Persistent staphylococcal bacteremia in an intravenous drug abuser.

A patient with methicillin-resistant Staphylococcus aureus bacteremia received vancomycin (MIC = 0.8 microgram/ml, MBC = 15 micrograms/ml) and heparin simultaneously through the same intravenous line to treat a septic deep venous thrombosis. Bacteremia persisted for 7 days. Bacteremia terminated when the simultaneous infusion of heparin and vancomycin through the same line was stopped. This suggested that an interaction between vancomycin and heparin may have occurred, which resulted in a reduction in vancomycin activity. To test for such an interaction, mixtures of heparin and vancomycin in various concentrations were made and tested for antimicrobial activity against the organisms in the patient. A precipitate formed at the concentrations achieved in the intravenous lines, and when the vancomycin concentrations were measured by bioassay, a 50 to 60% reduction in activity was noted. In contrast, when these solutions were prepared and mixed at microgram concentrations, a precipitate was no longer observed, and antimicrobial activity was not reduced. Heparin appeared to interact unfavorably with vancomycin at the concentrations in the intravenous lines when these drugs were administered simultaneously to patients. This may be the cause of poor therapeutic responses to vancomycin in some patients, especially those infected with tolerant organisms.

Among the causes of persistent or breakthrough bacteremia are a removable focus of infection, an incorrect choice or dosage of antibiotics, an undrained abscess, and an incorrect mode of drug administration. A case of persistent staphylococcal bacteremia was observed at our hospital. Persistence of the organism may have resulted from antibiotic inactivation because of simultaneous administration of heparin and vancomycin through the same intravenous line. This may have caused a chemical or physical reaction that resulted in some nullification of the antibacterial activity of vancomycin. This phenomenon has not been described previously for vancomycin and heparin, but Ragamey et al. have described decreased aminoglycoside activity when gentamicin was exposed to heparin in blood-sampling tubes (6). Vancomycin and heparin are known to be incompatible, probably because of a concentration-dependent acid-base reaction causing a precipitate in material (8) prepared in 5% dextrose for intravenous use.
Because of the high frequency of deep femoral vein thrombosis occurring with femoral vein drug injection, and also because of the frequency of methicillin-resistant Staphylococcus aureus bacteremia in drug abusers, heparin and vancomycin are often administered concomitantly. If the activity of vancomycin is decreased when mixed with heparin, a poor clinical outcome is possible for many of these patients. We report a patient in whom this may have occurred.

CASE REPORT
A 33-year-old male from Detroit was admitted to the Wayne County Hospital because of fever and right groin pain. The patient admitted to using heroin intravenously for the past 3 years; usually he injected it into the femoral veins. He often self-administered oral cephalexin concomitantly. Two days before admission he noted fever, shaking chills, and diaphoresis. He also noted swelling and tenderness of the right groin at the injection site. His temperature was 102°F (38.9°C), and he had a tender, indurated area overlying the right femoral triangle. No peripheral signs of endocarditis were observed, but there was a grade II/VI systolic ejection murmur noted at the lower left sternal border that increased with inspiration.
Since no peripheral venous access was available, a subclavian venous line was placed with a sterile technique. The patient was empirically treated with 1.0 g of vancomycin intravenously in 250 ml of 5% dextrose every 12 h (infused over 30 min) and with 90 mg of tobramycin intravenously every 8 h. Cultures of blood and of a subcutaneous aspirate from the right femoral triangle grew methicillin-resistant S. aureus. Tobramycin was discontinued once S. aureus was identified as the infecting organism (vancomycin MIC = 0.8 ,ug/ml, MBC = 15 ,ug/ml). Shortly after his admission, a venogram was performed and confirmed the suspicion of a deep venous thrombosis. A continuous infusion of heparin (25,000 U/day) in 1,000 ml of 5% dextrose was administered. Both heparin and vancomycin were infused through the same line. Fever persisted, and blood cultures remained positive for methicillin-resistant S. aureus for 7 days. No signs of cardiac vegetations were found on a twodimensional echocardiogram. No area of fluctuance developed in the right femoral triangle. The serum creatinine did not rise above 1.2 mg/dl during hospitalization. After 7 days another intravenous line was started, and the vancomycin and heparin were infused through separate lines at the identical dosage and frequency as before. Within 24 h the patient became afebrile, and subsequent blood cultures were negative (Fig. 1). The  completing a 4-week course of vancomycin and was ultimately lost to follow-up.

MATERIALS AND METHODS
The presumed alteration in the activity of vancomycin by heparin was tested by dissolving vancomycin in 250 nil of 5% dextrose to yield a concentration of 4 mg/ml; heparin was mixed with 1 liter of 5% dextrose solution to yield a concentration of 25 U/ml. These concentrations were equivalent to those in the intravenous line when vancomycin and heparin were administered to the patient sinmultaneously. A solution of vancomycin (4 mg/ml) Alone and a solution of heparin (25 mg/ml) plus vancomycin (4 mg/ml) were immediately diluted serially with human serum (AB positive) to provide various concentrations of each drug within the range of a bioassay systenm. The final calculated vancomycin concentrations ranged from 2.5 ,uglml to 100 ug/tnl, and the final heparin concentrations ranged from 0.01 to 0.625 U/ml. Serum was used to simulate the conditions when a mixture of vancomycin and heparin is infused intravenously. Aliquots (20-,uI) of the various dilutions were added to standard diffusion disks, which were dropped onto agar (antibiotic medium 5; Difco Laboratories, Detroit, Mich.) that had been inoculated with Bacillus subtilis. After plates were incubated for 4 h at 35°C, zone diameters were measured and compared with the zones around known concentrations of vancomycin (1).
When administered through separate intravenous lines, high concentrations of vancomycin and heparin do not mix. Each drUg is diluted within the intravascular space to much lower concentrations before their contact. Gradient plates were prepared for an assay of the vancomycin activity of the solutions that were initially prepared at these lower concentrations (2 to 25 ,ug/ml) expected in the serum of patients (7). Vancomycin, at concentrations of 2, 4, 5, 6, and 25 ,ug/ml, was mixed with heparin at a concentration of 0.5 U/ml, added to Mueller-Hinton agar, poured into inclined square plates, and allowed to harden. This mixture was prepared from solutions that were not permitted to mix at high concentrations. After returning plates to a horizontal position, plain Mueller-Hinton agar was added to level offthe top of each plate. Similar plates were prepared without heparin. Overnight broth cultures of three methicillin-resistant S. aureus strains (vancomycin MICs and MBCs were 1.6 and 1.6, 3.1 and 3.1, and 6.25 and 6.25 respectively) were streaked with a cotton-tipped applicator in the axis of the concentration gradient and inicubated at 35°C overnight. The physical length of the colony growth was measured and compared with the length of colonies in similar plates containing vancomycin alone. The MIC and MBC for all organisms were determined by the microtiter technique described by Harwick et al. (4).

RESULTS
When the mixture of high concentrations of vancomycin (4 mg/ml) plus high concentrations of heparin (25 U/ml) in 5% dextrose was diluted to concentrations within the range of the bioassay system (vancomycin at 2.5, 5, 10, 12.5, 25, 50, and 100 jig/ml and heparin at 0.01, 0.03, 0.06, 0.08, 0.16, 0.31, and 0.63 U/ml), the mean reduction in vancomycin activity was 42% (range, 28 to 56%). Vancomycin alone, prepared in a similar manner and diluted to the same 'r a E U %. A concentrations, showed no reduction in bioactivity (Table   1).
In contrast, the gradient plate assay showed no significant difference between the antibacterial effects of vancomycin alone in agar and of vancomycin mixed with heparin at much lower initial concentrations (2 to 25 ,ug/ml and 0.5 U/ml, respectively) (Fig. 2). Therefore, there was no observed effect of heparin on the bioactivity of vancomycin when the mixture was initially prepared at the concentrations expected in serum after infusion of the two drugs through separate intravenous lines.
At the higher concentrations that probably occur in intravenous tubing when the two drugs are infused simultaneously (1 to 5 mg of vancomycin per ml and 1 to 1,000 U of heparin per ml), a white precipitate was immediately noted in tubes containing both vancomycin and heparin. On the other hand, at concentrations attainable in serum (vancomycin at 5, 50, and 100 ,ug/ml and heparin at less than 1 U/ml), no precipitates formed.

DISCUSSION
The staphylococcal bacteremia persisting for 7 days in this patient was unusual and was presumably caused by inactivation of vancomycin by heparin in mixtures of the two agents in a single intravenous line. Korzeniowsky and Sande report that the mean duration of methicillin-susceptible staphylococcal bacteremia is 3 days (±1 day) in addicts with right-sided endocarditis treated with nafcillin and gentamicin (5). The diminished activity of vancomycin coupled with the relatively high MBC (tolerance) for this isolate may have prevented in vivo killing of the organism. Typical peak concentrations of vancomycin in serum after a 1-g intravenous dose are between 20 and 50 ,ug/ml (2). However, if a 50 to 60% reduction in the bioactivity of vancomycin occurs after reacting with heparin, the net bioactive vancomycin in the serum may be as low as 8 ,ug/ml. The usual MICs of vancomycin for methicillin-susceptible S. aureus range from 1.56 to 3.12 ,ug/ml (3). MICs and MBCs for methicillinresistant S. aureus in our laboratory are similar to those of methicillin-susceptible S. aureus, with MICs and MBCs ranging from 0.4 to 6.25 ,xg/ml. It is possible that the quantity of bioactive vancomycin was insufficient to kill the organism isolated from this patient. The vancomycin-heparin interaction probably occurred within the same intravenous tubing during administration of the compounds that were allowed to mix at high concentrations. Once concomitant administration of the two drugs through one intravenous line ceased and additional vancomycin was infused, bacteremia ceased. Unfortunately, concentrations of vancomycin were not measured in the serum when the two drugs were infused through the same line, so the precise effect of mixing the two drugs upon the bioactivity of vancomycin in the serum of our patient could not be determined.
The incompatibility of vancomycin and heparin prepared in dextrose-and-water solutions is well known. In 0.9o sodium chloride solutions, the two drugs are more compatible. It has been suggested that when the two drugs need to be administered through the same intravenous line, the solution should be prepared in 0.9% sodium chloride (8). Because of the frequent occurrence of deep venous thrombosis and methicillin-resistant S. aureus bacteremia in the intravenous drug abuser population, care should be taken to avoid administration of heparin and vancomycin through the same intravenous line. Vancomycin, in addition, is often administered to renal dialysis patients through a heparinized fistula. Studies are planned to determine if vancomycin is inactivated in this situation.
The gradient plate assay showed no significant effect of heparin on vancomycin activity at the usual serum concentrations. No alteration in the bioactivity of vancomycin would be expected, therefore, when heparin in infused through a separate intravenous line. Once diluted in the intravascular space after separate infusion of each drug, the concentration of both drugs is much lower, in the range tested by the gradient plate assay. If necessary, infusion of the two drugs through the same line could be done serially, with a 0.9% sodium chloride solution flushing the line between the two drugs to prevent mixing at high concentration.