Antimicrobial-induced endotoxaemia in patients with sepsis in the field of acute pyelonephritis.

EJ Giamarellos-Bourboulis, J Perdios, P Gargalianos, J Kosmidis, H Giamarellou 
 4th Department of Internal Medicine, University of Athens, Medical School, Athens General Hospital "G Gennimatas," Greece. , Greece

Correspondence Address:
E J Giamarellos-Bourboulis
4th Department of Internal Medicine, University of Athens, Medical School, Athens General Hospital «DQ»G Gennimatas,«DQ» Greece.
Greece

Full Text

In-vitro results on antimicrobial-induced endotoxaemia have raised fears of a probable perpetuation of the systemic inflammatory response after the start of antimicrobial chemotherapy. Administration of antimicrobial agents leads to bacterial killing and release of endotoxins from the bacterial cell wall. Published data indicates that administration of beta-lactam anti-bacterial agents is associated with a higher release of endotoxins (LPS) from the bacterial cell wall as compared to that observed in association with the fluoroquinolones.[1],[2],[3],[4],[5] It is feared that this release of endotoxins may trigger mononuclear cells to secrete pro-inflammatory cytokines.[6] Although in-vitro results have been verified in experimental animals,[7] clinical studies in human subjects are scarce and have mainly concentrated on imipenem and ceftazidime.[8],[9],[10]

The present study was undertaken to compare three different classes of antimicrobial agents for their ability to cause endotoxin release and to trigger the secretion of tumour necrosis factor alpha (TNF?). The agents selected were cefuroxime (as a representative for ?-lactam antimicrobial agents), netilmicin, an aminoglycoside and ciprofloxacin, which belongs to the fluoroquinolone group. Such a study has been performed in experimental animals,[11] but to the best of our knowledge, this is the first such study undertaken in human beings.

::   Patients and methods

Consecutive patients presenting with pyelonephritis and sepsis were enrolled prospectively in this study that was carried out between January 1998 and January 2001 after receiving an approval from the Ethics Committee of the General Hospital of Athens “G. Gennimatas”. The subjects’ epidemiological and clinical data are shown in [Table:1]. The diagnosis of pyelonephritis and sepsis was made on the basis of the following criteria:[11],[12],[13] a) case history compatible with acute pyelonephritis comprising at least two spikes of fever above 380C and presence of lumbar tenderness on examination, b) pyuria defined as presence of more than 10 polymorphs per high-power field under a light microscope, c) positive urinary culture with the colony count exceeding 104 cfu/ml for a single Gram-negative bacterial species, and d) the presence of at least two signs of the systemic inflammatory response syndrome i.e. fever above 380C, tachypnoea (more than 20 breaths/ minute), tachycardia (more than 90 beats per minute) and presence of leucocytosis or leucopenia.[14] Patients with history of taking any antimicrobial agent in the one-month period prior to presentation were excluded. All patients were subjected to a renal ultrasound examination. The findings of ultrasound examination are depicted in [Table:1].

Patients were randomly assigned to receive an intravenous antimicrobial agent (cefuroxime 1.5g tid or netilmicin 150mg bid or ciprofloxacin 400mg bid for a total of seven days). Three blood samples were collected by venipuncture from a forearm vein at fifteen-minute intervals before the start of therapy. Mean values of LPS and of TNF? of these determinations were applied as indicators of steady state levels before therapy. Blood samples were then drawn at exactly 0.5, 1, 1.5, 2, 3, 8 and 24 hours after the administration of the first dose of each antimicrobial agent. Samples were then collected daily and at 24-hour intervals only if fever (axillary temperature above 37.50C) persisted. The time period of blood sampling was selected in analogy to former studies in humans[8],[9] and in accordance to microbiological data of the kinetics of LPS release by bacteria after addition of antimicrobial agents in the growth medium.[2]

One aliquot of 5ml of each sample was cultured into broth-containing tubes (Becton Dickinson, Cockeysville Md) and incubated for a total of seven days at 350C. Another aliquot of 3ml was placed into a pyrogen-free tube (Oxoid Ltd, London, UK) and it was centrifuged for five minutes at 2,500g and 40C. The serum samples were kept refrigerated at -700C until being assayed. Identification of bacterial isolates from blood or urine was made by the API 20E system (bioMérieux, Paris, France).

Serum samples were processed for the determination of LPS and of TNF? as follows: Before being assayed for LPS, the serum sample was diluted 1:10 with pyrogen-free water (BioWhitaker, Walkersville, Maryland, USA) and incubated for five minutes at 700C in a shaking water bath. LPS level was then measured by the LAL QCL-1000 colorimetric assay (BioWhitaker, Walkersville, Maryland, USA, lower detection limit 0.1 EU/ml). False-positive results were excluded by the using serum samples obtained from healthy volunteers. The inter-day coefficient of variation of the assay was 10.2%. Concentrations of TNF? were determined by an enzyme-immunoassay (Amersham, London, UK, lower detection limit 10pg/ml). All determinations were performed in duplicate. The difference between the levels were analysed by one-way analysis of variance (ANOVA).[15] Values of P below or equal to 0.05 were considered as significant. To avoid random significances, a Bonferroni correction was applied.

Endotoxaemia was defined as any concentration of LPS in serum above 0.1 EU/ml.[16] Comparisons for the presence of endotoxemia among patients with fever persisting for 48 hours post-therapy initiation were performed by Fischer’s exact test (P<0.05).

::   Results

Bacteremia was detected pre-therapy in five, one and six patients in the cefuroxime, netilmicin and ciprofloxacin groups, respectively [Table:1]. None of the patients had bacteremia after the institution of therapy, irrespective of the antimicrobial agent used. The median concentrations of LPS before the start of therapy were 0.91, 0.64 and 1.33 EU/ml, respectively. The differences in these levels were not statistically significant. Signs of multiple organ failure with respiratory distress and intra-vascular coagulation were present in only one patient treated in the cefuroxime group, who was admitted in the ICU.

The concentrations of LPS and of TNF? of each study group over time after the initiation of treatment are shown in [Table:2]. The levels of these substances in 2-hour sample were significantly higher in the cefuroxime group than those in the netilmicin and ciprofloxacin groups. For all other timed-samples, there was no significant difference amongst the three groups.

Fever persisted for 48 hours after the start of therapy with cefutoxime, netilmicin and ciprofloxacin in 12, 10 and 11 patients, respectively. Endotoxaemia was detected in nine patients in each group. The difference amongst the three groups on this count was not statistically significant. All the patients who were detected to have endotoxemia presented with signs of sepsis. In contrast, patients who persisted to have fever for at least 48 hours after the initiation of therapy did not demonstrate any signs of sepsis. [Table:3] depicts the correlation between persistence of fever and the rate of endotoxaemia.

::   Discussion

In-vitro results on the release of LPS by antimicrobial agents acting on Gram-negative bacteria[3],[4],[5] raise a lot of dilemma whether antimicrobial chemotherapy might amplify the systemic inflammatory response. To clarify that hypothesis, cefuroxime, netilmicin and ciprofloxacin were administered intravenously to patients who presented with acute pyelonephritis caused by Gram-negative bacteria and signs of sepsis. The patients were closely monitored clinically and readings of LPS and TNF? serum levels were obtained at pre-determined intervals. Although all patients enrolled presented with sepsis, not all of them were detected to have bacteraemia. Other authors have demonstrated similar findings.[17] It should also be noted that previously reported clinical trials on the comparative effect of imipenem and ceftazidime on serum LPS of patients with urosepsis did not comprise patients with bacteraemia alone.[8]

Results revealed that concentrations of LPS in serum did not differ significantly among the three study groups over the first 24 hours after start of treatment with the exception of the two-hour sample. This sample was characterized by a peak of serum LPS and TNF? of patients treated with cefuroxime compared to patients receiving netilmicin or ciprofloxacin. A similar peak of LPS in serum was reported two hours after the administration of ceftazidime, which is an inhibitor of PBP-3 like cefuroxime.[9] Attenuation of the release of TNF? and of interleukin-6 (IL-6) has been shown in mice pre-treated with ciprofloxacin before challenge with LPS,[18] but similar results were not found in the present study. Irrespective of changes in the levels of LPS and TNF? throughout the study, none of the patients showed signs of sepsis after 48 hours of therapy and none of the patients had detectable bacteremia, demonstrating the success of the antimicrobial therapy.

No direct evidence was found that persistence of fever after 48 hours of therapy might be attributed to antimicrobial-induced release of LPS since the frequency of endotoxaemia was similar in patients at 48, 72 or 92 hours after the start of treatment, irrespective of the agent used.

Three studies have attempted to investigate the effect of antimicrobial agents on the in-vivo release of LPS in patients with infections caused by Gram-negative bacteria. Maury et al enrolled 18 patients with various infections. These patients were treated with a variety of ?-lactam agents and LPS levels in the serum were determined prior to and at one- and four- hours after the start of therapy.[19] The other two studies investigated the comparative effect of imipenem and ceftazidime on 68 patients with melioidosis[9] and on 33 patients with urosepsis.[8] These patients were followed up for a short period of six- to eight- hours after the initiation of therapy.

On the basis of findings obtained in this study, the authors agree with others that the pathogenesis of sepsis is a very complicated process,[20] so that it could not be expected to obey to the simplistic rules of the in-vitro conditions where single bacterial cells are exposed to a certain concentration of an anti-microbial agent in the close proximity of a mononuclear cell culture.

References