Epidemiological cut-off values for Flavobacterium psychrophilum MIC data generated by a standard test protocol.

Epidemiological cut-off values were developed for application to antibiotic susceptibility data for Flavobacterium psychrophilum generated by standard CLSI test protocols. The MIC values for ten antibiotic agents against Flavobacterium psychrophilum were determined in two laboratories. For five antibiotics, the data sets were of sufficient quality and quantity to allow the setting of valid epidemiological cut-off values. For these agents, the cut-off values, calculated by the application of the statistically based normalized resistance interpretation method, were ≤16 mg L(-1) for erythromycin, ≤2 mg L(-1) for florfenicol, ≤0.025 mg L(-1) for oxolinic acid (OXO), ≤0.125 mg L(-1) for oxytetracycline and ≤20 (1/19) mg L(-1) for trimethoprim/sulphamethoxazole. For ampicillin and amoxicillin, the majority of putative wild-type observations were 'off scale', and therefore, statistically valid cut-off values could not be calculated. For ormetoprim/sulphadimethoxine, the data were excessively diverse and a valid cut-off could not be determined. For flumequine, the putative wild-type data were extremely skewed, and for enrofloxacin, there was inadequate separation in the MIC values for putative wild-type and non-wild-type strains. It is argued that the adoption of OXO as a class representative for the quinolone group would be a valid method of determining susceptibilities to these agents.

Effect of incubation temperature and time on the precision of data generated by antibiotic disc diffusion assays.

The influence on the precision of disc diffusion data of the conditions under which the tests were performed was examined by analysing multilaboratory data sets generated after incubation at 35 °C for 18 h, at 28 °C for 24 h and 22 °C for 24 h and 48 h. Analyses of these data sets demonstrated that precision was significantly and progressively decreased as the test temperature was reduced from 35 to 22 °C. Analysis of the data obtained at 22 °C also showed the precision was inversely related to the time of incubation. Temperature and time related decreases in precision were not related to differences in the mean zone sizes of the data sets obtained under these test conditions. Analysis of the zone data obtained at 28 and 22 °C as single laboratory sets demonstrated that reductions of incubation temperature resulted in significant increases in both intralaboratory and interlaboratory variation. Increases in incubation time at 22 °C were, however, associated with statistically significant increases in interlaboratory variation but not with any significant increase in intralaboratory variation. The significance of these observations for the establishment of the acceptable limits of precision of data sets that can be used for the setting of valid epidemiological cut-off values is discussed.

Statistical characterisation of bacterial wild-type MIC value distributions and the determination of epidemiological cut-off values.

MIC distribution data were obtained from a variety of international sources, and pooled after selection by a defined criterion. Sixty-seven of these datasets were subjected to a range of statistical goodness-of-fit tests. The log-normal distribution was selected for subsequent modelling. Cumulative counts of MIC distribution data were fitted to the cumulative log-normal distribution using non-linear least squares regression for a range of data subsets from each antibiotic-bacterium combination. Estimated parameters in the regression were the number of isolates in the subset, and (the log(2) values of) the mean and standard deviation. Optimum fits for the cumulative log-normal curve were then used to determine the wild-type MIC range, determined by calculating the MICs associated with the lower and upper 0.1% of the distribution, rounding to the nearest two-fold dilution, and calculating the probabilities of values higher and lower than these values. When plotted logarithmically, histograms of MIC frequencies appeared normal (Gaussian), but standard goodness-of-fit tests showed that the two-fold dilution grouping of MICs fits poorly to a log-normal distribution, whereas non-linear regression gave good fits to population (histogram) log-normal distributions of log(2) MIC frequencies, and even better fits to log-normal cumulative distributions. Optimum fits were found when the difference between the estimated and true number of isolates in the fitted subset was minimal. Sixteen antibiotic-bacterium datasets were fitted using this technique, and the log(2) values of the means and standard deviations were used to determine the 0.1% and 99.9% wild-type cut-off values. When rounded to the nearest two-fold dilution, > or = 98.5% of MIC values fall within the cut-off value range. Non-linear regression fitting to a cumulative log-normal distribution is a novel and effective method for modelling MIC distributions and quantifying wild-type MIC ranges.

Purification and characterisation of a plasminogen-binding protein from Haemophilus influenzae. Sequence determination reveals identity with aspartase.

Plasminogen binding proteins have been described both for Gram positive and Gram negative bacteria. In the present work we describe the purification and characterization of a plasminogen binding protein from Haemophilus influenzae (strain HI-23459). Bacteria were sonicated in order to solubilize plasminogen-binding proteins. The supernatant was subjected to affinity chromatography on plasminogen kringle-4 fragment bound to Sepharose 4B and subsequently processed by ion-exchange chromatography on DEAE-Sepharose CL-6B. Characterization of the protein by SDS-PAGE displayed a single band with a molecular mass of about 55,000, both prior to and after reduction. The purified protein stimulates tPA (tissue plasminogen activator) catalysed plasminogen activation by a factor of approximately 300, mainly due to a decrease in K(m). Antibodies were raised in rabbits and used in quantitative and qualitative analysis. However, using a FITC-conjugate we failed to demonstrate the presence of the purified protein on the surface of intact bacteria. The corresponding gene was isolated from a lambda EMBL3 phage library prepared from chromosomal DNA from the same H. influenzae strain, using an oligonucleotide probe based on the NH2-terminal amino acid sequence. An open reading frame corresponding to 472 amino acid was found. The amino acid sequence of the translated gene demonstrates 97% identity with the recently published sequence from aspartate ammonia lyase (aspartase) from H. influenzae. Enzymatic analysis of the purified protein revealed a high aspartase activity.

Integrons and gene cassettes in clinical isolates of co-trimoxazole-resistant Gram-negative bacteria.

Despite a trend of declining consumption, resistance to co-trimoxazole has increased during a 12-year period in Stockholm. The molecular background to this surprising development was investigated by using PCR to screen for integrons and specific resistance genes, followed by sequence analysis of selected integrons, in 105 clinical urinary isolates of Gram-negative bacteria selected partly for trimethoprim resistance. Sixty-five integrons of class 1 or 2 were detected in a subset of 59 isolates, and of these positive isolates, all but one were resistant to trimethoprim. However, 11 isolates were resistant to trimethoprim, but negative for integrons. Isolates positive for integrons were resistant to an average of 4.2 antibiotics, compared with 1.9 antibiotics for integron-negative isolates. Despite this, the only gene cassettes identified in 19 class 1 integrons analysed were dfr and aadA cassettes. Thus, only resistance to trimethoprim, streptomycin, spectinomycin and sulphonamides could be explained by the presence of integrons in these isolates. A new dfr gene, named dfrA22, was discovered as a single gene cassette in a class 1 integron. In addition, sulphonamide resistance in many isolates was caused by carriage of sul2, which has no known association with integrons. Resistance to co-trimoxazole and many other antibiotics was thus not accounted for fully by the presence of integrons in these isolates.

Meropenem susceptibility breakpoint for Pseudomonas aeruginosa strains hyperproducing mexB mRNA.

Twenty-five isolates of Pseudomonas aeruginosa with different meropenem susceptibilities were subjected to quantitative RT-PCR for analysis of transcription levels of oprD, mexB and mexD, and, in selected isolates, PA3720, which is hyper-expressed in nalC efflux mutants. Regulator genes of efflux pump MexAB-OprM, mexR and PA3721 (putative) were sequenced in selected isolates. The potential for mathematical reconstruction of the ideal susceptible population using normalised resistance interpretation (NRI) was also studied. In three isolates with intermediate susceptibility to meropenem (according to Swedish breakpoints), a reduction in MIC from 4 to 2 mg/L was observed with efflux inhibitor MC-207,110. These isolates would be considered susceptible according to British Society for Antimicrobial Chemotherapy and NCCLS breakpoints. These three isolates had between 4.6- and 5.0-fold increases in mexB transcription. None of these isolates had significant nalB mutations, but an Ala145-->Val mutation was observed in PA3721 in two of the isolates. However, these isolates had moderately increased production of PA3720 only. Single-strain regression analysis did not detect any major biological differences between the different groups. Using NRI, a disk-diffusion susceptibility breakpoint of >/= 28 mm was generated. Isolates with intermediate susceptibility to meropenem, which are considered fully susceptible in many countries, displayed possible low-grade meropenem resistance mechanisms, implying that the susceptibility breakpoint should be reconsidered. The increased transcription of mexB mRNA in such isolates seems unrelated to nalB or nalC mutations.

Fractionation of rat IgG subclasses and screening for IgG Fc-binding to bacteria.

The four IgG subclasses of the rat, IgGl, IgG2a, IgG2b and IgG2c, were purified from normal serum by a combination of protein A-affinity chromatography and DEAE-cellulose chromatography. Purified, radiolabelled preparations of IgG were tested for binding to Gram-positive bacteria representing five different Fc-receptor (FcR) types. Distinct rat subclass-specific Fc-binding was noted to bacterial species belonging to different Fc-receptor types. Staphylococcus aureus (FcR I) strains bind IgGl and IgG2c as shown by others. Group C and G Streptococci (FcR III) bind all four subclasses of rat IgG. Streptococcus zooepidemicus strains (FcR V) also bind all four subclases but only to a lower degree. Human group A Streptococci (FcR II) and bovine group G Streptococci (FcR IV) do not bind any of the rat IgG subclasses. Elution studies on two strains. Staphylococcus aureus, Cowan I, and human group G Streptococcus, G 148, showed that both thiocyanate and pH-elution might be useful for the fractionation of IgG subclasses bound to bacterial cells. The present work indicates the possible use of bacterial cells as solid-phase absorbents in immunological studies of rat IgG.

Little difference between minimum inhibitory concentrations of Mycobacterium tuberculosis wild-type organisms determined with BACTEC MGIT 960 and Middlebrook 7H10.

The MIC wild-type (WT) distribution for Mycobacterium tuberculosis in BACTEC 960 MGIT is not defined, which may result in poor reproducibility for drug susceptibility testing (DST), as several DST methods with different breakpoints are in use. In a comparison between MGIT and Middlebrook 7H10 medium of seven first- and second-line drugs, including 133 MIC determinations of 15 WT isolates, we found an agreement of 91.7% within ± one MIC dilution step. The results confirm the agreement in MIC testing between 7H10 and MGIT and indicate that breakpoints could be harmonized in order to avoid misclassification.

Different protein A immunosorbents may have different binding specificity for rat immunoglobulins.

Purified polyclonal immunoglobulin preparations representing the 4 rat IgG subclasses were tested for binding to Staphylococcus aureus protein A attached to 3 different solid phases (Staphylococcus aureus Cowan I bacteria, Sepharose CL4B and Sepharose 6MB). Protein A Sepharose CL4B showed higher reactivity with IgG1 and IgG2b than the staphylococci, whereas protein A Sepharose 6MB showed a lower uptake of these subclasses. No differences were seen for IgG2a and IgG2c. Protein A on different solid phases cannot be used interchangeably without confirmation of binding specificity.

On the theory of the disk diffusion test. Evidence for a non-linear relationship between critical concentration and MIC, and its practical implications for susceptibility testing of Haemophilus influenzae.

The procedure for disk diffusion susceptibility testing has been worked out for rapidly growing non-fastidious bacteria. Using the general zone diameter breakpoints for interpretation of susceptibility, it was found that clinical isolates of H. influenzae were assigned to the wrong SIR category in fifty per cent of the strains for erythromycin, and ten per cent for doxycycline. New species- and laboratory-specific interpretive zone diameter breakpoints corresponding to the recommended MIC limits were therefore worked out. The Standard Curve regression Analysis (SCA) method used for this purpose is based on the correlation between zone size and disk content, using two reference strains with different MICs. In its original version (the Single strain Regression Analysis, SRA) only one reference strain was used. This equation was found not to be generally valid since the relationship between MIC and the critical concentration is not constant, as was originally assumed. The slope and intercept of the regression line obtained by SCA is species related, while a general regression line based on results from many different species assumes that there is the same relation between the zone size and MIC for all species. Breakpoints for erythromycin and doxycycline calculated by the SCA equation gave more accurate results in routine susceptibility testing of H. influenzae and reduced the error rate from fifty three and ten per cent to four and three per cent for the two antibiotics.

Distribution of presumptive pathogenicity factors among beta-hemolytic streptococci isolated from Ethiopia.

Beta-hemolytic streptococci are known to bind several mammalian proteins, which are presumed to be important in pathogenicity. The distribution of such binding structures was examined for mouse albumin, human serum IgA, human IgG, human fibrinogen, and human plasminogen. A total of 218 group A beta-hemolytic streptococci (GAS) were studied: 5 isolates from children with acute rheumatic fever (ARF), 18 from acute post-streptococcal glomerulonephritis (APSGN), 57 from tonsillitis, 52 from skin infections, and 86 from healthy carriers. Sixty-eight Streptococcus equisimilis and 20 group G streptococci were also included. Most of the S. equisimilis (60/68) and group G (14/20) were obtained from apparently healthy carriers. The results were evaluated with respect to T type, serum opacity reaction (SOR), site of isolation, and disease type. No direct correlation was detected between the protein-binding structures studied. There was no apparent correlation between any particular protein-binding structure and specific T type. Albumin-binding and IgA-binding activities were inversely correlated among skin and nephritis GAS isolates. A strong correlation was demonstrated between IgA-binding activity and SOR production, while albumin-binding activity correlated with SOR-negative strains. Albumin-binding levels in isolates from ARF, APSGN and tonsillitis were significantly higher than in isolates from healthy carriers (P < 0.001). A higher albumin-binding capacity was shown in skin isolates from APSGN than in isolates from impetigo (P < 0.001).

Antibiotic disk diffusion testing revisited. Single strain regression analysis. Review article.

The standardized (NCCLS, ICS, DIN etc.) disk diffusion method is the most widespread technique for antibiotic susceptibility testing. Interpretive zone breakpoints are calculated from the regular regression line between minimum inhibitory concentrations (MIC) of bacterial isolates and the corresponding inhibition zone diameters around the disk containing the antibiotic. Studies of the regression line has revealed marked differences between different bacterial species. A newly described equation, the single strain regression analysis (SRA) equation, can be used to determine the regression line constants for individual strains. This method was applied to ciprofloxacin and S. aureus, E. faecalis, E. coli, P. mirabilis, P. aeruginosa, and P. maltophilia. The slope and intercept constants were determined for all 40 strains and showed a strong similarity within each species. A close similarity was also observed between the two Pseudomonas species and between S. aureus and E. faecalis. When the regression lines calculated by SRA for individual strains were extrapolated towards higher MIC values, the lines obtained for the more susceptible strains predicted the zones of more resistant strains within the species. The applications of SRA to several other antibiotics and bacterial species in earlier studies were reviewed. One exception to the predictive power of SRA has been detected earlier, H. influenzae and erythromycin. This led to the formulation of the standard curve regression analysis (SCA) equation which requires the use of two or more strains. Methodological aspects of SRA/SCA applications were presented. Three areas are particularly well suited for the use of SRA/SCA: 1. Calculation of interpretive zone breakpoints corresponding to recommended MIC limits in the individual laboratory. 2. Analysis of the effects of various disk contents of antibiotic on the resulting inhibition zones for various bacteria when new antibiotics are introduced. 3. Analytical tool as part of external quality control programmes.

New receptor for human plasminogen on gram positive cocci.

180 bacterial strains representing 17 different species of gram positive cocci were tested for the ability to interact with human plasminogen. Receptors for plasminogen could be detected on 23/24 strains of S. pyogenes, 15/15 strains of S. equisimilis, 14/16 strains of human group G streptococci and 14/14 strains of S. pneumoniae. Eight of nineteen strains representing five species of alpha-hemolytic streptococci were also positive. S. equisimilis demonstrated the highest uptake with a median value of 58 per cent (20%-67%). On the other hand, all strains of S. agalactiae, the majority of S. faecalis and all S. aureus, S. epidermidis and S. saprophyticus strains tested were negative. The concentration of unlabelled plasminogen causing a 50 per cent reduction of bound tracer was between 50 and 150 mM. These estimates of the dissociation constant confirmed the specific nature of the interaction. Binding of plasminogen could be blocked by addition of plasmin-aprotinin complex, suggesting that plasminogen and plasmin bind to the same receptor. Binding was also blocked by the plasminogen fragment kringle 1-3, but not by miniplasminogen, a fragment containing kringle 5 and the B-chain region. As streptokinase interacts mainly with the B-chain of plasmin it is clear that the bacterial receptor for plasminogen is not identical to streptokinase.

Types of interpretive errors in susceptibility testing. Zone breakpoints for norfloxacin disk diffusion testing.

A total of 548 strains of the eleven most common urinary tract pathogens were investigated for possible errors in norfloxacin susceptibility tests comparing MIC determinations with disk diffusion assays. Most strains were found to be sensitive with MIC-90 values below 1.0 for the Enterobacteriaceae while the classical nalidixic acid resistant species, the gram-positive bacteria and Pseudomonas aeruginosa, were less susceptible to norfloxacin with MIC-90 above 1.0 mg/l. MIC-values close to interpretive MIC-limits were recorded for S. faecalis and S. agalactiae using the recommendations of the national Committee for Clinical Laboratory Standards (NCCLS) (susceptible, S less than or equal to 4.0) and for P. aeruginosa and S. aureus using the Swedish Reference Group for Antibiotics (SRGA) standards (S less than or equal to 1.0). Susceptibility interpretations for these species showed a lack of accuracy consistent with methodological problems of reproducibility, an error called type I. The changes in the MIC-limits required for these strains to correct the error would be S less than or equal to 4 for P. aeruginosa and S. aureus, S less than or equal to 8 for S. agalactiae and S less than or equal to 0.5 for S. faecalis. A type II error, occurring when a bacterial species shows a regression line different from the regular line, was also identified for S. saprophyticus. The use of breakpoints derived from single strains regression analysis corrected this error and also reduced the frequency of similar misinterpretations in other species. The term "species-specific MIC-limits" should be introduced along with the established concept of "species-specific interpretive zone breakpoints" to allow for the correction of type I interpretive errors. Type II errors can be corrected by using species-specific interpretive breakpoints, either issued by reference laboratories or derived by calculations from single-strain regression analysis in the individual laboratory.

Two types of receptors for human plasminogen on group G streptococci.

To investigate the nature of plasminogen binding to streptococci, strains selected for high reactivity with human plasminogen were examined for binding pattern against a panel of plasminogen fragments. The strains included human isolates of groups A, C and G as well as bovine isolates of group G. All strains reacted substantially with the plasminogen fragment kringle 1-3. Using the miniplasminogen fragment (kringle 5 and the B chain) a small but reproducible uptake was detected for human group G strains but not for group A or C strains. The group G strains of bovine origin on the other hand demonstrated high uptake of miniplasminogen, suggesting the possibility of an alternative plasminogen receptor for this species. This interpretation was supported by blocking experiments with the lysine analogue EACA where low concentrations (1 mM) completely blocked plasminogen binding to human streptococci, whereas a 100-fold higher concentration was needed for bovine group G strains. Scatchard plots with human isolates resulted in straight lines and Kd values were generally in the range of 20-80 nM. The number of receptors was estimated to be 45,000 for a selected group A strain and about 10,000 for the selected group C and G strains. Scatchard analysis with bovine group G isolates on the other hand revealed a two phase interaction, supporting the assumption of two different receptor structures on these strains. Kd for the first phase was estimated to be about 20 nM (10,000-20,000 receptors per bacterium), which was similar to the human strains, whereas the second phase was in the range of 400-500 nM (50,000 and 150,000 receptors per bacterium with two selected strains). Scatchard plots with the miniplasminogen fragment as ligand mimicked the phase two reaction with plasminogen, supporting the concept that this reaction represents a new and not previously described receptor. Both the receptor reacting with the kringle 1-3 portion and the one reacting with the miniplasminogen portion bound plasmin and plasminogen with similar affinity.

Non-immune Fab- and Fc- mediated interactions of avian Ig with S. aureus and group C and G streptococci.

Serum samples from 19 avian species representing 8 orders were tested for their capacity to inhibit the Fab- and Fc-mediated immunoglobulin binding to protein A-carrying S. aureus and protein G-carrying group C and G streptococci. Four species (mallard, dunlin, starling and blackbird) belonging to three different orders showed a high degree of Fc-mediated protein A- and protein G-reactivity. Five species demonstrated a high level and nine species exhibited a low level of Fab-mediated protein A-reactivity. The four species identified as Fc-reactive were capable of Fab-mediated immunoglobulin binding with streptococcal surface proteins but incapable of Fab-mediated protein A binding. SDS-PAGE analysis confirmed that the protein A-Sepharose affinity purified material contained proteins corresponding to immunoglobulin chains. Inhibition results by avian sera were confirmed by direct binding of protein A-reactive proteins to bacteria, by precipitation in gel and by Western blot analysis of binding to protein A and protein G, respectively.

Patterns of mutations in target genes in septicemia isolates of Escherichia coli and Klebsiella pneumoniae with resistance or reduced susceptibility to ciprofloxacin.

Thirty-two Escherichia coli and 21 Klebsiella pneumoniae septicemia isolates with varying degrees of resistance to ciprofloxacin were analyzed for the presence of point mutations within the quinolone-resistance target genes. The number of mutations observed in the resistant isolates agreed with the level of ciprofloxacin resistance in both species. Such isolates were also resistant to nalidixic acid. Isolates with borderline susceptibility to ciprofloxacin, on the other hand, behaved differently in the two species. In E. coli all the isolates harbored at least one mutation and these isolates were also resistant to nalidixic acid, while no mutations were detected in the K. pneumoniae isolates, and susceptibility to nalidixic acid was unpredictable. Therefore, nalidixic acid cannot be used as a class representative. Time-kill curve studies on an isolate with borderline susceptibility from each species showed higher degrees of resistance to ciprofloxacin in comparison to that of the wild-type E. coli. A previously unreported parC mutation, S57-->T, was detected in a resistant E. coli isolate and might expand the QRDR of this gene. Normalized resistance interpretations of histograms confirmed the setting of microbiological zone breakpoints for ciprofloxacin testing.

Comparison of 30 microgram and 120 microgram gentamicin disks for the prediction of gentamicin resistance in Enterococcus faecalis.

Single strain regression analysis was performed on PDM II medium for E. faecalis with 10, 30 and 120 micrograms gentamicin disks using E. faecalis, strain ATCC 29212 as the reference. This method permits the calculation of zone diameters corresponding to different MIC values for different disk contents. The lack of discrimination between normal low-level resistant strains and high-level resistance using the 10 micrograms disk was confirmed. However, both the 30 micrograms and 120 micrograms disks seemed to provide a separation of the normal low-level gentamicin-resistant population from strains with increased resistance. Since the 30 micrograms disk is used routinely in some countries, there should be no need for an extra high content disk in these laboratories. This was confirmed when 96 clinical isolates of E. faecalis were analysed and the results of routine disk diffusion tests were compared with the MIC values. Two of the strains showed high-level gentamicin resistance (greater than 2000 mg/l) and produced no zone of inhibition. The other 94 isolates showed gentamicin MIC values between 4-16 mg/l, and 72 of the MIC results were 8 mg/l. The zone diameters for these strains ranged between 15 and 25 mm with a mean of 18.2 and a median value of 18 mm. In order to include statistical considerations of the zone size populations for setting of breakpoints, a study of gentamicin zone size distributions was performed for several bacterial species. Inhibition zone diameter values around the 30 micrograms gentamicin disk for 2079 clinical isolates of E. faecalis, 2268 S. aureus, 3201 E. coli and 547 strains of P. mirabilis from different years were plotted as histograms. Tests for agreement with a Gaussian distribution showed that the histograms were slightly peaked and skewed towards higher zone values. Parametric and non-parametric statistical tests were compared and the results showed that means and medians were very similar and that parametric fractile estimations at the lower end of the histogram populations were conservative and could be used in view of a slightly lower rate of false resistance. The 1% parametric fractile of 12 mm was selected as a suitable breakpoint for the identification of normal, low-level resistant isolates of E. faecalis using the standardized disk test of the Swedish Reference Group for Antibiotics.

Epidemiological typing of Streptococcus pneumoniae from various sources in Sweden and India using Box A PCR fingerprinting.

Epidemiological typing of Streptococcus pneumoniae is essential to determine strain relatedness and also to trace resistant clones. The novel Box A PCR assay was used for characterization of S. pneumoniae isolates from two Scandinavian countries and to compare those from India on the Asian continent. In addition, the assay was employed to determine the clonality of 25 pneumococcal strains from an outbreak in a day-care centre in Linköping, Sweden. All 25 showed a unique pattern with 100% homology for 24 of them, thereby establishing the clonal nature of the outbreak. The pneumococcal strains involved in the outbreak belonged to serotype 9 and were resistant to penicillin, with an MIC value of 2 mg/l. Thirty-eight genotypes were obtained when the Box A results were analysed by computer (Molecular Analyst Software with GelCompar). The discriminatory index of the method was D=0.98, which indicates excellent performance. No major segregation of strains from the different geographical locations was observed when a lower level of similarity was used for typing (80%, 13 types). However, at the level chosen for genotyping, 95% (38 genotypes) there was a clear geographical segregation. No correlation between genotype and serotype was seen as strains from a common place of origin were most often of different serotypes. Computer-assisted analysis of the results of Box A PCR typing facilitated the evaluation.

Septic arthritis caused by a gram-negative bacterium representing a new species related to the Bordetella-Alcaligenes complex.

A knee-joint exudate culture yielded on two occasions a gram-negative bacterium. Regular methods for speciation did not provide an identification. The infection was successfully treated with ciprofloxacin. The unknown isolate, CCUG 36768, was subjected to further investigation, including 16S rDNA sequencing, protein profiling, cellular fatty acid analysis, and various biochemical tests, in order to produce a species identification. The 1469 bp-long 16S rDNA sequence did not reveal identity with any known species sequence. CCUG 36768 clustered in a group of species, including Alcaligenes defragrans, Denitrobacter permanens, Taylorella equigenitalis, Alcaligenes faecalis, and four strains of Alcaligenes species without a specific species name. Bordetella species also showed a high degree of similarity with CCUG 36768. Protein profiling, cellular fatty acid analysis and computer-assisted analysis of biochemical profiles indicated similarity with Bordetella-Alcaligenes species, often close to B. holmesii and B. avium. API 20 NE indicated the profile of Moraxella species of poor identity. It is concluded that CCUG 36768 represents a new bacterial species of pathogenic potential in humans. It is related to the Bordetella-Alcaligenes group. Powerful new methods for speciation are available and it is recommended that unknown isolates from normally sterile sites be submitted for further analysis. Several isolates are required for the definition of new species.