of IL-10 administration in humans. Seventeen healthy volunteers received a single i.v. bolus injection of either human IL-10 (1, 10, or 25 micrograms/kg) or placebo. Routine safety parameters, lymphocyte phenotypes, T cell proliferative responses, and stimulus-induced cytokine production were assessed before and 3, 6, 24, and 48 h after injection. There were no adverse symptoms or signs after IL-10 administration. A transient neutrophilia and monocytosis that peaked at 6 h (45-160% above base line) was observed. However, lymphocyte counts fell by 25% 3 and 6 h after the injection (p < 0.01). In particular, lymphocytes expressing the T cell surface markers CD2, CD3, CD4, CD7, and CD8 were significantly decreased. Mitogen-induced T cell proliferation was suppressed by up to 50% (p < 0.01) in the two higher dose groups. Significant dose-dependent inhibition (65-95%) of TNF-alpha and IL-1 beta production from whole blood stimulated ex vivo with endotoxin occurred after each dose of IL-10. In contrast, there was no reduction in the production of their respective antagonists, TNF soluble receptor p55 or IL-1 receptor antagonist. We conclude that a single intravenous injection of IL-10 is safe in humans, has inhibitory effects on T cells, and suppresses production of the pro-inflammatory cytokines TNF-alpha and IL-1 beta.

A randomized, controlled trial of IL-10 in humans. Inhibition of inflammatory cytokine production and immune responses

In vitro, IL-10 inhibits T cell proliferation and LPS-induced monocyte production of IL-1, TNF-alpha, IL-6, and IL-8. We studied the safety and immunomodulatory effects of IL-10 administration in humans. Seventeen healthy volunteers received a single i.v. bolus injection of either human IL-10 (1, 10, or 25 micrograms/kg) or placebo. Routine safety parameters, lymphocyte phenotypes, T cell proliferative responses, and stimulus-induced cytokine production were assessed before and 3, 6, 24, and 48 h after injection. There were no adverse symptoms or signs after IL-10 administration. A transient neutrophilia and monocytosis that peaked at 6 h (45-160% above base line) was observed. However, lymphocyte counts fell by 25% 3 and 6 h after the injection (p &lt; 0.01). In particular, lymphocytes expressing the T cell surface markers CD2, CD3, CD4, CD7, and CD8 were significantly decreased. Mitogen-induced T cell proliferation was suppressed by up to 50% (p &lt; 0.01) in the two higher dose groups. Significant dose-dependent inhibition (65-95%) of TNF-alpha and IL-1 beta production from whole blood stimulated ex vivo with endotoxin occurred after each dose of IL-10. In contrast, there was no reduction in the production of their respective antagonists, TNF soluble receptor p55 or IL-1 receptor antagonist. We conclude that a single intravenous injection of IL-10 is safe in humans, has inhibitory effects on T cells, and suppresses production of the pro-inflammatory cytokines TNF-alpha and IL-1 beta.

Impaired endotoxin-induced interleukin-1 beta secretion, not total production, of mononuclear cells from ESRD patients

Lipopolysaccharide (LPS)-induced interleukin-1 beta (IL-1 beta) and tumor necrosis factor alpha (TNF alpha) production and secretion from peripheral blood mononuclear cells (PBMC) were determined in a longitudinal study with repeated measurements in PBMC from patients with chronic uremia not on hemodialysis (N = 8), end-stage renal disease (ESRD) patients (N = 8), and healthy controls (N = 7). ESRD patients were studied while using low-flux Cuprophan dialyzers and again using high-flux AN 69 dialyzers. Total (cell-associated plus secreted) LPS-induced IL-1 beta production was enhanced in uremic patients, but similar to controls in ESRD patients on Cuprophan. In contrast, LPS-induced IL-1 beta secretion (secreted amounts in % of total production) was similar to controls in uremic patients, but significantly reduced in ESRD patients on Cuprophan (P < 0.01). During AN 69 hemodialysis, LPS-induced total IL-1 beta production remained unchanged but IL-1 beta secretion increased significantly (P < 0.05) compared to Cuprophan dialysis. Increased IL-1 beta secretion coincided with a suppression in PGE2 synthesis (P < 0.02). Similarly, blockade of endogenous PGE2 by indomethacin increased LPS-induced IL-1 beta secretion (P < 0.01) but did not enhance total IL-1 beta production in PBMC from controls and patients on Cuprophan hemodialysis. Neither total production nor secretion of TNF alpha was different comparing the three study groups. We conclude that LPS-induced IL-1 beta secretion, but not total production, is impaired in PBMC from ESRD patients on long-term Cuprophan hemodialysis. This functional change in the PBMC response is specific for IL-1 beta, not due to uremia per se but hemodialysis-dependent and reversible. Hemodialysis with AN 69 suppresses endogenous PGE2 synthesis in PBMC which is associated with increased LPS-induced IL-1 beta secretion in the presence of unchanged total IL-1 beta production. We speculate that PGE2 could inactivate the IL-1 beta converting enzyme which is essential for processing and secretion of mature IL-1 beta.

Cytokines in human renal interstitial fibrosis. II. Intrinsic interleukin (IL)-1 synthesis and IL-1-dependent production of IL-6 and IL-8 by cultured kidney fibroblasts

We compared cytokine production from transformed human fibroblast cell lines derived from either a kidney with interstitial fibrosis or a normal kidney to that from primary human foreskin fibroblasts. Fibrosis-derived as well as normal renal fibroblasts, but not skin fibroblasts, spontaneously produced the chemokine, IL-8, and the growth promoting cytokine, IL-6. Spontaneous IL-8 and IL-6 synthesis by renal fibroblasts was dependent on the intrinsic release of IL-1, since blocking IL-1 receptors with IL-1 receptor antagonist (IL-1Ra) partially inhibited the constitutive production of these cytokines. Both kidney cell lines had detectable mRNA and protein for IL-1 alpha and IL-1 beta. Renal and skin fibroblasts stimulated by picomolar concentrations of exogenous IL-1 or TNF-alpha produced large amounts of IL-6 and IL-8, whereas nanomolar concentrations of basic fibroblast growth factor did not. Fibrosis-derived cells expressed less high affinity IL-1 receptors (600 receptors/cell; KD = 0.6 pM) compared to normal renal fibroblasts (1000 receptors/cell). However, fibrosis-derived renal fibroblasts produce three- to fourfold more IL-8 and IL-6 in response to picomolar concentrations of IL-1 beta compared to cells derived from a normal kidney. As this enhanced production is not due to increased numbers of IL-1 receptors, we speculate that post-receptor responsiveness to either endogenous or exogenous IL-1 is greater in fibrosis-derived renal fibroblasts than in cells from normal kidneys.

Cytokines in human renal interstitial fibrosis. I. Interleukin-1 is a paracrine growth factor for cultured fibrosis-derived kidney fibroblasts

We studied the role of interleukin-1 beta (IL-1 beta) and basic fibroblast growth factor (bFGF) in the proliferative response of transformed human renal interstitial fibroblast cell lines established from either a kidney with glomerulonephritis and interstitial fibrosis or a normal kidney in comparison to primary human foreskin fibroblasts. Growth of fibrosis-derived renal fibroblasts was inhibited in the presence of IL-1 receptor antagonist (IL-1Ra) by 35% (P < 0.005), suggesting that these cells produce IL-1 and possess IL-1 receptors as part of paracrine growth. In contrast, spontaneous proliferation of fibroblasts derived from a normal kidney or normal skin were not inhibited by IL-1Ra. In fibrosis-derived but not in normal renal cells, fibronectin synthesis was increased 2.2-fold (P < 0.01) in the presence of IL-1Ra. Addition of exogenous IL-1 beta or bFGF stimulated proliferation of skin fibroblasts. In contrast, growth of fibrosis-derived renal fibroblasts was stimulated by IL-1 beta and unchanged by bFGF. Growth of normal kidney fibroblasts was unaffected by bFGF and inhibited by IL-1 beta. We conclude that compared to normal fibroblasts, fibrosis-derived renal fibroblasts have a different cytokine-response profile, are IL-1-dependent, produce IL-1 as a paracrine growth factor and do not proliferate to bFGF, a classical fibroblast growth factor.

The role of plasma coating on the permeation of cytokine-inducing substances through dialyser membranes

We studied the effects of coating of dialyser membranes with plasma proteins on the permeation of bacteria-derived cytokine-inducing substances (CIS). An in vitro dialysis circuit using polysulphone (PS) or modified cellulose triacetate (mCT) dialysers was used. Precoating of the dialysers was performed by recirculation of 10% normal human plasma for 30 min in the blood compartment and subsequent rinse with pyrogen-free saline. Samples from the blood compartment were tested for induction of interleukin-1 alpha (IL-1 alpha), interleukin-1 beta (IL-1 beta) and tumour necrosis factor (TNF alpha) at various time points after challenging the dialysate with sterile culture supernatants from Pseudomonas aeruginosa. Contamination of the dialysate resulted in the appearance of CIS in the blood compartment of both polysuphone modified cellulose triacetate (IL-1 alpha: PS, time 0: 81 +/- 11 pg/ml, time 60 min: 4747 +/- 1822 pg/ml, P < 0.05; mCT, time 0: 235 +/- 141 pg/ml, time 60 min: 1632 +/- 531 pg/ml, P < 0.05). The plasma protein layer reduced the penetration of CIS significantly only for polysulphone (IL-1 alpha: PS, time 60: 4747 +/- 1822 versus 880 +/- 525 pg/ml, P < 0.05; modified cellulose triacetate, time 60 min: 1632 +/- 531 pg/ml versus 930 +/- 326 pg/ml). Samples from the blood compartment contained < 6 pg/ml LAL-reactive material at all time points. We conclude that plasma coating of polysulphone dialysers reduces the permeability for CIS derived from Pseudomonas, either by reducing the effective pore size or by adsorption of proteins that bind CIS.(ABSTRACT TRUNCATED AT 250 WORDS)

Plasma levels of tumor necrosis factor (TNF) and soluble TNF receptors in kidney transplant recipients

Tumor necrosis factor-alpha is elevated in plasma during kidney transplant rejection. However, the measurement and biological activity of TNF alpha is influenced by inhibitory soluble TNF receptors. We therefore determined plasma levels of TNF alpha and the 2 soluble TNF receptors, the 55-kDa TNF receptor (TNF-sR55) and the 75-kDa TNF receptor (TNF-sR75), by immunoassays in 25 patients before and daily after kidney transplantation. Plasma samples were retrospectively assigned to 3 groups: (1) patients with well-functioning grafts (n = 14); (2) patients with biopsy-proven graft rejections (n = 7 patients with 10 rejections); and (3) patients with episodes of CsA nephrotoxicity (n = 4 patients with 9 samples). On the day of biopsy-proven graft rejection, TNF alpha increased from 8.6 +/- 0.9 pg/ml to 14.8 +/- 3.5 pg/ml (P < 0.02), TNF-sR55 from 6.6 +/- 1.3 ng/ml to 9.0 +/- 1.2 ng/ml (NS), and TNF-sR75 from 10.3 +/- 1.0 ng/ml to 15.3 +/- 2.0 ng/ml (P < 0.01). During episodes of CsA toxicity, TNF alpha levels did not change, TNF-sR55 increased from 5.2 +/- 0.5 ng/ml to 10.5 +/- 0.5 ng/ml (P < 0.01), and TNF-sR75 increased from 10.2 +/- 0.8 ng/ml to 17.5 +/- 0.9 ng/ml (P < 0.01). There was a strong correlation between serum creatinine and plasma TNF-sR55 (r = 0.7, P < 0.001) and TNF-sR75 (r = 0.7, P < 0.001), but not with TNF alpha. Therefore, levels of TNF-sR55 and TNF-sR75 were corrected for serum creatinine. An index expressing TNF alpha over actively released soluble receptors (index = TNF alpha/(corr.TNF-sR55 + corr.TNF-sR75)) detected rejection episodes with a sensitivity of 70-80% and a specificity of 89%. We conclude that the measurement of plasma TNF alpha in combination with its soluble receptors is superior to isolated TNF alpha determinations in discriminating acute graft rejection from episodes of CsA toxicity in kidney transplant recipients.