Beta-lactam antibiotics

Neutropenia

Neutropenia due to beta-lactam antibiotics has been reviewed [ ]. It usually occurs after high-dose therapy lasting more than 10 days, and the frequency rises with cumulative dose. It is often preceded by a fever or rash, usually lasts less than 10 days, and is uncommonly associated with infectious complications or death. Although any beta-lactam can cause neutropenia, there seems to be a high incidence associated with the prolonged use of cefepime or piperacillin + tazobactam.

While in large series of several thousands of patients, neutropenia has generally been reported as an adverse effect in under 0.1–1.0% [ ], an overview in 1985 estimated that neutropenia (neutrophil count below 1.0 × 10 ⁹ /l) occurs in up to 15% of all patients treated with high-dose intravenous beta-lactam antibiotics for more than 10 days [ ]. In subsequent series of patients treated for several weeks with various beta-lactam antibiotics, up to 25% developed neutropenia [ , , , , ].

In one series, 22 of 128 patients receiving cloxacillin for staphylococcal infections became neutropenic [ ]. Neutropenia appeared, on average, 23 days after the start of therapy. The same authors, in a somewhat bigger population, found neutropenia in 1.1% of patients who received cumulative doses of oxacillin below 150 g, but in 43% (22 of 51) who received more than 150 g [ ]. Similarly, in 132 patients, cefapirin in a cumulative dose of less than 90 g did not cause neutropenia, but did in 26% (five of 19) of those who used higher total doses [ ].

In addition, for a given compound, higher daily doses increase the risk of neutropenia. In one study, seven of 14 patients became neutropenic with a mean dose of penicillin G of 17 g/day after 9–23 days [ ], while in another study only 12 of 193 patients developed neutropenia with a mean dose of 11 g/day for an average duration of 20 days [ ]. A considerable extension of the aforementioned study [ ] corroborated this: neutropenia occurred in 35% of those treated with a mean daily dose of 17 g of penicillin G for an average of 23 days, while it was found in only 8% of those who received 12 g for 22 days [ ].

Epidemiological studies [ , ] as well as single cases of severe neutropenia observed with newer compounds have invariably confirmed the dose- and time-dependent pattern described above. For example, cefepime, a fourth-generation cephalosporin, possibly or probably caused neutropenia in only 0.2% of 3314 treatment courses, while 7.1% of those who received cefepime for several weeks developed neutropenia [ ]. Accordingly, high-dose cefepime (150 mg/kg/day) was given for 7–10 days to 43 children for bacterial meningitis without causing neutropenia [ ], while there were two cases in adults after total doses of 112 g (over 28 days) and 120 g (30 days) respectively [ ].

It is therefore not surprising that after consecutive or simultaneous treatment with more than one beta-lactam antibiotic, neutropenia is similarly observed, suggesting additive toxicity [ , , ].

There is so far no clear evidence about the different risks of different compounds. The data best fit the assumption that the risk of neutropenia correlates with the cumulative dose, or probably more precisely with the area under the serum concentration versus time curve (AUC). Hence, renal insufficiency is a potential risk factor. In addition, beta-lactam-antibiotic-induced leukopenia has been associated with hepatic dysfunction [ ].

Recovery in most cases is rapid and uneventful. In patients who were re-exposed to the same or other beta-lactam antibiotics, there was similar dependence of neutropenia on the duration of treatment and the cumulative dose [ ]. Whether the use of hemopoietic growth factors, and in particular G-CSF, is useful is unclear. There are case reports of positive clinical effects [ ]. However, the recovery time in these reports did not differ from that observed in a large population of untreated patients [ ]. Theoretically, early use of growth factors could even be counterproductive, since some toxic effects of beta-lactam antibiotics on bone marrow cells appear to be related to the S-phase of the cell cycle [ ]. On the other hand, G-CSF maintained the proliferative activity of bone marrow cells exposed to ceftazidime in vitro, if it was added at the beginning of the culture process [ ].

Neutropenia is accompanied by fever, eosinophilia, and/or a rash in more than 80% of cases.

Hemolytic anemia

Immune hemolytic anemia was originally described with penicillin G, but subsequently also with other penicillins and cephalosporins. It is usually seen during treatment with very high doses after the so-called “drug absorption” mechanism. The beta-lactam antibiotic binds covalently to the erythrocyte surface, forming complete antigens, which can in turn bind drug-specific circulating IgG antibody. Typically, direct and indirect Coombs’ tests are positive, but complement is not activated [ ]. Rarely, other immunological mechanisms have been observed, for example the so-called “innocent bystander” type of hemolysis [ ], in which complement can be detected on the erythrocyte surface. Some cephalosporins, clavulanic acid, and imipenem + cilastatin can cause positive direct antiglobulin tests [ ]. The phenomenon is due to non-specific serum protein absorption on to the erythrocyte membrane and is not related to immune hemolytic processes. Detection of non-immunologically bound serum proteins is improved if the reagents used include additional anti-albumin activity [ ]. The phenomenon is a known source of difficulties in evaluating suspected immune hemolysis or routine cross-matching of blood products [ ].

The true frequency of the phenomenon is unclear, since it has not been positively sought. However, in a 20-year retrospective analysis of 73 patients with drug-dependent antibodies to 23 different drugs from an immune hematological reference laboratory in the USA, cephalosporins were at the top of the list (n = 37), followed by penicillins and/or penicillin derivatives (n = 12), non-steroidal anti-inflammatory drugs (n = 11), and others (n = 13) [ ].

Eosinophilia

Virtually all beta-lactam antibiotics can cause eosinophilia, either isolated or in the context of very different reactions.

Bleeding disorders

Treatment with beta-lactam antibiotics can result in impaired hemostasis and bleeding [ ]. The true incidence of bleeding is difficult to assess, since many non-antibiotic factors can be involved, such as malnutrition with vitamin K depletion [ ], renal insufficiency [ ], and serious infection [ ]. Cancer, the use of cytotoxic drugs, and surgery have made conclusive interpretation of coagulation disorders difficult [ ]. Between the different beta-lactam antibiotics, the reported incidence of clinical relevant bleeding varies widely, and was highest with moxalactam (22% of patients), now withdrawn [ ]. With other cephalosporins, bleeding was observed with frequencies ranging from 2.7% (cefazolin/cefalotin) to 8.2% (cefoxitime) [ ]. Two basic mechanisms have been proposed: altered coagulation and altered platelet numbers and function.

Altered coagulation : Both direct inhibition of the hepatic production of vitamin K-dependent clotting factors and alterations in the intestinal flora, with subsequent reduction of microbial supply of vitamin K, have been implicated [ , ]. The relative role of either mechanism is difficult to assess, but experimental support for the flora theory is weak [ , ].

Several of the cephalosporins that contain either a non-substituted N -methylthiotetrazole (NMTT) side chain, such as cefamandole, cefamazole, cefmenoxime, cefmetazole, cefoperazone, cefotetan, and moxalactam, as well as a substituted NMTT side chain (ceforanide, ceforicid, or cefotiam), or the structurally similar N -methylthiotriazine ring in ceftriaxone and the 2-methyl-1,2,4-thiadiazole-5-thiol (MTD) ring of cefazolin interfere with vitamin K-dependent clotting factor synthesis in the liver (factors II, VII, IX, and X). The molecular mechanism involves dose-dependent inhibition of microsomal carboxylase function, as shown in animals [ ], and inhibition of the epoxide reductase system in both animals and man [ ]. Cefoxitin, a non-NMTT compound, was implicated significantly more often than the NMTT-containing compounds cefamandole and cefoperazone [ ].

The NMTT must leave the parent antibiotic to inhibit the carboxylation reaction [ ]. The NMTT molecule leaves the parent cephalosporin either during spontaneous hydrolysis in the blood or during nucleophilic cleavage of the beta-lactam ring by intestinal bacteria, and is reabsorbed from the gut into the portal circulation [ ]. Studies in healthy volunteers show compound-related differences in the ability of NMTT antibiotics to generate free NMTT, reflecting drug-specific differences in susceptibility to in vitro hydrolysis or differences in gut NMTT production, which may be a function of biliary excretion of the drug [ ].

Altered platelet numbers and function : Platelet dysfunction occurs dose-dependently with carbenicillin, ticarcillin, and, infrequently, other broad-spectrum penicillins [ ], but the NMTT cephalosporin moxalactam has also been associated with altered platelet function in both healthy subjects and in patients treated with standard regimens [ ]. In contrast, clinical studies including cefotaxime, ceftizoxime, cefoperazone, and ceftracone did not show platelet dysfunction attributable to these compounds [ ]. There is evidence that beta-lactam-antibiotic-induced platelet dysfunction is at least partially irreversible [ ].

From a practical point of view it can be concluded that:

• 1.

  The use of cephalosporins containing an NMTT side chain is associated with a risk of dose-dependent inhibition of vitamin K-dependent clotting factor synthesis.

• 2.

  Platelet dysfunction occurs primarily with the broad-spectrum penicillins, but the NMTT cephalosporins, notably moxalactam, have also been implicated; monitoring of bleeding time should be considered in patients at risk (bleeding history, clinical bleeding, concomitant thrombocytopenia, or the use of other drugs known to interfere with platelet function.

• 3.

  The presence of non-antibiotic factors, such as therapy with vitamin K antagonists or NSAIDs, renal insufficiency, hepatic dysfunction, impaired gastrointestinal function, and malnutrition, can increase the risk of bleeding in cephalosporin-treated patients; close monitoring of homeostasis (prothrombin time, bleeding time), as well as prophylactic supplementation with vitamin K or, if necessary, therapeutic administration of fresh-frozen plasma and/or platelets is warranted according to the clinical context.

Thrombocytosis

Thrombocytosis is frequently mentioned as an adverse effect of beta-lactam antibiotics. However, it has been suggested that this reflects healing from infection rather than toxicity [ ].

Antibiotic-induced diarrhea

There are three types of antibiotic-induced diarrhea:

• simple diarrhea due to altered bowel flora; this is quite common, for example it occurs in about 8% of patients who take ampicillin [ ];

• diarrhea due to loss of bowel flora and overgrowth of Clostridium difficile , with toxin production; this is much less common;

• a rare form of diarrhea that is due to allergy.

Almost all antibacterial agents have been observed to cause diarrhea in a variable proportion of patients [ , ]. The proportion depends not only on the antibiotic, but also on the clinical setting (in-patient/out-patient), age, race, and the definition of diarrhea. Severe colonic inflammation develops in a variable proportion of cases, and in some cases pseudomembranous colitis occurs [ ]. Since 1977, much evidence has accumulated that the most important causative agent in antibiotic-associated diarrhea is an anaerobic, Gram-positive, toxin-producing bacterium, C. difficile [ ].

Pseudomembranous colitis was known before the introduction of antimicrobial agents and can still occur without previous antibiotic use, for example after antineoplastic chemotherapy [ ] or even spontaneously. However, the number of cases has increased dramatically since antibiotics began to be used [ ]. Patients treated with lincomycin or clindamycin, cephalosporins, penicillinase-resistant penicillins, or combinations of several antibiotics are at especially high risk [ ]. A low risk is usually associated with sulfonamides, co-trimoxazole, chloramphenicol, and tetracyclines [ ]. Although few data have yet been published on this subject for the quinolones, they seldom seem to cause diarrhea and pseudomembranous colitis [ ].

Presentation : In pseudomembranous colitis the stools are generally watery, with occult blood loss, which is seldom gross. Common findings include abdominal pain, cramps, fever, and leukocytosis. Especially severe forms can run such a rapid course that diarrhea does not occur; they present with symptoms of severe toxicity and shock [ ]. As a rare complication, marked dilatation of the colon and paralytic ileus can develop, that is, toxic megacolon.

Pseudomembranes are described as initially punctuate creamy to yellow plaques, 0.2–2.0 cm in size, which may be confluent, with “skip areas” of edematous mucosa. Histologically they are composed of fibrin, mucous, necrotic epithelial cells, and leukocytes.

An acute colitis, different from pseudomembranous colitis, was observed in five patients taking penicillin and penicillin derivatives [ ]. There was considerable rectal bleeding. The radiographic findings were those of ischemic colitis (spasm, transverse ridging, “thumbprinting,” and punctuate ulceration). On sigmoidoscopy and biopsy, the mucosa was normal, except for an inflammatory cell infiltration in one case. Conservative treatment resulted in rapid remission.

Occurrence and frequency : Clostridium difficile has been isolated in 11–33% of patients with antibiotic-associated diarrhea, 60–75% of patients with antibiotic-associated colitis, and 96–100% of patients with pseudomembranous colitis [ , , ]. However, about 2% of the adult population are asymptomatic carriers [ ]. Primary symptomless colonization with C. difficile reduces the risk of antibiotic-associated diarrhea [ ]. Infants up to 2 years seem to be refractory to pseudomembranous colitis, although a high percentage may be carriers of C. difficile [ , ]. The reasons for this are unknown. It has been speculated that infants lack receptors for the toxin.

There have been several reports of frequent diarrhea in patients treated with combinations of ampicillin or amoxicillin with beta-lactamase inhibitors, such as sulbactam or clavulanic acid [ ]. A double-blind crossover study in healthy volunteers showed disturbances of small bowel motility after oral co-amoxiclav [ ].

The appearance of pseudomembranous colitis in clusters of patients [ ] may explain the wide variation in occurrence, and suggests that the disease may result from cross-contamination among patients rendered susceptible by antibiotic treatment. This is especially true for epidemic outbreaks in hospitals, where the disease may be considered a nosocomial infection favored by serious illness, frequent and prolonged use of broad-spectrum antibiotics (especially cephalosporins), and poor compliance with the rules of hospital hygiene [ ]. In such an epidemic, a variable proportion of patients will harbor the organism as asymptomatic carriers. An additional possible explanation for the large differences in reported frequencies may be the use of different methods of detection and differences in the definition of the disease. If colonoscopy was routinely performed in all patients with diarrhea taking clindamycin, pseudomembranous colitis was found in as many as 10% [ ].

Although the first antibiotics reported to cause pseudomembranous colitis were lincomycin and clindamycin, the disease was later described with all other antimicrobial drugs, even topically applied [ ]. Vancomycin [ ] and metronidazole [ ], which may be used as specific treatments, have also been implicated.

Susceptibility factors : Besides the type of antibiotic therapy, other factors such as the age of the patient, the severity of the underlying disease, colonic stasis, cytostatic therapy, surgical interventions, and gastrointestinal manipulations are predisposing factors for antibiotic-associated colitis [ ].

It is still not established if there is a correlation between toxin production or genotype of the C. difficile and the clinical manifestations of the infection [ , ]. Although hospital-acquired antibiotic-associated colitis is by far the major problem, community-acquired diarrhea associated with C. difficile has also been described [ ].

Mechanism : Clostridium difficile produces two well-characterized toxins [ , ]—toxin A, an enterotoxin, and toxin B, an extremely potent cytotoxin—which are thought to be responsible for the disease. The toxigenicity of toxins A and B varies between different strains of C. difficile and seems to correlate with symptomatic disease [ ]. Pseudomembranes were found in a higher percentage of patients with stools positive for cytotoxin than in patients whose stools were positive for C. difficile , but toxin-negative [ ]. Although there is also a high association with C. difficile (about 20% are toxin-positive) in antibiotic-associated diarrhea without pseudomembranes, it is possible that this microorganism plays no pathogenic role in some of these usually milder forms of the disease. In these cases the diarrhea may be due to impaired metabolism of carbohydrates, altered fatty acid profiles, or the composition and deconjugation of bile acids by quantitatively and qualitatively altered fecal flora [ , , ].

Diagnosis : The diagnosis of antibiotic-related colitis should be considered in any patient with severe diarrhea during or within 4–6 weeks after antibiotic therapy. The single best diagnostic procedure is sigmoidoscopy, although in a number of cases the typical pseudomembranous lesions may be seen only above the rectosigmoid area [ ]. Radiographic investigations (barium enema and air contrast) may show typical findings, but are dangerous in advanced cases and should be avoided. Computerized tomography showed typical but not pathognomonic patterns in two patients [ ].

Clostridium difficile can be cultured from the stool, and toxins A and B can be assessed by different techniques [ ]. The most accurate method is still a cytotoxin tissue culture assay. This detects the cytopathic effect of cytotoxin B, which can be neutralized by Clostridium sordellii antitoxin, but it takes 24–48 hours to show a result. Alternative tests that produce faster results have been developed. A latex agglutination test lacks sensitivity and specificity, and does not distinguish toxigenic from non-toxigenic strains. An enzyme immunoassay for toxin A may be an acceptable alternative to the cell cytotoxin assay and the results are rapidly available. A dot immunobinding assay has not yet been extensively studied [ ].

Management : Therapy consists of withdrawal of the antibiotic when diarrhea occurs and replacement of fluid and electrolyte losses. In less severe cases of antibiotic-associated diarrhea, no further treatment is needed. However, in patients with pseudomembranous colitis, a more intensive approach is usually required. When a toxic syndrome develops, fluid losses within the bowel can be very large. In these cases, a central venous line offers the chance to measure central venous pressure. Usually there is also loss of serum proteins and in some cases blood, which need appropriate replacement. In the rare cases with fulminant colitis and toxic megacolon, surgical intervention may be necessary [ , ].

In pseudomembranous colitis (typical endoscopic findings, positive test for C. difficile or its toxin), the preferred treatment is oral metronidazole, 250 mg qds or 500 mg tds [ , ]. Metronidazole is as effective as vancomycin 125–250 mg qds, which is significantly more expensive [ ]. Oral bacitracin 25 000 U qds [ ] and oral teicoplanin [ ] are acceptable alternatives.

Relapses are similarly frequent after treatment with metronidazole and vancomycin [ ]. In 189 adult patients, a first relapse occurred in up to 24% and a second relapse in 46% [ ]. Relapse may be due to sporulation of C. difficile and not to the development of resistance. Relapses usually respond to further courses of the initial treatment. Some alternative treatments have been proposed for repeatedly relapsing cases, including the combination of vancomycin with rifampicin for 10 days [ ].

The role of anion exchange resins (colestyramine and colestipol), which bind C. difficile toxin, is still controversial [ ]. If ion exchange resins are given at all, they should not be given together with vancomycin, because they also bind the antibiotic [ ]. Attempts to restore the intestinal flora with Lactobacillus GG [ ], or with fecal enemas [ ] from healthy volunteers have shown some favorable results in less severe cases. However, esthetic and infectious concerns may be an obstacle. It also has been suggested that treatment with Saccharomyces boulardii may help prevent the development of antibiotic-associated diarrhea [ ]. Its value in the prevention and treatment of relapses has still to be demonstrated. Antimotility agents have been associated with an increased incidence of antibiotic-related diarrhea and can worsen symptoms when the disease is already established [ ]. They should therefore be avoided.

There is little evidence that re-exposure to the same antibiotic that caused pseudomembranous colitis confers a further risk for relapse. Still, it would be wise to avoid the antibiotics that are most often related to pseudomembranous colitis in a patient who has had this complication.