Antibiotic-induced endotoxin release has been demonstrated in vitro to depend on the type of antibiotic action and whether the cell wall remains intact [3, 4].
World J. Surg. 23, 75–79, 1999
WORLD Journal of
SURGERY © 1999 by the Socie´te´ Internationale de Chirurgie
Clinical Relevance of Antibiotic-induced Endotoxin Release in Patients Undergoing Hepatic Resection Masashi Ishikawa, M.D., Takayuki Miyauchi, M.D., Keiko Yagi, M.D., Hiroshi Chikaishi, M.D., You Fukuta, M.D., Hidenori Miyake, M.D., Masamitsu Harada, M.D., Sirou Yogita, M.D., Seiki Tashiro, M.D. First Department of Surgery, University of Tokushima, School of Medicine, 3-18-15, Kuramoto-cho, Tokushima City 770, Japan It has been proved that antibiotics binding to penicillin-binding protein 3 (PBP3) are associated with the greater release of endotoxin than those that bind to PBP2 in both in vitro and animal models. The aim of this study is to evaluate the potential clinical implications of antibioticinduced endotoxin release after hepatic resection. Forty-five patients who underwent hepatic resection in our clinic were enrolled. The patients were divided into two groups. Group A (n 5 26): antibiotics that bind primarily to PBP3, including cefmetazole (CMZ), latamoxef (LMOX), flomoxef (FMOX), were used. Group B (n 5 19); antibiotics that bind to both PBP2 and PBP3, including cefazolin (CEZ), cefoperazone (CPZ), cefotiam (CTM). Postoperative complications, liver functional tests, and chemical mediators [endotoxin, interleukins (IL-6, IL-8), tumor necrosis factor a (TNFa), granulocyte colony-stimulating factor (G-CSF), hepatotrophic growth factor (HGF) were examined after hepatic resection. There were no significant differences in the backgrounds of the two groups. Eight patients in each group developed postoperative complications; in particular, 9 of 13 patients with biliary tract carcinoma developed postoperative complications. No significant elevation of peripheral blood endotoxin was noted by the endospecy method, in any of the patients, although six died following sepsis. Pre- and postoperative levels of cytokines showed no significant difference between the two groups. Our data suggest that clinical antibiotic-induced endotoxin release would not occur after hepatic resection regardless of the antibiotic, probably owing to continuous scavenging of endotoxin from peripheral blood.
Patients with gram-negative infections often deteriorate despite effective antibacterial therapy. This is a manifestation of antibiotic-induced endotoxin release [1, 2]. Antibiotic-induced lysis of bacterial cells can result in increased plasma levels of endotoxin and other harmful bacterial products. Endotoxin is a cell wall component of gram-negative bacteria with the biologic activity of lipopolysaccharide. Antibiotic-induced endotoxin release has been demonstrated in vitro to depend on the type of antibiotic action and whether the cell wall remains intact [3, 4]. In the case of b-lactam antibiotics, the endotoxin-enhancing properties are closely related to the affinity for penicillin-binding proteins (PBPs) in the bacterial cell wall . Antibiotics with the highest affinity for PBP2 cause conversion of the bacteria to rounded This International Society of Surgery (ISS)/Socie´te´ Internationale de Chirurgie (SIC) article was presented at the 37th World Congress of Surgery International Surgical Week (ISW97), Acapulco, Mexico, August 24 –30, 1997. Correspondence to: M. Ishikawa, M.D.
spheroplast forms without cell lysis or endotoxin release. Antibiotics with the highest affinity for PBP3 cause conversion of the bacteria to long filament forms with a largest release of endotoxin. Antibiotics with a high affinity for PBP2 and PBP3 have an intermediate action. Endotoxin itself does not exhibit cytotoxicity. It has become clear that many of the features of toxemia result from a complex cascade of endogenous mediators triggered by endotoxin released from the bacterial cell wall . Although the release of endotoxin as a result of the rapid killing of bacteria has been considered to be potentially important in the clinical setting, some studies [7, 8] have not shown any increase in endotoxin with antibiotic therapy. The clinical relevance of the differences between antibiotics with respect to the release of endotoxin is unknown owing to a lack of relevant studies. Endotoxemia is frequently observed with obstructive jaundice, liver cirrhosis, and after hepatectomy . The proposed mechanisms of endotoxemia after hepatectomy are damage to the Kupffer cells, the presence of peripheral arteriovenous shunts, and deterioration of immunity to harmful substances in the blood. Studies have shown that a conventional Limulus test is not specific to endotoxin because of the presence in amebocyte lysate of a b-D-glucan-sensitive factor [10, 11]. An endotoxin assay using a synthetic chromogenic peptide as substrate (endospecy method) has been developed . This assay is more sensitive than the previous gelation assay and provides quantitative measurement of endotoxin. A randomized clinical study was performed to explore the potential clinical implications of the endotoxin - releasing properties of antibiotics in patients who had undergone hepatectomy. Postoperative infections, serum endotoxin levels, and levels of other chemical mediators were followed after operation. Subjects and Methods Patients Forty-five patients (33 men, 12 women; mean age of 60.6 6 10.8 years) who underwent hepatectomy in our clinic between 1994 and 1996 were enrolled in the study. The operations were
World J. Surg. Vol. 23, No. 1, January 1999
Table 1. Postoperative infections in patients undergoing hepatectomy. Site of infection
Abdominal cavity Sepsis Lung Biliary tract Urinary tract Peritonitis
6 6 4 2 1 1
Table 4. Postoperative infections and surgical techniques.
No. of patients
60.0 6 14.3
61.0 6 8.7
HCC CBT CGB Meta HCC CBT CGB Meta
Table 2. Cultured microorganisms among 45 patients undergoing hepatectomy. Cultured microorganisms
No. of cases
Staphylococcus Streptococcus Enterococcus Neisseria Klebsiella Escherichia coli Enterobacter Pseudomonas aeruginosa Citrobacter Acinetobacter Serratia Candida Total
3 1 21% 1 3 2 3 12 71% 3 2 1 1 2 } 8% 34 } 100%
Table 3. Postoperative infections from the view point of penicillinbinding proteins (PBPs).
No. of patients
No. of infected cases/ death
62.4 6 9.3
58.2 6 12.6
HCC CBT CGB Meta HCC CBT CGB Meta
14 7 1 4 10 2 3 4
HCC: hepatocellular carcinoma; CBT: carcinoma of the biliary tract; CGB: carcinoma of the gallbladder; Meta: metastatic carcinoma.
performed for hepatocellular carcinoma (n 5 24), carcinoma of the biliary tract (n 5 9), carcinoma of the gallbladder (n 5 4), or metastatic carcinoma (n 5 8). We performed 26 partial hepatic resections, segmentectomies, 11 lobectomies, and 1 extended lobectomy. Therapy The patients were randomized to receive intravenous cefmetazole, latamoxef, and flomoxef (group A) or cefazolin, cefoperazone, and cefotiam (group B). The antibiotics in group A have a high affinity for the PBP3 of gram-negative bacteria, and the antibiotics in group B have a high affinity for both PBP2 and PBP3 . The antibiotics were given twice a day (60-minute infusions), and the duration of treatment was 120 hours after operation. No
5 6 3 2 19 3 1 6
Reconstruction of the biliary tract (no. of patients) 8
Fig. 1. Serial changes of serum endotoxin levels in 45 patients undergoing hepatectomy. a. Endospecy method. b. Toxicolor method. *p , 0.05.
other antibiotics were used. After the 120 hours the antibiotics were stopped or changed to an appropriate regimen. Follow-up Liver function tests, postoperative complications, and the general condition of the patient were assessed after hepatectomy. At 0, 3, 24, 72, and 168 hours blood was collected in pyrogen-free tubes and immersed in ice for determination of levels of endotoxin and chemical mediators. Samples were centrifuged at 1500 3 g for 20 minutes. Serum was collected, immediately frozen, and stored at 280°C until testing. The endotoxin levels were determined by two methods, using the endospecy method (Seikagakukougyo, Osaka, Japan) and the toxin color test (Seikagakukougyo). These methods employ the Limulus assay and have a detection limit of 1 pg. The normal rarge is less than 10 pg/ml. Interleukin-6 (IL-6), IL-8, and granulocyte-colony stimulating factor (G-CSF) levels were measured by an enzyme-linked immunoassay (ELISA) kit (Immunotech, Marseille, France). The hepatocyte growth factor (HGF) concentration was measured by an ELISA kit (Ootsuka, Tokushima, Japan). Tumor necrosis factor a (TNFa) levels were also measured by ELISA kit (Ootsuka, Tokushima, Japan). The detection limit of TNFa in plasma was 50 pg/ml. Statistical Analysis For comparison of continuous variables, Student’s t-test was used. Differences between the two groups with respect to variable times were compared using the analysis of variance (ANOVA) for
Ishikawa et al.: Antibiotic-induced Endotoxin
Fig. 2. Serial changes of serum cytokine levels in 45 patients undergoing hepatectomy. a. Interleukin-6. b. Interleukin-8. c. Hepatocyte growth factor. d. Granulocyte-colony stimulating factor. No significant differences regarding changes of serum cytokines were seen between the two groups. POD: postoperative day.
repeated measures. A p value of , 0.05 was considered statistically significant. Results Of the 45 patients, 16 had a total of 20 postoperative infections (Table 1). Four patients developed sepsis, two patients within 1 week and the other two within 1 month of the operation. These four patients and two of the six patients who developed an intraabdominal abscess died of multiple organ failure. Of the pathogens cultured, 71% were gram-negative. Some of the patients had more than one gram-negative pathogen grown in culture (Table 2). Five patients who had a long history of antibiotic therapy showed a positive Canditec test. Eight patients in each group developed postoperative infections (Table 3). There were five deaths in group A and one death in Group B. Of the nine patients who underwent reconstruction of the biliary tract, eight had postoperative infections (Table 4). There were no significant correlations between the occurrence of infection and the presence of liver disease or the type of surgery. Endotoxin Levels The plasma endotoxin levels were measured preoperatively and on postoperative days (POD) 1, 3, and 7. The preoperative endotoxin level by the endospecy method in all patients was 4.1 6 0.9 pg/ml (Fig. 1a). The levels remained low postoperatively in all
patients, including those who developed sepsis. No significant differences were seen between groups A and B. On the other hand, markedly elevated endotoxin concentrations by the toxicolor method was observed after hepatectomy (Fig. 1b). There was also no significant differences between groups A and B. Chemical Mediator Levels Serum IL-6 levels increased in all patients reaching a peak immediately after the operation and falling within 72 hours (Fig. 2a). Neither the absence or presence of infections nor the type of antibiotics had a significant influence on the IL-6 concentration. Although the IL-8 levels tended to be higher in patients with an infection, no significant differences were seen between groups A and B (Fig. 2b). The TNFa levels were less than the detection limit in all but two of the patients, including those who developed sepsis. The patients who subsequently developed an infection showed a significant increase in HGF levels form POD 3 to POD 7 (Fig. 2c). There was, however, no significant difference between the two groups of antibiotics used. The same trend was observed in the granulocyte count after hepatectomy. G-CSF levels followed the same pattern as the IL-6 and IL-8 levels (Fig. 2d). Discussion In 1895 it was recognized that the treatment of infections could worsen the clinical condition of patients with syphilis due to the
breakdown products of the bacteria. It was termed the JarischHerxheimer reaction . Reilly et al.  reported in 1950 that the release of endotoxin as a result of rapidly killing bacteria was potentially important. However, much remains unknown regarding the mechanism of the endotoxin release and its adverse influence on the host. During the 1980s in vitro studies [14, 15] showed that differences in endotoxin release can be attributed to different models of antibacterial activity, not to the number of bacteria. It is well known that antibiotics with selective affinity for PBP2 result in greater endotoxin release compared to antibiotics with a high affinity for PBP3. Some investigators [16, 17] using animal models of infection with gram-negative bacteria have reported that endotoxin release as a result of antibiotic therapy paralleled clinical deterioration. In patients with traumatic gramnegative bacterial sepsis  and urosepsis , it has been reported that antibiotics associated with a greater release of endotoxin were also associated with greater mortality. Prins et al.  have reported that cytokine levels such as IL-6, IL-8, and TNFa increased 10% to 40% after ceftazidime treatment compared with no increase in the imipenem-treated patients. However, Harthug et al.  and Brandtzaeg et al.  found no increase in plasma endotoxin levels in patients after unspecified antibiotic treatment. Endotoxin release is thought to trigger the cascade of clinical findings and the release of multiple endogeneous mediators associated with severe infection [20 –22]. Therefore we studied antibiotic-induced endotoxin release in patients after hepatectomy indirectly by measuring plasma chemical mediator levels. Our study did not show any increase in endotoxin by the endospecy method with antibiotic therapy after hepatectomy, whereas marked elevations of endotoxin levels were observed by the toxicolor method, as many investigators have reported previously [9, 23]. However, there were no significant differences in endotoxin release following different antibiotic treatment. No significant differences in IL-6, IL-8, TNFa, HGF, or G-CSF levels were seen regardless of the antibiotic used. However, the HGF levels and granulocyte counts were elevated on POD 3 in patients who developed an infection. Higher mortality was noted in the group treated with antibiotics with a high affinity for PBP3 (19.2% vs. 5.3%, p 5 0.3), but all of the deaths occurred more than 7 days after the antibiotic treatment was stopped or changed. The existence of a causal link between antibiotic therapy and an adverse outcome due to acute release of endotoxin was difficult to show from our study. However, our results differed from previous in vitro [14, 15] and vivio [17–19] studies that a 3- to 20-fold increase in the total concentration of endotoxin was a consequence of antibiotic action on gram-negative bacteria. This difference could have been caused by a difference in the severity of the infections and bacterial load, neutralization of endotoxinbinding proteins or antibodies to endotoxin, or phagocytosis induced immediately after endotoxin release.
Conclusions Antibiotic-induced endotoxin release does not usually occur after hepatectomy, regardless of the antibiotic used. However, antibiotics associated with a greater release of endotoxin should not be administered to critically ill septic patients.
World J. Surg. Vol. 23, No. 1, January 1999
Re´sume´ On sait que le largage d’endotoxines est plus important quand on utilise des antibiotiques qui se lient `a la prote´ine 3 (PBP3), par rapport `a l’utilisation des antibiotiques en liaison avec la prote´ine 2, et ce aussi bien dans les mode`les animaux qu’in vitro. L’objectif de cette ´etude a ´ete´ d’e´valuer les implications cliniques potentielles du largage d’endotoxine en rapport avec des antibiotiques apre`s re´section he´patique. On a ´etudie´ 45 patients ayant eu une re´section he ´patique dans notre ´etablissement. On a divise ´ les patients en deux groupes: Groupe A (n 5 26); antibiotiques qui se liaient principalement `a la PBP3, comprenant le cefmatazole (CMZ), latamoxef (LMOX) and flomoxef (FMOX), et Groupe B (n 5 19); antibiotiques qui se combinaient avec la PBP2 et la PBP3, comprenant le cefazoline (CEZ), le cefoperazone (CPZ) et le cefotiam (CTM). Les complication postope´ratoires, les tests de fonction he´patique et les me´diateurs cliniques (endotoxine, IL-6, IL-8, TNF-(, G-CSF et hepatocyte growth factor (HGF)) ont ´ete´ analyse´s apre`s re´section he ´patique. On n’a retrouve ´ aucune diffe´rence significative entre les deux groupes en ce qui concerne les caracte´ristiques de la population. Huit patients dans chaque groupe ont de´velopp“e une complication postope ´ratoire et en particulier, 9 des 13 patients ayant un cancer des voies biliaires ont de´veloppe ´ des complications. On n’a mis en ´evidence aucune diffe´rence significative en ce qui concerne le taux d’endotoxines dans le sang pe´riphe´rique par la me ´thode d’endospe ´cie bien que six patients sont de´ce´de´s apre`s leur sepsis. Les taux de cytokines pre´ et postope´ratoires ne diffe ´raient pas entre les deux groupes. Nos re´sultats sugge`rent que le largage d’endotoxine induit par les antibiotiques ne se produit pas dans de nombreux cas de re ´section he´patique vraisemblablement en raison de la clairance continue d’endotoxines dans le sang pe´riphe´rique. Resumen Se ha demostrado, tanto “in vitro” como en modelos animales experimentales, que los antibio ´ticos capaces de adherirse a la prote“ina 3 ligada a la penicilina (PBP3), producen una mayer liberacio ´n de endotoxina que aquellos que se adhieren a la PBP2. El objetivo del presente estudio consiste en averiguar si esta accio ´n liberadora de endotoxinas, producida por los antibi“oticos, tiene repercusiones clı´nicas favorables sobre pacientes hepatectomizados. La poblacio ´n estudiada de enfermo con reseccio ´n hepa´tica fue 45. Se dividio ´ en 2 grupos: Grupo A (n 5 26) a los que se les administro ´ antibio ´ticos capaces de adherirse, ante todo, a la PBP3. Se emplearon los siguientes antibio ´ticos: CMZ, LMOX y FMOX. Grupo B (n 5 19) a los que se administro ´ antibio ´ticos capaces de adherirse tanto a la PBP2 como a la PBP3. Se emplearon los siguientes antibio ´ticos CEZ, CPZ, CTM. Despue´s de la reseccio ´n hepa´tica se evaluaron: las complicaciones postoperatorias, pruebas de funcio ´n hepa´tica y determinados mediadores, tales como: endotoxina, IL-6, IL-8, TNF-a, G-CSF y HGF. No se encontraron diferencias significativas entre ambos grupos; 8 pacientes, de cada grupo, presentaron complicaciones postoperatorias; sin embargo, combiene sen ˜alar que el mayor nu ´mero de complicaciones postoperatorias se observo ´ en 9 (de un total de 13) pacientes con carcinoma del tracto biliar. No se constato ´, en ningu ´n paciente, incrementos significativos de endotoxina en sangre perife´rica, a pesar de que 6 murieron por septicemia. Los niveles plasma´ticos de citocinas pre y postoperatorios, no revel-
Ishikawa et al.: Antibiotic-induced Endotoxin
aron diferencia significativa alguna entre ambos grupos. Nuestros hallazgos sugieren, que la liberacio ´n de endotoxinas inducida por la administracio ´n de antibio ´ticos, demostrada “in vitro” y en modelos experimentales, no se produce en pacientes hepatectomizados, a pesar de la variedad de antibio ´ticos utilizados. Probablemente, esto sea debido a que en sangre perife´rica se produce una continua depuracio ´n de endotoxinas. References 1. Shenep, J.L., Flynn, P.M., Barrett, F.F., Stidham, G.L.: Serial quantitation of endotoxemia and bacteremia during therapy for gramnegative bacterial sepsis. J. Infect. Dis. 157:565, 1998 2. Hurley, J.C.: Antibiotic-induced release of endotoxin: a reappraisal. Clin. Infec. Dis.. 15:840, 1992 3. Jackson, J.J., Kropp, H.: b-Lactam antibiotic-induced release of free endotoxin: in vitro comparison of penicillin-binding protein (PBP)2specific imipenem and PBP 3-specific ceftazidime. J. Infect. Dis. 165:1033, 1992 4. Crosby, H.A., Bion, J.F., Penn, C.W., Elliot, T.S.J.: Antibiotic-induced release of endotoxin from bacteria in vitro. J. Med. Microbiol. 40:23, 1994 5. Tuomanen, E., Gilbert, K., Tomasz, A.: Modulation of bacteriolysis by cooperative effects of penicillin-binding proteins la and 3 in Escherichia coli. Antimicrob. Agents Chemother. 30:659, 1986 6. Matsuda, K., Shibata, K., Sanda, M., Nakagawa, S., Kawakami, M.: Comparative effect of imipenem and other b-lactam antibiotics on the release of endotoxin from gram-negative bacteria. Chemotherapy 41:345, 1993 7. Harthug, S., Bjotvatn, B., Osterud, B.: Quantitation of endotoxin in blood from patients with meningococcal disease using a Limulus lysate test in combination with chromatogenic substrate. Infection 11:192, 1983 8. Brandtzaeg, P., Kierulf, P., Gaustad, P., Skulberg, A., Bruun, J.N., Halvorsen, S., Soresen, E.: Plasma endotoxin as a predictor of multiple organ failure and death in systemic meningococcal disease. J. Infect. Dis. 159:195, 1989 9. Nagata, Y., Tanaka, N., Orita, K.: Endotoxin-induced liver injury after extended hepatectomy and the role of Kupffer cells in the ract. Surg. Today 24:441, 1994 10. Kakimuma, A., Asano, T., Tori, H., Sugino, Y.: Gelation of Limulus amebocyte by an antitumor (1 3 3)-b-D-glucan. Biochem. Biohys. Res. Commun. 101:434, 1981
79 11. Obayashi, T., Tamura, H., Tanaka, S., Ohki, M., Takahashi, S., Arai, M., Masuda, M., Kawai, T.: A new chromogenic endotoxin-specific assay using recombined Limulus coagulation enzymes and its clinical applications. Clin. Chim. Acta 149:55, 1985 12. Tramont, E.C.: Treponema pallidum (syphilis). In Principles and Practice of Infectious Disease (3rd ed.). New York, Churchill Livingstone, 1990, pp. 1794 –1808 13. Reilly, J., Compagnon, A., Tournier, P.: Observations anatomoclinques. Ann. Med. 51:599, 1950 14. Shenep, J.L., Mogan, K.A.: Kinetics of endotoxin release during antibiotic therapy for experimental gram-negative bacterial sepsis. J. Infect. Dis. 150:380, 1984 15. Cohen, J., McConnell, J.S.: Release of endotoxin from bacteria exposed to ciprofloxacin and its prevention with polymyxin B. Eur. J. Clin. Microbiol. 5:13, 1986 16. Tauber, M.G., Shibl, A.M., Hackharth, C.J., Larrich, J.W., Sande, M.A.: Antibiotic therapy, endotoxin concentration in cerebrospinal fluid, and brain edema in experimental Escherichia coli meningitis in rabbits. J. Infect. Dis. 156:456, 1987 17. Rokke, O., Revhaug, A., Osterud, B., Giercksky, K.E.: Increased plasma levels of endotoxin and corresponding changes in circulatory performance in a porcine sepsis model: the effect of antibiotic administration. Prog. Clin. Biol. Res. 272:247, 1988 18. Mock, C.N., Jurkovick, G.J., Pries, D.J., Maier, R.V.: Clinical significance of antibiotic endotoxin-releasing properties in trauma patients. Arch. Surg. 130:1234, 1995 19. Prins, J.M., van Agtmael, M.A., Kuijper, E.J., van Deventer, S.J.H., Speelman, P.: Antibiotic-induced endotoxin release in patients with gram-negative urosepsis: a double-blind study comparing imipenem and ceftazidime. J. Infect. Dis. 172:886, 1995 20. Hale, D.J., Robinson, J.A., Loeb, H.S., Gunnar, R.M.: Pathophysiology of endotoxin shock in man. In Handbook of Endotoxin, R.A. Proctor, editor. Amsterdam, Elsevier, 1986, pp. 1–17 21. Michie, H.R., Manogue, K.R., Spriggs, D.R., Revhaug, A., O’Dwyer, S., Dinarello, C.A., Cerami, A., Wolff, S.M., Wilmore, D.W.: Detection of circulating tumor necrosis factor after endotoxin administration. N. Engl. J. Med. 318:1481, 1988 22. Suffredini, A.F., Fromm, R.E., Parker, M.M., Brenner, M., Kovacs, J.A., Wesley, R.A., Parillo, J.E.: The cardiovascular response of normal human to the administration of endotoxin. N. Engl. J. Med. 321:280, 1989 23. Nolam, J.P.: Endotoxin, reticuloendothelial function and liver injury. Hepatology 1:458, 1981