Microbiological and Clinical Aspects of Infection Associated with Stenotrophomonas maltophilia

Taxonomy S. maltophilia is the only member of the genusStenotrophomonas (Stenos, Greek: narrow;trophos, Greek: one who feeds; monas, Greek: a unit, monad; i.e., a unit feeding on few substrates; andmalt, Old English: malt; philos, Greek: friend; i.e., a friend of malt). The genus was proposed in 1993 by Palleroni and Bradbury (313) after many years of debate regarding the appropriate taxonomic position of this organism. The type strain was isolated in 1958 by Hugh from an oropharyngeal swab from a patient with an oral carcinoma (184) and named Pseudomonas maltophilia (185, 186). Subsequently, Hugh and Ryschenkow reclassified “Bacterium bookeri,” which had been isolated from pleural fluid in 1943 by J. L. Edwards, asP. maltophilia (184). Pseudomonas melanogena isolated from Japanese rice paddies in 1963 was later recognized as Pseudomonas maltophilia (235), as were strains which had initially been characterized asPseudomonas alcaligenes by Komagata (184, 234). Similarly, strains of Alcaligenes faecalis as described by Ulrich and Needham (418) were subsequently reclassified by Hugh and Ryschenkow (186) as P. maltophilia. Later, the use of DNA-rRNA hybridization techniques revealed the presence of five rRNA homology groups in the genusPseudomonas, as it then was, and showed that the rRNA cistrons of the P. maltophilia type strain ATCC 13637 were most similar to those of three strains of Xanthomonas(314).

This information was used by Swings et al. in 1981 to propose thatP. maltophilia be reclassified in the genusXanthomonas as X. maltophilia (402). In addition, they cited several other factors to support their case. These included the guanine-plus-cytosine (G+C) content (P. maltophilia, 63 to 67.5%; Xanthomonas, 63 to 70%); comparative enzymology, particularly the absence of NADP-linked dehydrogenases; the occurrence of the same type of ubiquinones (P. maltophilia and Xanthomonas spp. both possess ubiquinones with eight isoprene units, whereas all otherPseudomonas strains possess nine units); and similar cellular fatty acid composition. A subsequent analysis of whole-cell proteins patterns (424) appeared to support this proposal. Additional evidence from isoelectric focusing studies of outer membrane esterase (87) confirmed that P. betle andP. hibiscicola were synonyms of X. maltophilia(383, 421). However, the proposed reclassification ofP. maltophilia as X. maltophilia did not meet with universal approval (312), and the controversy about the taxonomic status of this bacterium in the genus Xanthomonasremained unresolved (53).

The controversy was partly because some of the evidence put forward by Swings et al. (402) was subsequently contradicted by the findings of later studies. During an extensive DNA-rRNA hybridization study of the genus Pseudomonas, different melting temperatures were reported for the same 27 strains ofXanthomonas used in the proposal of Swings et al., despite the use of identical experimental conditions and the same X. campestris type strain as reference rRNA (96). The range for X. maltophilia was 76.5 to 78°C, almost overlapping the range for the genus Xanthomonas reported by Swings et al. (78.8 to 81.5°C) but further removed from that (80.0 to 81.5°C) reported later by De Vos and De Ley (96). In addition, Yang et al., using polyamine and fatty acid analysis (460, 461), demonstrated that X. maltophiliapossessed profiles distinct from other species within the genusXanthomonas. Furthermore, the possession of ubiquinones with eight isoprene units was shown not to be limited to X. maltophilia and other members of the genus Xanthomonasby Oyiazu and Komagata, who cited that they were found in several other pseudomonads including P. avenae and P. palleroniof rRNA group III (311).

These factors, along with supplementary evidence, were used to request a reinterpretation of the taxonomic position of X. maltophilia (425). This was supported by experiments with a Xanthomonas-specific 16S rDNA sequence as the primer for PCR (254). Xanthomonads were recognized by the presence of a single 480-bp PCR fragment; however, X. maltophiliastrains produced additional PCR fragments, leading the author to conclude that X. maltophilia does not belong in the genusXanthomonas.

Continuing dissatisfaction with the classification of this organism finally gave rise to the proposal in 1993 to create the new genusStenotrophomonas with S. maltophilia as the sole member (313). More recently, Nesme et al. confirmed the distinction between S. maltophilia and members of the genusXanthomonas by using restriction mapping of PCR-amplified 16S rRNA genes (293).

Recently, Drancourt et al. proposed a new species;Stenotrophomonas africana, which is biochemically identical to S. maltophilia except for its inability to assimilatecis-aconitate. Genotypic analysis, however, revealed only 35% DNA homology between the two species (104).

Morphology, Culture Characteristics, and Growth RequirementsCells of S. maltophilia are straight or slightly curved nonsporulating gram-negative bacilli that are 0.5 to 1.5 μm long. They occur singly or in pairs and do not accumulate poly-β-hydroxybutyrate as intracellular granules. They are motile by means of several polar flagella. The colonies are smooth, glistening, with entire margins and are white to pale yellow. Although Nord et al. (298) observed beta-hemolysis in 11 of 31 strains they examined, the bacterium is not generally considered to be beta-hemolytic. However, on blood agar there may be a greenish discoloration of the medium around confluent growth (149). On clear media, some strains may exhibit a brownish discoloration. This phenomenon has been attributed to a secondary chemical reaction among extracellular products and is more intense when these strains are incubated at 42°C (183) or cultured on media with a high tyrosine content (41).

S. maltophilia is an obligate aerobe. Growth does not occur at temperatures lower than 5°C or higher than 40°C and is optimal at 35°C. Margesin and Schinner reported the isolation from an alpine environment of a strain of the bacterium which was capable of growth at 10°C (258). Methionine or cystine is required for growth by most but not all strains (193, 260).

Biochemical CharacteristicsThe bacterium has been regarded as relatively metabolically inactive (395), but a variety of unusual substrates, including the aminoglycoside antibiotic streptomycin (127), may be metabolized (37, 39, 55, 191, 209, 244, 273, 291, 333,398) and strains of the bacterium have been investigated for their potential use as biodegradative agents. Selected biochemical characteristics and microbiological reactions of S. maltophilia are shown in Table 1.

Cellular Constituents and Antigenic StructureSomatic O antigens and flagellar H antigens have been described (186). A total of 31 O antigens have been identified, and these have been used to type S. maltophilia in epidemiological studies (220, 374, 375). The structures of seven of these antigens, including O3, which represents the most frequently occurring serotype, have been determined (98, 292,448-452). The repeating units of these antigens are all branched and contain between three and five saccharides including relatively uncommon moieties such as 6-deoxy-l-talose, 3-acetamido-3,6-dideoxy-d-galactose, and 4-acetamido-4,6-dideoxy-d-mannose. Cross-reactions with the O antigens of S. maltophilia and Brucella spp. have been reported (76). Cross-reactions withRenibacterium salmoninarum have also been observed (18,57), and Orrison et al. reported nonreciprocal cross-reactivity between Legionella pneumophila and S. maltophilia(309). Preparations of the bacterium are also capable of agglutinating mouse, rabbit, and human spermatozoa (26).

Moss et al. investigated the cellular fatty acid composition ofS. maltophilia and identified the presence of relatively large amounts (26% of the total) of a branched-chain 15-carbon fatty acid:13-methyltetradecanoic acid (282). In addition, they identified three additional branched-chain hydroxy-fatty acids, i.e., 2-hydroxy-9-methyldecanoic, 3-hydroxy-9-methyldecanoic, and 3-hydroxy-11-methyldodecanoic acids, which they did not detect in a range of other bacteria examined (281). They suggested that the uniqueness of the fatty acid profile of S. maltophiliacould be of value in the identification of the bacterium.

GeneticsLittle is known of the genetic structure of S. maltophilia. A gene of 1,362 bp encoding a 48.5-kDa chorionic gonadotrophin-like hormone has been sequenced (164, 166). This hormone bears an immunological resemblance to human chorionic gonadotrophin (hCG) (163, 165) (see below). The gene encoding the L1 β-lactamase of S. maltophilia(blaS) was cloned and expressed in Escherichia coli by Dufresne et al. (106) and has been sequenced by Walsh et al. (439). Very recently, the same group published the sequence of the L2 β-lactamase (440). The sequences of the alkA and alkB genes, which code for alkane hydroxylase and rubredoxin reductase, respectively, have also been determined (245). Of 18 strains of S. maltophiliaanalyzed in one study, 5 were shown to harbor plasmids (380), but further analysis was limited.

Habitat S. maltophilia is found in a wide variety of environments and geographical regions, including Antarctica (430), and occupies ecological niches both inside and outside hospitals. It has been isolated from a number of water sources including rivers, wells, a hypereutrophic lake, and bottled water (in which it may be responsible for altered organoleptic properties), and sewage (13, 167, 176, 183, 184, 185, 219, 288, 319). The bacterium has also been recovered from a variety of soil (44, 85,127, 400) and plant rhizosphere environments, including grasses, sugarcane and palms (202), from wheat (203), cabbage, rape, mustard, corn (86), beet (86, 243), bananas, cotton, beans, tobacco (183), rice paddies (235), citrus plants (1), orchids (447), irises (25), legume inoculants which use nonsterile peat as a carrier (306), and stored timber (108). Iizuka and Komagata reported the isolation of S. maltophilia from oil brines and other related materials from oil fields in Japan (191). S. maltophilia will also grow on loofah sponges (51).

The potential use of S. maltophilia as a biological control agent in rhizosphere environments and against agricultural pests has been explored (31, 32, 112, 173, 232, 239, 242, 264, 382). Given the broad range of phytopathogenic fungi inhibited by the bacterium, it is not altogether surprising that S. maltophilia has recently been reported to inhibit the growth of human fungal pathogens such as Candida spp. andAspergillus fumigatus (213). The proposal that the observed inhibition is due to pyrrolnitrin production (213) merits further investigation, although it has also been suggested that chitinolytic activity exhibited by the bacterium may be of importance in this respect (232). Recently, Jakobi et al. reported the production of maltophilin, a novel macrocyclic lactam agent with anti-fungal activity against saprophytic, phytopathogenic, and human isolates, by S. maltophilia(195). Food sources for the bacterium include frozen fish (186), milk (185, 201), poultry eggs (7,469), and lamb carcasses (381). In the laboratory, the organism has been isolated from contaminated cell cultures (395) and an isolate of S. maltophiliacontaminating Sephadex columns was reported to tolerate 1% sodium azide (48).

The bacterium has also been isolated from a wide range of nosocomial sources, which are listed in Table 2.

The possibility of S. maltophilia being found in the domestic setting, in contrast to hospitals, has received comparatively little attention. In their investigation of the prevalence of pseudomonads in the domestic environment of cystic fibrosis sufferers and of control families, Mortensen et al. recorded low rates of isolation (4 of 407 cultures) of S. maltophilia, although it should be noted that they did not use a selective medium specifically intended to recover this bacterium (280). This is significant, because isolation from environmental samples and other heavily contaminated materials can be enhanced by the use of selective media (44, 202, 216).

Although relatively little information is available on the ability ofS. maltophilia to survive drying, Rosenthal reported that whereas the bacterium was frequently isolated from wet environments such as sink traps, recovery from dry sites such as hospital floors was very uncommon (363). This is in agreement with the findings of Moffet and Williams (274), who cultured S. maltophilia from various nosocomial water samples but noted that isolation of the bacterium from dry surfaces of respiratory therapy equipment was comparatively rare. Hirai (177) investigated the ability of a single clinical strain of S. maltophilia to survive when inoculated onto cotton lint and glass plates. On the former, viable bacteria could not be identified 7 h after inoculation; whereas on the latter, viable organisms were recovered after the same period of incubation although less than 1% of the initial inoculum survived. The D value (the time required for 90% reduction of the initial inoculum) for survival on glass plates was 2.4 h. Although the presence of proteins, such as bovine serum albumin, increased the survival of other gram-negative bacteria, this was not observed with S. maltophilia(177).

Human CarriageFew studies have investigated the carriage rate of S. maltophilia in humans. Isolation from feces was first reported by Hugh and Ryschenkow in 1961 (186). Using a selective medium, von Graevenitz and Bucher recorded a fecal carriage rate of 10.9% (14 of 128 samples) in outpatients with diarrheal illness or their contacts (436). Rosenthal, however, was unable to isolate S. maltophilia from the rectal swabs of 50 ambulatory nursing home patients (363), although it should be noted that their isolation medium was of poorer selectivity than the medium of von Graevenitz and Bucher.

In a small series of 12 patients with hematologic malignancy, Kerr et al. (215) reported a fecal carriage rate of 33%. In contrast, only 2 of 69 (2.9%) subjects in a control group were fecal excreters of S. maltophilia. There have been few systematic studies of carriage rates at sites other than the lower gastrointestinal tract. Although carriage on the hands of health care workers during nosocomial outbreaks of S. maltophiliainfection has been noted (419, 433), Khardori et al. (220) were unable to isolate the bacterium from skin swabs taken from hospital personnel. S. maltophilia has been isolated from the oropharyngeal swabs of healthy adults (183), although another study examining 200 throat or pharyngeal washings from healthy individuals participating in a health-screening program did not reveal the presence of the bacterium in any sample (363). Similarly throat swabs from hospital staff taken during the investigation of the nosocomial outbreak, reported by Khardori and colleagues, did not yield the bacterium (220). Respiratory tract carriage rates in individuals with cystic fibrosis are discussed in the section on respiratory tract infection, below. Carriage rates of 0% were recorded for 50 skin swabs of patients attending a cancer-screening clinic and for 50 vaginal swabs taken from gynecology clinic attendees (363).

Even less is known about animal sources of S. maltophilia. The bacterium has been isolated from fish, raw cows’ and ewes’ milk, rabbit feces (186, 248), lizards, frogs (183), the mouths and cloacae of captive snakes (105), and the gastrointestinal tracts of laboratory animals (135, 410). The bacterium has also been implicated in fleece rot of sheep (253). Lysenko and Weiser reported the recovery of S. maltophilia from the nematode Neoaplectana carpocapsae(252).

Following the isolation of S. maltophilia from the ovarian fluid of a sow, Richert and Ryan (353) and others (2) demonstrated that the bacterium was capable of binding hCG. Subsequently, it was shown that an hCG-like protein (xCG) fromS. maltophilia has a high degree of homology with the β subunit of hCG (164). An autocrine and/or paracrine function for xCG has been suggested following observations that when added to growth media, it (and hCG) causes changes in both the cell growth cycle and bacterial morphology (65). The nucleotide sequence of the chorionic gonadotropin-like protein receptor has been published (164, 166).

Virulence FactorsLittle is known of the putative virulence factors of S. maltophilia. The failure to distinguish between colonization and infection has fostered the belief that S. maltophilia is an organism of very limited pathogenicity. These views may have been reinforced by studies in which cases of infection with the bacterium were not associated with an unfavorable clinical outcome. For example, in their review of 106 intensive care unit patients infected or colonized with the bacterium, Villarino et al. were unable to attribute the death of any patient directly to S. maltophiliainfection (433). Gardner et al. (145) suggested that S. maltophilia was often associated with clinically overt infection only when acting synergistically with other pathogens. Early attempts to induce infection in an experimental animal model appeared to support the hypothesis that S. maltophilia does not cause serious sepsis when whole bacteria are administered intravenously to mice.

Although extracts of bacterial cultures were associated with some toxicity, it was less severe than that in similar preparations of strains of E. coli (160). Elaboration of a range of extracellular enzymes by S. maltophilia, including DNase, RNase, fibrinolysin, lipases, hyaluronidase, protease, and elastase, was described in several early reports (9, 42, 43, 52, 196, 298,302, 303, 377), and Bottone et al. have proposed that these enzymes may play a role in the pathogenesis of S. maltophilia-associated infection (52). They reported a case of ecthyma gangrenosum in a leukemic patient with S. maltophilia bacteremia. Because bacterial production of protease and elastase is deemed important in the pathogenesis of cases of ecthyma gangrenosum associated with P. aeruginosasepticemia, they examined the S. maltophilia isolate for elaboration of these enzymes and found that it was an “avid protease and elastase producer” (52). A larger study of 52 strains of clinical and environmental origin examined the production of nine extracellular enzymes. No strain produced lecithinase, hyaluronidase, or chondroitin sulfatase, whereas all the strains produced protease and elastase. Elastase production was variable, but there was no significant difference between clinical and environmental strains. Although DNase, lipase, and fibrinolysin were produced by all strains at 20°C, clinical isolates were significantly more likely to produce these enzymes when grown at 37°C (409).

Adherence to plastic is considered an important property of bacteria commonly implicated in line-related colonization and infection, and strains of S. maltophilia of both clinical and environmental origin have been reported to adhere to several types of plastic materials including intravenous cannulae (214).

Recently Jucker et al. investigated the adherence of a strain ofS. maltophilia to both glass and Teflon (200). They noted that adhesion to these negatively charged materials was promoted considerably by the positive charge manifested by this strain at physiologic pH.

The ability of S. maltophilia to survive and multiply within total parenteral nutrition and other types of intravenous infusates may also contribute to the pathogenesis of intravenous line-related infections (266, 323). Similarly, the ability of this bacterium to grow in dialysis fluids with the release of low-molecular-weight pyrogens, which are apparently different from endotoxin-like moieties as detected in the Limulus amebocyte lysate assay, may be important in the pathogenesis of pyrogenic reactions during hemodialysis (138).

The role of a cell wall-associated immunoglobulin G binding protein in the pathogenesis of S. maltophilia infection (162) requires further investigation. Serum resistance is a phenomenon frequently exhibited by gram-negative bacteria associated with septicemia, and a study of a small number of strains of S. maltophilia noted that clinical isolates were much more likely to exhibit this property than were environmental isolates (217).

Virulence factors of importance in the pathogenesis of S. maltophilia-associated pulmonary infection remain to be characterized. Grant et al. (158) were unable to demonstrate adherence of the bacterium to cultured hamster tracheal cells.

Laboratory Isolation S. maltophilia can be isolated quite readily from normally sterile sites on a wide range of commonly used solid media, including blood, heated blood, MacConkey, and cysteine lactose electrolyte-deficient agars. It will grow overnight at temperatures ranging from 20 to 37°C in air and will also grow in 5% CO₂. Most blood culture systems are capable of growingS. maltophilia with various degrees of efficiency. Published evaluations are summarized in Table 3.

It is difficult to determine the relative merits of the various systems with respect to isolation of S. maltophilia because of the small numbers of strains in many studies. It should be noted, however, that S. maltophilia has been isolated from terminal subcultures (false negative) taken from a number of systems including Oxoid Signal (396), BACTEC NR 660 (297), BacT/Alert (357), and Difco ESP (278), although on most occasions other bottles within each set were positive. In one study, the Isolator system detected significantly more positive cultures (P < 0.01) and more sepsis episodes (P < 0.02) with S. maltophilia than did aerobic bottles of the Difco ESP system (227).

Selective MediaA number of selective media have been developed for the isolation of S. maltophilia from clinical or environmental specimens likely to be contaminated with other bacteria. Early examples, such as mineral acetate lactate methionine agar were of relatively poor selectivity (363). Antimicrobial agents were added to later medium formulations in an attempt to improve this. Juhnke and Des Jardins developed Xanthomonas maltophilia selective medium for the isolation of S. maltophilia from soil and rhizosphere environments (202). This medium contains six antibacterial agents, i.e., cephalexin, bacitracin, penicillin G, novobiocin, neomycin, and tobramycin, and two antifungal agents, i.e., nystatin and cycloheximide. Maltose and bromothymol blue were also included to facilitate the identification of S. maltophiliacolonies, which, because of acid production from maltose, appear as orange with a yellow halo (202). In an investigation into environmental reservoirs and hand carriage of S. maltophiliaby health care personnel, Villarino et al. used tryptic soy agar supplemented with 5% sheep blood to which 5 μg of gentamicin per ml had been added, although no comparator medium was tested (433).

Several media which take advantage of the inherent resistance ofS. maltophilia to carbapenems have also been developed. Imipenem added to blood or MacConkey agar has been used to isolateS. maltophilia from stool (436) and sputum (154) samples. MacConkey agar which incorporated imipenem at a concentration of 10 μg/ml was more inhibitory for some strains than was imipenem-blood agar (436). Khardori et al. (220) used MacConkey agar supplemented with 25 μg of imipenem per ml, which might have contributed to the low rate of isolation of S. maltophilia from environmental sources during their investigation of a nosocomial outbreak. The imipenem-containing medium of Bollet et al. (44) incorporates methionine, thus taking advantage of the requirement of most strains of the bacterium for this amino acid (193,260). Media which contain imipenem as the sole selective agent may fail to inhibit the growth of some organisms found in stool cultures, particularly Enterococcus faecium andCandida spp. These problems can be circumvented by the further addition of vancomycin and amphotericin B to imipenem-containing agar (216) and the use of a mannitol-bromothymol blue indicator system to facilitate the differentiation of S. maltophilia (which does not produce acid from mannitol) from other imipenem-resistant gram-negative bacteria. This medium was shown to enhance the isolation of S. maltophilia from sputum samples taken from patients with cystic fibrosis (92) and the feces of patients with hematological malignancy (216).

Identification in the Clinical Laboratory S. maltophilia can be identified in the clinical laboratory by standard biochemical and other tests (151) (Table 1). In addition, Townsend et al. suggested that in vitro synergism between aztreonam and clavulanic acid by using disks containing 30 and 10 μg of drug, respectively, can be used in the presumptive identification of the bacterium (412). Furthermore, several commercial products which are intended to aid in the identification of gram-negative non-fermentative rods have been tested for their ability to accurately identify S. maltophilia (19, 70, 110, 205, 228, 229, 263, 271, 304, 332,352, 356, 385, 387, 394, 407, 435, 441). Details of studies which have examined the utility of these methods, including automated and semiautomated systems, for the identification of this bacterium are listed in Table 4.

The limited number of S. maltophilia strains used in many of these evaluations (e.g., the study of Varettas et al., in which a single strain was used [426]), however, precludes definitive conclusions about the relative abilities of each system with respect to correct identification of this particular organism. Where comparative studies have evaluated automated and nonautomated systems, there have been no significant differences in their ability to identifyS. maltophilia (304, 356).

Given its erstwhile relative infrequency of isolation from clinical material and uncertain taxonomic status, it is perhaps not surprising that several early reports noted that the identity of S. maltophilia could be confused with that of other bacteria and that it had been misidentified as Pseudomonas alcaligenes(192), Bordetella bronchiseptica, andAlcaligenes faecalis (186). Later reports, however, emphasize that misidentifications still occur. One strain ofPseudomonas stutzeri was misidentified as S. maltophilia by the Radiometer Sensititre AP80 (394), an isolate belonging to Flavobacterium group IIB was misidentified by the Vitek AutoMicrobic system (385), and two isolates were misidentified as Pseudomonas diminuta by the PASCO MIC-ID system (352).

Misidentification of S. maltophilia can have important clinical and psychosocial implications. In a survey of 32 “Burkholderia cepacia” strains isolated from cystic fibrosis patients sent to the Canadian Pseudomonas Repository Laboratory, 3 were subsequently identified as S. maltophilia(59). Misidentifications of this type are of particular concern since 20% of cystic fibrosis patients colonized with B. cepacia are known to suffer a syndrome of pneumonitis, septicemia, and rapid clinical deterioration and thus its isolation is understandably viewed with great alarm by these patients and their care givers.

Conversely, some systems have misidentified other organisms as S. maltophilia. In an evaluation of the RapID NF Plus, one strain ofB. cepacia needed additional biochemical tests before it could be correctly distinguished from S. maltophilia(229). In addition to methods which rely on the biochemical reactions for the identification of S. maltophilia, several other techniques have undergone evaluation. These include gas-liquid chromatography (240), direct-probe mass spectrometry (343), and molecular methods. Zechman and Labows (468) used automated headspace concentration gas chromatography to identify S. maltophilia and were able to distinguish the bacterium from P. aeruginosa, B. cepacia, P. fluorescens, and P. putida. Veys et al. correctly identified 59 strains of S. maltophilia by the presence of three characteristic peaks consisting of 3-hydroxydodecanoate, 13-methyltetradecanoate, and 12-methyltetradecanoate (431). An automated gas-liquid chromatography system, the Microbial Identification System (Microbial ID, Newark, Del.), correctly identified 45 of 47 (96%) S. maltophilia strains tested, with the others being misidentified asXanthomonas campestris (310).

Molecular methods for the identification of S. maltophiliahave, to date, received comparatively little attention. By using single-strand conformation polymorphism electrophoresis of PCR-amplified 16S rRNA fragments, species-specific patterns forS. maltophilia have been observed (445). Tyler et al. (417) were able to distinguish between S. maltophilia and seven species of Pseudomonas by identifying differences in the sequences of 16S to 23S rDNA internal transcribed spacers.

Typing SchemesA variety of typing schemes have been developed in an attempt to elucidate the epidemiology of infections caused by S. maltophilia, and they have been used, with various degrees of success, to investigate nosocomial outbreaks. Phenotypic methods include serotyping based on heat-stable O antigens (220, 374, 375,433) and lipopolysaccharide typing (12).

Typing based on comparison of antibiograms is of limited applicability because of a lack of variation in resistance patterns expressed by most strains (419). Similarly, biotyping is poorly discriminative because of a paucity of interstrain variability manifested by the bacterium. More recently, genotypic typing techniques have been the subject of much attention. Methods evaluated have included ribotyping (36, 114, 147), pulsed-field gel electrophoresis (PFGE) (241, 369, 405, 419, 462), random amplification of polymorphic DNA (RAPD) (67, 84, 419, 462), and enterobacterial repetitive intergenic consensus PCR (ERIC-PCR) (67). Other techniques which have been used to study the epidemiology of S. maltophilia are pyrolysis mass spectrometry (308) and multilocus enzyme electrophoresis (376).

An outbreak involving 45 patients on a shock-trauma intensive care unit was investigated by serotyping (433). Of 22 patient isolates available for typing, 19 (86%) were of serotype 10. A total of 11 of 17 samples from inanimate environmental sites (including respirometers, a tracheal suction catheter, components of a ventilation circuit, and water from a sink) and 3 of 6 hand cultures from nurses and respiratory therapists were positive for S. maltophilia. Of these 14 isolates, 13 were also of serotype 10. Reemphasis of the importance of hand washing before and after all patient contact, implementation of secretion and drainage precautions for all patients with S. maltophilia would infection, the use of gloves when suctioning or manipulating the respiratory tracts of ventilated patients, and high-level disinfection of respirometers between use in different patients were the measures used to successfully bring an end to the outbreak.

Serotyping was also used to analyze 52 clinical isolates collected from 35 cancer patients over a 6-month period (220). Again, serotype 10 was the most common type (16 of 52 isolates [31%]). Eight other types were found among the remaining 36 isolates. Three environmental isolates were found on the medical intensive care unit, two of which were from water faucets and one was from a water sample. One of these belonged to serotype 10. The number of S. maltophilia isolates decreased following the earlier introduction of appropriate antibiotic therapy for S. maltophiliainfection and an increase in the temperature of the boiler for hot water from 40 to 43.3°C. In a more recent study, Schable et al. examined 900 isolates from 10 countries and noted that 795 (88.3%) were serotypable. The most common serotypes were 10, 3, and 19 respectively, and they were predominantly associated with respiratory and blood isolates (375). This typing system is, however, of limited applicability because reagents are not widely available and because, as shown in the study of Schable et al., more than 10% of isolates are untypable. In addition, this method appears less discriminatory than other techniques (376).

The technique of pyrolysis mass spectrometry relies on the analysis of the products of thermal degradation of materials (337). It was used to investigate a suspected outbreak of S. maltophilia cross-infection and colonization on a heart-lung transplant intensive care unit (308). The respiratory tracts of five patients became colonized or infected over a 19-day period. All five had been nursed in either of two single cubicles equipped with positive-pressure ventilation. Extensive environmental sampling failed to reveal a source of S. maltophilia, although unauthorized reuse of “single-use” disposable nebulizers was uncovered. These had been cleaned but not resterilized. Twelve “cleaned” nebulizers were sampled prior to their intended reuse, and S. maltophilia was cultured from one of them. Although the five patient strains and the nebulizer strain were all different, the authors considered the outbreak to be related to reuse of disposable nebulizers, and the outbreak ceased when this practice was terminated.

Ribotyping has been used in the investigation of cases of nosocomialS. maltophilia bacteremia which occurred over 13 months in seven patients residing on four different wards of a pediatric hospital (36). Four patients (three on a gastroenterology ward, one on the intensive care unit) had the same ribotype, although the mode of transmission between these patients was not readily apparent. Ribotyping was also used to characterize 77 consecutive clinicalS. maltophilia isolates in a Danish hospital (147). Great diversity was found among the ribotypes of the hospital strains, and no single-strain outbreak was detected. Potential reservoirs for these strains were not determined. Wust et al. (458) used ribotyping to demonstrate that increasing antimicrobial resistance in serial isolates of S. maltophilia from a patient with cystic fibrosis was due to acquisition of resistance by the original strain rather than infection by other strains.

Several reports evaluating the usefulness of PFGE have been published. Thirty S. maltophilia strains from three countries (Brazil, Switzerland, and the United States) were analyzed by PFGE ofXbaI or SpeI digests (369). With the exception of four strains from Brazil collected during an outbreak on a dialysis unit, all the strains were unrelated. Additional details relating to the Brazilian outbreak were, unfortunately, not given.

Ten strains from five patients and two environmental strains relating to a S. maltophilia outbreak on a hematology unit were examined by PFGE following chromosomal DNA digestion withDraI (405). The five patients each had unique strains; however, the environmental isolates, one from a kitchen water faucet and one from a shower head, each matched the profile of one of the patient isolates.

PFGE following DraI digestion was also used to analyze 109 clinical strains collected over a 10-month period in a Canadian hospital (462). This revealed 62 unique DNA restriction profiles. Respectively, 23, 11, 6, 6, and 3 strains had concordant profiles in each of five types. Four profiles were seen among 8 strains (2 strains with each profile), and the remaining 52 strains had unique profiles. Two profiles accounted for 32 (61.5%) of the 52 isolates from the intensive care units but only for 2 of 31 ward isolates. This was suggestive of nosocomial transmission of these two strains on the intensive care units but no details of any further investigations were given.

PFGE of XbaI or SspI digests of chromosomal DNA was used to examine 64 clinical S. maltophilia isolates collected from 60 patients and the hands of one nurse over a 9-month period (419). Eight of the patients and the nurse were involved in an outbreak of S. maltophilia on an intensive care nursery. Although isolates from six of the eight patients had the same profile, the isolate from the hands of the nurse was unique. Turner-Hubbard et al. (416) used SpeI-based PFGE to demonstrate that isolates from bronchoalveolar lavage fluid were indistinguishable from those from bronchoscopes which had undergone defective sterilization procedures. More recently, Fabe et al. used this technique after XbaI or DraI digestion to examine strains from a hematology unit. They observed great genomic diversity among the strains analyzed but reported that person-to-person spread had occurred between several patients (117). An investigation into pulmonary colonization in cystic fibrosis patients by PFGE failed to demonstrate evidence of a common source of exposure or person-to-person spread among five patients (437).

RAPD, also known as arbitrarily primed PCR and multiple arbitrary amplicon profiling, has also been reported to be a useful tool for typing S. maltophilia. Its advantage over PFGE is that it is a more rapid and less labor-intensive method. However, there have been disagreements about which is the more discriminatory (420,462).

RAPD as a single typing method has been used to type 130 S. maltophilia strains (51 from the hospital environment, 48 clinical, and 31 from other environments) collected over an 8-month period in a French hospital (84). Sixteen cases where pairs were indistinguishable by RAPD profiles were identified: nine clinical, five from the hospital environment, and two from soil. Some evidence of cross-infection between hematology and intensive care unit patients was identified, but there were no matches between environmental and clinical strains.

ERIC-PCR has also been used to successfully type S. maltophilia (67). A total of 38 clinical strains were examined, 9 from a burn unit, 20 epidemiologically unrelated isolates, and 9 from the same cystic fibrosis patient. The burn unit strains and the 20 other unrelated strains were independent. The nine strains collected from the sputum of the cystic fibrosis patient over a 22-month period were, however, indistinguishable. The authors compared ERIC-PCR with RAPD. Both yielded rapid, reproducible, and discriminatory fingerprints of S. maltophilia. However, the authors were of the opinion that interpretation of results was easier with ERIC-PCR.

Schable et al. used multilocus enzyme electrophoresis, in conjunction with serotyping, to investigate an outbreak of S. maltophilia infection on a shock-trauma intensive care unit (376). Although most of the outbreak strains belonged to serotype 10, strains of this serotype unrelated to the outbreak yielded different electrophoretic types on multilocus enzyme electrophoresis (376).

Despite the number of reported outbreaks in which there was identification of clusters of strains indistinguishable on typing (36, 369, 419, 433, 462), exact sources and modes of transmission of S. maltophilia have proved difficult to elucidate. The great diversity of strains identified within given institutions (84, 147), the relative frequency of isolation from a wide variety of environmental sites, and the culture from some of these sites of strains indistinguishable from associated clinical strains (405) suggest that multiple independent acquisitions from environmental sources could be an important mode of transmission of S. maltophilia.

BacteremiaBacteremia is a common manifestation of S. maltophiliainfection (116, 134, 197, 220, 222, 238, 241, 279, 284, 285, 301,432) and appears to be increasing in frequency (238). Bacteremia may be secondary to a pulmonary, urinary, or gastrointestinal source (8, 388, 467, 470) although it is often not possible to determine the portal of entry. The importance of intravascular devices in the pathogenesis of S. maltophiliabacteremia is, however, becoming increasingly recognized (10, 28,78, 115, 116, 138, 279, 284, 432), and in cases where there is no obvious source of infection it has been suggested that these devices may be the primary source. For example, Muder et al. report that in their series of 91 patients with S. maltophilia bacteremia, 56% did not have a clinically apparent portal of entry but 84% of these individuals had a central venous catheter in place (284). In several instances, it has been possible to identify environmental reservoirs from which S. maltophiliawas likely to have been acquired. Flaherty et al. (130) investigated an outbreak of gram-negative bacteremia, including three cases caused by S. maltophilia and an additional case from which both S. maltophilia and Enterobacter cloacae were isolated from blood cultures. The source was identified as contaminated O-rings in reprocessed hemodialyzers. High-level O-ring contamination during simulated dialysis was demonstrated by Bland et al. to produce blood compartment contamination with S. maltophilia (40). Inadequate disinfection during reprocessing of capillary dialyzers was also implicated in another outbreak of S. maltophilia bacteremia (422). A cluster of cases reported by Fisher et al. (129) traced the source to contaminated pressure monitoring systems during open-heart surgery.

It is important to distinguish true S. maltophiliabacteremia from pseudobacteremia (197) arising from contamination of blood cultures when samples used to fill tubes for coagulation studies (which contain nonsterile anticoagulants such as sodium citrate) are then used to inoculate blood culture bottles (8, 378). This may be a more frequent occurrence than is generally recognized (8).

Relapsing S. maltophilia bacteremia has been described (36, 115, 285). Molecular typing techniques demonstrated that blood culture isolates from a patient who had experienced three episodes of bacteremia were indistinguishable.

Case/fatality ratios associated with S. maltophiliabacteremia may be high (36, 115, 285). In a series of 32 cases, Jang et al. noted that 22 (69%) were associated with a fatal outcome; 13 (41%) of these were deemed to be directly attributable to septicemia (197). Some investigators have also reported that polymicrobial bacteremia with S. maltophilia is a frequent occurrence (115). In their study, Jang et al. (197) noted that 44% of 32 cases of bacteremia were polymicrobial in origin. According to some investigators, the acute mortality rate associated with these polymicrobial bacteremias is not significantly different from that associated with bacteremia due toS. maltophilia alone (115, 197), although this is in contrast to the reports of others (238). S. maltophilia septicemia may be complicated by disseminated intravascular coagulation and purpura fulminans (208, 267). Ecthyma gangrenosum is also a well-recognized complication (52) (see the section on skin and soft tissue infection, below).

EndocarditisThere have been several reports of S. maltophiliaendocarditis (15, 100, 115, 128, 168, 180, 197, 285, 378, 389,399, 456, 464, 467, 470). Most of the cases have occurred in intravenous drug abusers or as a complication of prosthetic valve surgery. In the latter, infection often arises early in the postoperative period, possibly due to contaminated vascular access devices (115), although late-onset infections have been described (100, 115). Semel et al. reported a patient being treated for streptococcal prosthetic valve endocarditis who developedS. maltophilia superinfection, which they attributed to reflux of contaminated anticoagulant from collection tubes into the patient’s venous system (378). Endocarditis in the absence of drug abuse or cardiac valve abnormalities is uncommon (115,389). For example, Gutierrez Rodero et al. described a case of endocarditis following infection of a ventriculoatrial cerebrospinal fluid (CSF) shunt in a patient with otherwise normal native cardiac valves (168). The prognosis of S. maltophiliaendocarditis is variable. Although favorable outcomes with antimicrobial therapy alone have been recorded (168), surgical intervention to replace the infected valve may be necessary (467). The infection may be complicated by valve ring or myocardial abscesses and septic emboli (197, 285, 467).

A variety of other cardiovascular manifestations of S. maltophilia infection have been reported, including pericarditis (115, 470), infection of the pseudopericardial sac in an artificial heart recipient (101), and bacteremia in association with mitral valve prolapse (289).

Respiratory Tract Infection S. maltophilia has been reported to account for 5% of nosocomial pneumonias (3), although several large studies of the etiology of nosocomial pneumonia have failed to find any cases associated with this bacterium (77, 351, 411). Nosocomial pneumonia has been observed during outbreaks of S. maltophilia infection, with several cases occurring over relatively short periods (220, 241, 308, 433).

The respiratory tract is the most common site of isolation of S. maltophilia in hospitalized patients, accounting for the origin of 56 to 69% of isolates (84, 147, 241, 279), although the majority of patients (53 to 71%) with S. maltophilia-positive respiratory tract cultures are colonized rather than infected at this site (220, 241, 433). More rigorous diagnostic criteria are necessary, therefore, if the distinction between colonization and true infection is to be made (137).

S. maltophilia nosocomial pneumonia is associated with mechanical ventilation, tracheostomy, previous exposure to broad-spectrum antibiotics, the use of respiratory tract equipment such as nebulizers (116, 220, 233, 241, 279, 308, 433, 470), and therapy with aerosolized polymyxin (122). Many patients also have preexisting lung conditions such as chronic obstructive pulmonary disease, bronchiectasis, kyphoscoliosis, or endobronchial obstruction (241, 247, 470). Pulmonary infection following lung transplantation has also been reported (465).

Respiratory tract involvement with S. maltophilia is associated with significantly increased mortality (279). Kollef et al. (233) reported that the isolation of a “high-risk” pathogen such as S. maltophilia was the most important predictor of mortality in late-onset ventilator-associated pneumonia. The mortality rate in neutropenic patients with pneumonia is 40% (220).

The source of S. maltophilia in these patients is often unclear (172). Although ventilator circuits have yielded the organism when sampled (255, 308, 433) this may have represented contamination from the patient rather than the source itself. Klick and Du Moulin (230) reported a cluster of cases on an intensive care unit in which an oxygen analyzer contaminated with S. maltophilia was implicated, and Korn et al. also described an outbreak associated with reusable in-line ventilator temperature probes (236). In their study, Turner-Hubbard et al. attributed the isolation of S. maltophilia from bronchoalveolar specimens to faulty bronchoscope sterilization protocols (416). Interestingly, they did not observe clinically overt infection as a result of using contaminated bronchoscopes.

Although several extensive studies of the etiology of community-acquired pneumonia have failed to identify a single case in which S. maltophilia was the cause (22, 54, 119, 120,206, 261), a few cases of community-acquired pneumonia have been described (372); these mostly occurred in patients with predisposing conditions such as rheumatic heart disease and bronchiectasis (194, 388). The case reported by Irifune et al. is noteworthy because the isolate exhibited a mucoid phenotype (194). Gardner et al. (145) reported a case ofS. maltophilia bacteremia secondary to aspiration pneumonia.

The first reported isolation of S. maltophilia from the respiratory tracts of patients with cystic fibrosis was in 1975 in Denmark by Frederiksen et al. (133). The prevalence ofS. maltophilia in this patient group has been increasing, but variation between centers is well recognized. The prevalence rose from <2% in 1975 to 19% in 1993 in Denmark (133) and from 0% in 1983 to 10% in 1990 in one British center (154). It is now 19% in another British center (93) and over 30% in one Spanish center (16). The prevalence has generally been lower in the United States, ranging from 1.8 to 8.7% (89,283). The reasons for this increase are not clear, but one study suggested that the amount of antipseudomonal antibiotics with which these patients were treated and the duration of hospitalization were important risk factors (154). S. maltophilia may persist for long periods in the sputum of patients colonized with the bacterium (24). Recently, in a comparison of four different methods for preculture treatment of sputum samples from cystic fibrosis patients, Hoppe et al. found no significant effect on S. maltophilia isolation rates (181), although the use of selective media may affect the ability of clinical laboratories to demonstrate the presence of the bacterium in sputum (92).

There is no evidence to date of any nosocomial transmission of S. maltophilia to cystic fibrosis patients (94, 437).

The prognostic significance of S. maltophilia in the respiratory tract of cystic fibrosis patients is unclear. Some studies have shown no evidence of an adverse clinical outcome (89), whereas others have reported progressive deterioration in pulmonary function, particularly in patients chronically colonized for long periods with bacterial counts of S. maltophilia in excess of 10⁵ to 10⁶ CFU/ml of sputum (16, 207,231). For a more detailed discussion of S. maltophiliain cystic fibrosis, the reader is referred to the recent review by Denton (91).

Instances of true infection of the upper respiratory tract appear very rare. The clinical significance of the isolation of S. maltophilia from pharyngeal swabs (185, 186, 401) is uncertain. Harlowe reported a single case of S. maltophiliaacute mastoiditis in a patient who developed acute otitis media after swimming (170).

Central Nervous System InfectionMeningitis caused by S. maltophilia is uncommon (183). In neonates and infants, the infections appear spontaneous in onset (90, 373), whereas those in adults are secondary to neurosurgical procedures (153, 285, 295, 415). The presence of foreign material such as CSF shunts, ventriculostomy tubes, and Ommaya reservoirs may be important in the pathogenesis of these infections (295). Additionally, a spinal epidural catheter was deemed to have predisposed a patient to the development of a spinal epidural abscess (438). Adult cases of S. maltophilia meningitis in the absence of antecedent neurosurgery are very rare (327). The recently proposed species, S. africana, as described by Drancourt et al., was isolated from the CSF of a human immunodeficiency virus-seropositive Rwandan refugee with clinical signs of meningoencephalitis (104).

Ophthalmologic InfectionOcular infections associated with S. maltophilia appear to be increasing in incidence (331). A variety of ophthalmologic syndromes have been reported, including conjunctivitis, keratitis, dacryocystitis, and preseptal cellulitis (28, 331, 373,386, 390, 401). Conjunctivitis, keratitis, and corneal ulcers in contact lens wearers from whom the bacterium has been isolated have also been described (237, 331, 393). Chen et al. (68) reported a case of endophthalmitis following the implantation of a sustained-release preparation of ganciclovir. Endophthalmitis has also been described following cataract extraction (204). S. maltophilia has been isolated from contact lenses or lens care paraphernalia (73, 103, 211,331), and it has been suggested that synergy between S. maltophilia and Acanthamoeba spp. may be important in the pathogenesis of contact lens-associated amebal keratitis (49,50).

Urinary Tract InfectionAlthough S. maltophilia has been frequently isolated from urine specimens, early reports acknowledged that there was insufficient evidence to support an unequivocal role for this bacterium as a urinary tract pathogen (145, 150, 180, 287, 401). Therefore, it has come to be considered an uncommon cause of urinary tract infection (470). In most patients, the infection is acquired in the hospital setting (241) and is usually secondary to urinary tract surgery or instrumentation, including catheterization (148, 220, 470), or against a background of structural urinary tract abnormality (428). Wishart and Riley (455) reported an outbreak of nosocomial infection in which the source was identified as a contaminated disinfectant used for bladder instillation. Community-acquired infection appears rare (265).

Peritonitis in patients undergoing chronic ambulatory peritoneal dialysis has been observed (29, 324, 403, 404). This may follow Tenckhoff cannula exit-site infection, although most S. maltophilia exit-site infections do not appear to progress to peritonitis (83).

Urethritis, periurethral abscess, and epididymitis have also been reported (150, 180, 401). Garcia et al. described the isolation of the bacterium from semen specimens from fertile men (140).

Skin and Soft Tissue Infection S. maltophilia is a relatively frequent isolate from wounds and other skin lesions (148, 150, 180, 220, 279,390). However, a paucity of clinical details, coupled with incomplete information on microbiologic and other diagnostic criteria, often make it difficult to establish an unequivocal role for S. maltophilia as a pathogen in this context (250, 262, 330,429), particularly when the organism is isolated in mixed culture. Improved diagnostic criteria may, in the future, allow the distinction between true infection and colonization (241).

Wound sepsis may follow accidental injury (148, 150, 330) including accidents in the workplace (4, 109, 155). Infection also occurs after surgical trauma (148, 150) and at sites where cutaneous defenses have been breached iatrogenically, such as tracheostomies and suprapubic (115), Tenckhoff, and vascular catheter sites (115, 429).

Primary and metastatic S. maltophilia cellulitis is reported increasingly from patients with solid-organ or hematologic malignancy. In a series of 114 cases of S. maltophilia infection in cancer patients, Vartivarian et al. reported 11 cases of cellulitis (429). Metastatic cellulitis may manifest as ecthyma gangrenosum. Although more commonly associated with P. aeruginosa bacteremia, this syndrome is a recognized feature in systemic S. maltophilia infection in oncology patients (52, 115, 197, 215, 285, 429). Recently, a syndrome of metastatic cellulitis associated with nodular skin lesions in individuals with leukemia has been described (429). These lesions may mimic disseminated fungal infections and are associated with a poor prognostic outcome. S. maltophilia is also associated with mucocutaneous and perineal lesions in patients with malignancy (287, 334, 429).

Other manifestations of S. maltophilia soft tissue infection include umbilical cellulitis (455), prepatellar bursitis (318), infections of burn wounds (210), cat scratches, and human bite wounds (390).

Bone and Joint InfectionBone and joint infections are uncommon. They may follow orthopedic surgery (148, 150, 172) or trauma (17). Sequeira et al. reported a case of pubic symphysis infection in an intravenous drug abuser from whom S. maltophilia and P. aeruginosa were isolated (379).

Gastrointestinal InfectionAlthough asymptomatic fecal carriage of S. maltophiliahas been noted in several reports (see the section on habitat, above), the bacterium appears to be an infrequent cause of gastrointestinal infection. Isolation from ascitic fluid and from intra-abdominal abscesses (including liver abscesses) has been recorded (102, 107,286, 390), and five cases of cholangitis caused by S. maltophilia, all of which were secondary to biliary tract obstruction associated with malignancy, have also been described (317, 470). Zuravleff and Yu (470) cited a case of bacteremia subsequent to gastroenteritis, although it is unclear whether the two events were causally related. Minah et al. noted the appearance of S. maltophilia in the oral flora of cancer patients undergoing myelosuppressive chemotherapy, but they did not observe symptoms of acute infection in these individuals (272).

It has been suggested that cell wall defective variants of S. maltophilia may play a role in the pathogenesis of Crohn’s disease and ulcerative colitis (321, 322). Graham et al. used DNA hybridization techniques to identify sequences with homology to S. maltophilia in patients with chronic inflammatory bowel disease (CIBD) but concluded that the role of the bacterium in the natural history of CIBD has not been established unequivocally (158). Additional immunologic, serologic, and immunocytochemical evidence does not support its role in CIBD (190, 443, 444).

In Vitro Susceptibility TestingInfections caused by S. maltophilia are particularly difficult to manage because clinical isolates are frequently resistant to many antimicrobial agents (see below). These difficulties are compounded by methodological problems associated with susceptibility testing. Disk diffusion methods are deemed inaccurate and of poor reproducibility (178, 268, 315, 316, 462), with marked variations in results observed when zone diameters were measured after 24 and 48 h of incubation (315, 316). Quinolone agents, particularly ciprofloxacin, appear most problematic in this respect (20, 161, 316, 354), although Arpi et al. reported that disk diffusion was as reliable as Etest and agar dilution in determining the susceptibility of 124 clinical isolates of S. maltophilia to ciprofloxacin (11). In a study of 33 clinical isolates comparing agar dilution and disk diffusion tests, Hohl et al. (178) observed a 12% very major error rate and 58% minor error rate with ciprofloxacin. Although most workers have noted that erroneous disk diffusion tests overstate the activity of quinolones against S. maltophilia, false resistance has also been reported (161). The National Committee for Clinical Laboratory Standards currently recommends agar or broth dilution methods for S. maltophilia (290). These methods, however, are time-consuming, and alternative methods such as the Etest (AB Biodisk Solna Sweden) have been evaluated (316). Yao et al. compared the Etest with an agar dilution method by using 176 clinical isolates of S. maltophilia and reported an overall agreement of 94% between the two techniques (463), although Pankuch et al. noted that MICs of ticarcillin-clavulanate obtained with by technique did not correspond well to those obtained by four other methods (316).

A number of factors have been cited as affecting results of susceptibility testing. With respect to drugs of the β-lactam class, these include both the type (111) and nutrient concentration of susceptibility medium used (45). Hawkey et al. noted that the concentration of Zn²⁺ in the test medium influences the observed susceptibility of S. maltophilia to imipenem, although variations in Ca²⁺ and Mg²⁺concentration had no such effect (171). A similar phenomenon was not observed for the Zn²⁺ concentration and meropenem (75). This observed effect does not appear to be mediated by a direct effect of the zinc ions on the zinc-dependent L1 β-lactamase (see below) of S. maltophilia (46). Divalent cation concentrations have, however, been reported to affect the MICs of the carboxy- and ureidopenicillins (82, 124). Similar effects have been observed with tetracycline, polymyxin B, aminoglycosides, and trimethoprim-sulfamethoxazole (82,124).

Tilton et al. observed differences in the MICs of gentamicin and tobramycin in a comparison of broth and agar techniques (408). The incubation temperature also influences the results obtained on susceptibility testing, with aminoglycosides and polymyxin B appearing less susceptible when the bacteria are incubated at 30°C (320, 365, 442).

Resistance of Antimicrobial Agents S. maltophilia is resistant to many currently available broad-spectrum antimicrobial agents. Particularly noteworthy is the resistance to drugs of the β-lactam class. Initially, resistance to these agents was thought to be conferred by two enzymes designated L1 and L2 (370, 371). The former, produced by virtually all wild-type strains (329), belongs to the metalloenzyme family, although the deduced amino acid sequence shows significant differences from that of other enzymes of this class such as the β-lactamases produced by Aeromonas hydrophila andBacillus cereus (126, 439). These enzymes are dependent on the presence of Zn²⁺ at the active site. Although the zinc moiety in L1 can be replaced by Co²⁺, Cd²⁺, or Ni²⁺, these enzymes are less active than the native enzyme (35). The holoenzyme, which consists of a tetramer of four equal subunits, hydrolyzes a broad range of drugs of the β-lactam class, albeit with different degrees of catalytic efficiency (125). It exhibits predominantly penicillinase activity, although it is unable to hydrolyze aztreonam. It has been the subject of much attention, however, because of its ability to hydrolyze carbapenems such as imipenem and meropenem. It is not susceptible to β-lactamase inhibitors, such as clavulanate, although Payne et al. have very recently reported the inhibition of L1, and other metallo-β-lactamases by mercaptoacetic acid thiol ester derivatives (328). Following experiments with a purified enzyme fromBacillus cereus, they proposed that hydrolysis of these compounds by metallo-β-lactamases yields mercaptoacetic acid, which forms a disulfide link with a cysteine residue at the active site (328). In contrast to L1, L2, which exists as a dimer in its native state, is an enzyme with a serine active site. Comparison of the amino acid sequence of L2 with other enzymes has shown that it is most closely related to the TEM β-lactamase (440). It exhibits principally cephalosporinase activity but also hydrolyzes aztreonam (370). Unlike L1, this enzyme is susceptible to β-lactamase inhibitors. Although L2 is more susceptible to clavulanate than to tazobactam or sulbactam, there appears to be no correlation between the susceptibility to a β-lactamase inhibitor and the MIC of a β-lactam–inhibitor combination (341). Both L1 and L2 are chromosomally encoded and are inducible, and there is evidence that they share regulatory components (5, 364). Bonfiglio et al., however, recently described a clinical isolate ofS. maltophilia with low-level expression of the L1 β-lactamase which retained inducibility for the L2 enzyme (47). In addition, Payne et al. (329) reported another clinical isolate of the bacterium which elaborated high levels of carbapenemase without induction and suggested that this strain might have a mutation in its regulatory gene(s) which allowed the constitutive production of significant levels of enzyme.

The existence of additional β-lactamases produced by S. maltophilia was first suggested by Cullman and Dick (80), who described the presence of six distinct enzymes in 20 clinical strains of the bacterium. Subsequent reports have confirmed the heterogeneity of β-lactamase production in S. maltophilia, with the additional enzymes categorized as either serine enzymes or metalloenzymes (80, 326). Poor diffusion of penicillins and cephalosporins across the bacterial cell membrane has also been suggested as contributing to resistance to these compounds (79, 269), although it is unclear whether this arises from quantitative or qualitative differences in porin channels (79, 459). More recently, there have been instances of transferable cephalosporin and aztreonam resistance from nosocomial strains of S. maltophilia to recipient strains of E. coli, Proteus mirabilis, and P. aeruginosa(14, 188). Kelly et al. reported transferable penicillin and cefazolin resistance in a clinical strain of S. maltophiliaassociated with a 5.6-kb plasmid (212). There have been no recorded instances of transferable carbapenem resistance.

Resistance to aminoglycosides mediated by aminoglycoside-modifying enzymes is apparently uncommon in S. maltophilia(335). King et al. (225) described the production of aminoglycoside 6′-N-acetyltransferase by a single isolate of the bacterium. Vanhoof et al. (423), in their study of six clinical isolates, reported the expression of an aminoglycoside-O-nucleotidyltransferase and aminoglycoside 6′-N-acetyltransferase in one strain. Reduced uptake of aminoglycosides is possibly the most important mechanism which determines resistance to these compounds. Temperature-dependent aminoglycoside resistance was initially attributed to changes in outer membrane profiles (446).

More recent evidence from strains exhibiting both temperature-dependent and -independent resistance suggests that an altered cell surface microenvironment secondary to changes in the O side chains of lipopolysaccharides (347, 348, 423) is a more likely mechanism, with changes of the phosphate content of lipopolysaccharide at lower growth temperatures correlating well with decreased aminoglycoside susceptibility (349). Yu et al. reported the emergence of amikacin resistance in an isolate from a patient with infective endocarditis (467). Resistance was transferable to a strain of P. aeruginosa, but the mechanism of resistance was not investigated.

Resistance to quinolone agents is less well characterized. Resistant mutants are readily selected in vitro and are associated with qualitative and quantitative changes in outer membrane proteins (246). Cross-resistance to chloramphenicol and doxycycline in quinolone-resistant isolates has been reported by Lesco-Bornet et al. (246). The development of antimicrobial resistance in vivo has received comparatively little attention (69, 294). Manian et al. noted that 50% of 10 isolates lost susceptibility to at least one antibiotic during repeated isolations from intensive care unit patients (256). However, serial isolates were not subjected to a pheno- or genotypic typing technique, and the possibility of exogenous acquisition of a more resistant strain cannot be excluded. However, Garrison et al. used PFGE to demonstrate that resistant strains of S. maltophilia recovered from patients following therapy were indistinguishable from pretreatment isolates. In addition, they developed an in vitro pharmacodynamic chamber model, which permitted the observation of emergence of multiple resistance phenotypes during exposure to ceftazidime, ciprofloxacin, gentamicin, and ticarcillin-clavulanate (146).

Recently, Alonso and Martinez reported the presence of at least one multidrug resistance system in S. maltophilia, selected for by exposure to low concentrations of tetracycline. This energy-dependent efflux mechanism is effective against quinolones and chloramphenicol, as well as tetracyclines, but not against aminoglycosides or β-lactam drugs (6).

Selection of Antimicrobial AgentsThe selection of agents for use in the management of S. maltophilia infection presents a challenge to laboratarians and clinicians alike, not merely because of the aforementioned problems associated with susceptibility testing and the inherent resistance of the bacterium to many agents. While there is no shortage of in vitro studies of the activity of antimicrobial agents against S. maltophilia, there is a paucity of clinical investigations to determine the optimal therapy of infections associated with the bacterium; in particular, there have been no controlled trials, and recommendations for therapy are often based on retrospective studies and anecdotal reports.

Table 6 shows the susceptibility ofS. maltophilia strains to selected antimicrobial agents. The wide range of values is, at least in part, explicable by differences in the susceptibility testing methods used by the investigators which, as noted above, may significantly influence the observed results. Differing criteria for the definition of susceptibility and resistance and also the fact that some investigators have tested limited numbers of isolates also compound the difficulties in comparing studies.

There have also been very few studies which have examined the antimicrobial susceptibility of strains from disparate locations. Sader et al. tested isolates from the United States, Brazil, and Switzerland, although their results do not give a detailed comparison between country of origin and antimicrobial susceptibility (369). Despite their limitations, it is possible to draw general conclusions from the available data and to make broad recommendations for the therapy of S. maltophilia infection. With few exceptions (142, 427) most studies have found trimethoprim-sulfamethoxazole to be active against most strains of the bacterium, and this drug has long been regarded as the agent of choice for the therapy of S. maltophilia infection (113, 118,178, 276, 355). This is despite the view that this drug is only bacteriostatic for most isolates, a fact which has prompted some authorities to recommend that this compound be used at close to the maximum tolerated doses (427). Toxicity of the sulfonamide component of the combination may prove to be a limitation of this approach (113). Resistance to trimethoprim-sulfamethoxazole has long been recognized (276). Although resistance to this drug may be increasing, Vartivarian et al. noted an increase in the trimethoprim-sulfamethoxazole susceptibility of isolates at the M. D. Anderson Cancer Center over a 12-year period. They attributed this to the discontinuation of this drug as a commonly used agent of antibacterial prophylaxis. Not unexpectedly, they observed a parallel increase in resistance to the quinolones—the compounds which replaced trimethoprim-sulfamethoxazole in this role—over the same period (427). Recently, Muder et al. suggested that a combination of trimethoprim-sulfamethoxazole and either ticarcillin-clavulanate or an extended-spectrum cephalosporin may be superior to trimethoprim-sulfamethoxazole alone (284).

There is near-universal agreement that the penicillins and cephalosporins, with a few exceptions, exhibit poor activity againstS. maltophilia. The resistance of the bacterium to carbapenems, including newer members of this class (406) such as biapenem (367), is well recognized, although variations in the activity of these compounds have been noted. Piddock and Turner reported that three strains of S. maltophiliawhich were resistant to imipenem were susceptible to meropenem (336). Although several studies have shown that most strains tested are susceptible to the oxa-β-lactam moxalactam (Table 6) (397), the hematologic side effects of this agent have precluded its continuing use in clinical practice. Of the extended-spectrum cephalosporins, ceftazidime appears to possess reasonable activity (144), although it must be stated that this is highly variable between strains, and this compound cannot be recommended for use as empirical therapy. Ticarcillin-clavulanate has been noted by several investigators to exhibit good activity againstS. maltophilia, and it has been suggested that this agent should be used as the agent of choice in individuals intolerant of trimethoprim-sulfamethoxazole. In contrast, other β-lactam–β-lactamase inhibitor combinations such as piperacillin-tazobactam (58, 122, 391), amoxicillin-clavulanate, and ampicillin-sulbactam (142) are poorly active against the bacterium. Only Fujita et al. have reported that 8 of 10 strains isolated from patients with S. maltophilia pneumonia were susceptible to a combination of cefoperazone and sulbactam (137).

Similarly, other combinations which have been evaluated, such as aztreonam-sulbactam, aztreonam-tazobactam (142), and clavulanic acid combined with carbenicillin, imipenem, ceftazidime, or aztreonam (142, 143), do not offer increased activity against the bacterium with the exceptions of 2:1 and 1:1 (wt/wt) aztreonam-clavulanic acid (11, 143). However, the differing pharmacokinetics of these two agents dictate that levels of clavulanic acid in plasma decrease more rapidly than those of aztreonam, and the clinical applicability of this combination may be limited (143,427). Rouse et al. reported synergy with a combination of cefoperazone and sulbactam in the treatment of S. maltophilia experimental pneumonia in mice (366), and Elkhaili et al. observed increased in vitro activity of a cefepime-clavulanic acid combination compared to the activities of the agents when used singly (113). It is unlikely that agents of the trinem (formerly tribactam) class of β-lactams will have any useful role in the management of S. maltophilia infection (454).

Susceptibility to early examples of the quinolone class of antimicrobial agents, including ciprofloxacin (Table 6), varies markedly, but there have been anecdotal reports of response to these agents in systemic S. maltophilia infection (294,457). However, both superinfection by S. maltophiliain patients receiving quinolones and therapeutic failures of S. maltophilia infection have been documented (69, 294), and the use of these compounds as empirical therapy must be considered with care. Newer quinolones, such as clinafloxacin, sparfloxacin, and trovafloxacin, appear to be more active against the bacterium than do the older agents in this class (23, 34, 62, 71, 88, 121, 131, 136,182, 199, 221, 223, 224, 241, 251, 259, 315, 344-346, 350, 359-361,368, 369, 391, 427). Pankuch et al., for example, reported that 93% of 25 ciprofloxacin-resistant strains were susceptible to clinafloxacin (315). Although these in vitro studies appear promising, there are at present few data on the use of the newer quinolones in the clinical setting, and the results of clinical studies are awaited with interest.

The activity of the aminoglycosides, including newer members of this class such as isepamicin (198, 221), against S. maltophilia is disappointing, and these agents, in general, play a very limited role as monotherapy. In vitro synergy has been observed when these agents have been used in combination with other antimicrobial agents (see below). The observation that liposomal preparations of some aminoglycosides are associated with decreased MICs compared with those of the free compound deserves further investigation (307).

Minocycline and doxycycline exhibit good in vitro activity againstS. maltophilia, in marked contrast to tetracycline itself (Table 6) (300, 408), although clinical experience with these agents is extremely limited. Chloramphenicol has also been reported by several investigators (Table 6) as having good activity against the bacterium, and there are anecdotal reports of therapeutic success with this drug (20). The potential for hematologic side effects is of particular importance, and since many patients at risk of S. maltophilia infection have underlying disease-related or iatrogenic myelosuppression, chloramphenicol is not widely used in clinical practice. Because of the disappointing outcomes achieved with monotherapy in the management of S. maltophilia infection, it is not surprising that several investigators have proposed the use of combinations of antimicrobial agents, with much effort focusing on the identification of those two- or three-drug combinations which demonstrate in vitro synergy against the bacterium (30, 66, 72, 124, 156, 286, 299, 340, 362,466).

It is difficult to draw firm conclusions from these studies, not only because of strain-to-strain variation and the wide variety of combinations tested but also because of the differing methods used. These include broth and agar dilution, checkerboard, and half-checkerboard techniques, as well as time-kill curves. Indeed, variations within studies have been observed in which combinations identified to be synergistic by the checkerboard method were found not to be so by killing-curve criteria (66, 124, 340). In addition, synergy may be strain dependent (88, 156). Furthermore, although combinations with demonstrable synergy in the laboratory setting have been associated with successful therapeutic outcomes, other combinations deemed to be nonsynergistic in vitro have also been associated with favorable clinical results (467). Moreover, it is important to note that although synergy between combinations of drugs may be demonstrable, synergy may not occur at clinically achievable concentrations (72). For example, in their investigation of the use of aerosolized amiloride in the therapy of Pseudomonas infections in cystic fibrosis patients, Cohn et al. were able to demonstrate synergy between amiloride and tobramycin against three strains of S. maltophilia(74). However, this phenomenon was observed only at levels of amiloride in excess of those achievable in clinical practice. Studies in which synergy was demonstrated are summarized in Table7.

Few data on the use of combinations of antimicrobial agents in clinical practice have been published, and they are derived from anecdotal evidence (168, 467) and noncontrolled studies (284), but the utility of these therapeutic approaches clearly deserves further study.

Management of Infection by Means Other than Antimicrobial AgentsIn addition to antimicrobial chemotherapy in the management ofS. maltophilia infection, several investigators have stressed the importance of removing infected vascular access devices (27, 115, 429) or prosthetic material (327, 415). Muder et al., however, reported the successful management of central venous catheter-related bacteremia without removal of the device (284).

PrognosisSeveral investigators have examined factors which influence the outcome in S. maltophilia infection. In their prospective multicenter study of 91 cases of bacteremia, Muder et al. (284) found that hematologic malignancy, transplantation, neutropenia, immunosuppressive therapy, and a high severity of illness score (based on temperature, presence of hypotension, mental status, and the need for ventilatory support) were important. In contrast to other investigators (238), Muder et al. did not find a significant difference in acute mortality if S. maltophiliabacteremia was polymicrobial or monomicrobial in origin. They were unable to demonstrate a significant association between survival and administration of appropriate antimicrobial therapy (unlike another study [220]), although they acknowledged that the number of patients classified as being inappropriately treated may have been too small to allow meaningful statistical analysis.