- Email: [email protected]
Inter-strain comparison by pyrolysis mass spectrometry in the investigation of Staphylococcus aureus nosocomial infection
Journal
of Hospital
Infection
(1991)
19, 41-48
Inter-strain comparison by pyrolysis mass spectrometry in the investigation of Staphylococcus aurezzs nosocomial infection F. K. Gould*,
R. Freeman?, P. R. Sisson$, N. F. Lightfoot
B. D. Cookson§
and
*Department of Microbiology, Freeman Hospital, f-Medical School and $Regional Public Health Laboratory, Newcastle upon Tyne and §Division of Hospital Infection, Central Public Health Laboratory, Colindale Avenue, London Accepted for publication
23 July
1991
Summary:
Pyrolysis mass spectrometry (PyMS) was used to examine isolates of Staphylococcus aureus from an outbreak of wound infections on a cardiothoracic surgical unit, some of which were thought to have been related to a point-source in the operating theatre. The PyMS results were compared with the results of phage typing. Both methods suggested that a single strain of S. am-em, of phage pattern 29/52/52A/79/80/81, was responsible for some of the wound infections. but PvMS also identified two natients with nhaee non-typable isolates. Ph’age typing indicated four staff members as p&siGe carriers of the epidemic strain, but PyMS indicated only two. Epidemiological enauirv confirmed that one of the two members of staff identified bv both methods was likely to have been the source of the theatre-based infection. PyMS is a rapid and relatively inexpensive technique for the investigation of nosocomial S. aureus infection and was more discriminatory than phage typing in this instance. Keywords: infection.
Pyrolysis
mass spectrometry;
Staphylococcus
aureus;
nosocomial
Introduction of outbreaks of wound infection due to Staphylococcus a common feature of hospital cross-infection control. A particularly important aspect is that of tracing a common source when epidemiological evidence suggests that a member of staff may be carrying and shedding an epidemic strain. In such investigations samples from apparently affected patients, other patients, staff and the environment will inevitably yield many isolates of S. aweus. Accurate characterization of these isolates is necessary to identify the putative epidemic strain. Phage The
investigation
aureus remains
Correspondence Newcastle upon
to: Dr R. Freeman, Tyne NE2 4HH.
01954701/91/090041
f08
Microbiology
Department,
$03.00/O
Medical
School,
0
41
Framlington
1991 The Hospital
Infection
Place,
Society
42
F. K. Gould
et al.
typing is the usual method employed since isolates of different phage types are unlikely to be related. Conversely, those of similar or even identical phage type are not necessarily a single strain and under such circumstances the method may provide insufficient discrimination. Phage typing is usually only performed in reference centres and the results may not be available immediately. The detection of a common source outbreak of nosocomial S. aUreUS infection and the institution of appropriate control measures may, therefore, be delayed. Pyrolysis mass spectrometry (PyMS) gives useful and rapid inter-strain comparisons for many bacterial species, including staphylococci.’ The method is based on the principle that pyrolysis of pure bacterial cultures generates mass ion spectra which are characteristic of the individual pyrolysed bacteria. Computerized analysis of the data is used to select those mass ions showing the best discrimination between the various spectra. These form the basis of a mathematically-derived analysis of similarities and differences which is then used for inter-strain comparison. In essence, the system produces a similarity matrix, using the selected mass ions as unweighted characteristics. At present the system remains comparative and reproducibility is restricted to individual batch examination.2 We now report the application of PyMS to the investigation of a presumed outbreak of nosocomial S. aUreUS infection. Materials
and methods
The outbreak In November 1990 an increase in S. aUreUS sternotomy wound infections was noted in the Cardiothoracic Centre. Surveillance records from the preceding several months indicated that the increase was recent. The affected patients were distributed across several convalescent wards and included at least one patient on the Cardiothoracic Transplant Unit as well as patients undergoing conventional cardiac surgery. Initial epidemiological investigation failed to establish whether the increase was due to a failure in postoperative wound technique or due to a point source, but several patients developed their infection soon after operation, suggesting the possibility of intra-operative contamination. In anticipation of the delay expected before phage typing results of the original isolates, it was decided immediately to organize nasal swabs from medical and nursing staff in the wards and the theatre suite. Any S. aureus isolates of similar sensitivity (penicillin resistant, otherwise fully sensitive) to those present in the postoperative wounds could then be examined at the same time. In all, 46 isolates (31 patient isolates, including two isolates from patients on an unrelated orthopaedic ward, and 15 isolates from the nares of members of staff) were collected and sent for phage typing. All 46 isolates were also examined by PyMS. Sample preparation and PyMS analysis The 46 cultures were encoded and examined
by two of the investigators,
Pyrolysis
of nosocomial
S. auf-em
43
who were unaware of the epidemiological data or of the subsequent results of phage typing until analysis was completed. Each slope was subcultured in duplicate onto plates of a single batch of diagnostic sensitivity test agar (DST; Oxoid) and incubated overnight at 37°C. Smears of each subculture in triplicate were made on pyrolysis foils (Horizon Instruments Ltd., Heathfield, Kent). The foils were inserted into pyrolysis tubes (Horizon Instruments) and dried in a hot-air oven at 80°C for 5 mins. They were then pyrolysed at 530°C as a single batch on a Horizon Instruments PYMS 200X pyrolysis mass spectrometer, as previously described.3 Integrated ion counts at mass intervals of 11 to 200 were recorded on floppy disk with the pyrolysis sequence number and total ion count for each sample. The duplicate subcultures were given consecutive PyMS numbers such that isolate 1 was pyrolysed as l/2; isolate 2 as 314, and so on.
Data analysis As a first step, differences in spectra solely attributable to variations in the amount of material pyrolysed per foil were identified and the data were corrected (normalization). The normalized spectral data for each subculture were analysed for within-group and between-group variation. Discriminant analysis was then performed on two separate datasets, one comprising the data from all the even-numbered subcultures and the second containing the data from the odd-numbered subcultures. In each set the 30 most discriminatory mass ions were subjected to principal component (PC) and canonical variate (CV) analysis. The meaned results of the triplicate samples for each subculture were then represented on an ordination diagram of PCCV 1 versus PCCV 2. The two resultant ordination diagrams (one for even-numbered and one for odd-numbered subcultures) were examined for clusters of apparently closely related isolates. Editing of the database to produce datasets containing only those isolates contained in such a cluster allowed further analysis for similarity and difference within the cluster. When the number of isolates to be thus analysed was less than 10 a final analysis was performed combining data from both sets of subcultures. The end-point of this analysis was a number of isolates in which the means of the triplicate analyses for each subculture of the same isolate were shown to differ as least as much from each other as from different isolates. Such isolates were regarded as indistinguishable by PyMS. The principles of the analysis have been described previously.3
Phage typing Phage typing was carried out at the Division of Hospital Infection, Central Public Health Laboratory, Colindale, London, by standard methods,4 using the International Standard Set of phages.
44
F. K. Gould
ef al.
Results
The ordination diagram produced from the analysis of the odd-numbered subcultures is seen in Figure 1. Two apparent clusters of isolates are seen. The larger diffuse cluster contains data from 25 patient isolates and 13 isolates from staff samples, and the smaller cohesive cluster contains data from six patient isolates and two staff samples. A similar ordination diagram was produced from the analysis of the even-numbered subcultures (not shown), and the distribution of isolates between the two clusters was the same. Analysis of the combined data from both subcultures of the isolates in each separate cluster showed that the isolates within the larger diffuse cluster were distinct since the differences between different isolates were always greater than the differences between two subcultures of the same isolate. In contrast, the analysis of the smaller cluster showed that these eight isolates were indistinguishable by PyMS since the differences demonstrated between means of triplicates of two subcultures of the same
;----\ , I’ ’ /’ ; ,’ ::
I’
n
,’ ’ ,’ ;
CPm
q+ n
:: ! 25 patients
:’ *\
: 113 staff
‘:
I’ : :
-10.2 t
-16.0 h__, -16
-8
0 Combined
8 PCCV group
16 score
24
:
I
Figure 1. Ordination diagram of the first two principal component canonical variates (PCCV 1 and 2) for odd-numbered subcultures of 46 isolates of Staphylococcus aweus examined by pyrolysis mass spectrometry. The distinct cluster seen on the right contains six patient isolates (solid squares) and two staff isolates (open squares).
Pyrolysis
of nosocomial
45
S. aureus
isolate were at least as great as any differences shown between the means of triplicates of any two separate isolates. The six patient isolates thus identified comprised five isolates from sternotomy wound infections on the affected wards in the cardiothoracic centre and one isolate from an orthopaedic ward. The two staff isolates in this same closely related cluster came from nasal swabs of, respectively, a theatre nurse and a cardiopulmonary bypass technician. The results of phage typing showed that 43 of the 46 isolates were typable, either at RTD or RTD x 100. Only one phage pattern (susceptibility to phages 29/S2/52A/79/80/81, either as a complete pattern or with less than two strong differences) was found in both staff samples and in some of the patient isolates and was, therefore, potentially that of an epidemic strain with a point source in the staff. On this basis four staff members and three cardiothoracic patients, together with one orthopaedic patient, were potentially linked. All four members of staff were treated with mupirocin nasal ointment, which eliminated the organism in each case. The results of both investigations are summarized in Table I, from which it can also be seen that six of the eight isolates found to be indistinguishable by PyMS were of phage pattern 29/52/52A/79/80/81 but two were phage non-typable. The other 38 isolates, shown by PyMS to be not only distinct from the clustered isolates but also from each other, included two isolates phage typed as 29/52/52A/79/80/81 and a varied collection of other phage groups and types. Both methods identified the same isolate from an orthopaedic patient epidemiologically unrelated to the cardiothoracic infection incident as being indistinguishable from the epidemic strain.
Table
Isolate
I. Comparison of results of phage typing and pyrolysis mass spectrometry analysis isolates of Staphylococcus aureus shown to be similar by either OY both methods Phage type 29/52/52A/79/80/81
source
Patients Cardiothoracic Orthopaedic Cardiothoracic Cardiothoracic Cardiothoracic Cardiothoracic
patient wound patient wound patient wound patient wound patient wound patient wound
Staff A. Surgeon B. Theatre nurse C. Theatre nurse D. Cardiopulmonary bypass technician
PyMS
analysis
Non-typable Yes Yes Non-typable Yes Yes
Clustered Clustered Clustered Clustered Clustered Clustered
Yes Yes Yes Yes
Not clustered Clustered Not clustered Clustered
on ten
result
46
F. K. Gould
et
al.
When both sets of results were compared with the epidemiological evidence it was found that staff members A, B and C were unable to be patients, but D (a circumstantially linked to all the identified cardiopulmonary bypass technician) had been present in the theatre suite on all relevant occasions, albeit briefly for some. A particularly important finding was that A, a surgeon, had not been present at any of the relevant operations. As part of the resultant enquiries it also emerged that traffic in and out of the cardiothoracic theatre suites had been unusually, and to some extent, unnecessarily, busy during the studied period and this may have contributed to the general increase in postoperative infections in the cases which both methods demonstrated to be unrelated to the common source. Discussion
Both methods (phage typing and PyMS) suggested that there was a small common-source outbreak of S. LZUY~ZLSwound infection that was contributing to the overall increase in staphylococcal infection in the cardiothoracic surgery patients. The results of phage typing showed that three cardiothoracic patients were infected with S. UUY~US type and that staphylococci of the same phage type were 29/52/52A/79/80/81 present in the nasal swabs of four members of staff. Since it was impossible further to distinguish between the isolates by this method all four staff members were treated with mupirocin, although it was clear from enquiries that three of them were probably not the point source of the outbreak and may themselves have become passively colonized by an outbreak strain. Follow-up studies have shown no evidence of further point-source infections due to S. uureus in sternotomy wounds. Comparison of the two methods shows that PyMS identified the same patients as phage typing but added two more patient isolates which were phage non-typable. PyMS also identified the same staff member as the point source, but with greater discrimination, the relevant member of staff being one of only two possibilities rather than four. The inclusion of phage non-typable isolates of S. UUY~USin a single strain identified by other methods, other members of which are phage typable, is not unexpected. Similar observations have resulted when S. aureus isolates have been examined by other methods, including restriction endonuclease analysis.’ Conversely, the possibility that isolates of S. aUreUS belonging to an identical phage susceptibility pattern are not necessarily a homogeneous strain has recently been shown in studies comparing the results of phage typing and electrophoretic typing methods in methicillin-resistant S. uweza (MRSA).‘j Both phenomena were also demonstrated in a PyMS study of an epidemiologically well-characterized outbreak of Listeriu monocytogenes infection.7 PyMS is not a typing method since it does not assign a permanent characterization to the examined organisms. Pyrolysis mass spectrograms
Pyrolysis
of nosocomial
S. aureus
47
vary with the duration and conditions of incubation of the organisms, leading to variability between batches. When, however, the organisms are prepared under identical conditions and analysed in a single machine batch, comparisons between isolates are objective and highly discriminatory. PyMS inter-strain comparison allows the identification of closely-related groups of isolates from within a larger collection of similar organisms and provides the essential information necessary to detect an outbreak, indicate the likely source and, therefore, institute control measures. PyMS is rapid and consumable costs are cheap in comparison to conventional techniques. Although the method is currently a research tool and the capital costs are c. Es0 000, the consumable costs are less than El per sample. The same apparatus and reagents can be used for a wide range of different bacterial species and the same method has been successfully applied in outbreaks due coagulase-negative staphylococci,’ to Streptococcus pyogenes,8 Salmonella,’ maltophilia’O penicillin-resistant and Listeria,7 Xanthomonas pneumococci. l1 This small study suggests that PyMS can also be used for the investigation of nosocomial S. aureus infection. The relative merits of PyMS and other new methods of strain characterization such as restriction fragment length polymorphism (RFLP) have yet to be established. More extensive application will be necessary before the value of PyMS in S. aweus epidemiology is determined. However, the speed, comparatively low cost and versatility of PyMS make it an ideal method for the initial screening of suspected outbreaks of allowing more expensive and time-consuming nosocomial infection, methods to be used more cost-effectively when formal typing is still necessary. References 1.
2. 3. 4. 5. 6. 7. 8. 9.
Freeman R, Gould FK, Wilkinson R, Ward AC, Lightfoot NF, Sisson PR. Rapid mass spectrometry of coagulase-negative inter-strain comparison by pyrolysis staphylococci from persistent CAPD peritonitis. Epidemiology and Infection 1991; 106: 239-246. Pitt TL. Editorial: Burnt offerings or typing system? J Hasp Infect 1991; 17: 157-158. Aries RE, Gutteridge CS, Ottley TW. Evaluation of a low-cost, automated pyrolysis mass spectrometer. J Anal Appl Pyrol 1986; 9: 81-98. Blair JE, Williams REO. Phage typing of staphylococci. Bull Wld Hlth Ovg 1961; 24: 771-784. Rhinehart E, Shlaes DM, Keys TF et al. Nosocomial clonal dissemination of methicillin-resistant Staphylococcus uureus: Elucidation by plasmid analysis. Arch Int Med 1987; 147: 521-524. Gaston MA, Duff P’S, Naidoo J et al. Evaluation of electrophoretic methods for typing methicillin-resistant StaPhvlococcus aweus. 7 Med Microbial 1988: 26: 189-l 97. Freeman R, Sisson PR, Lightfoot NF, McLauchlin J. Analysis of epidemic and sporadic strains of Listeria monocytogenes by pyrolysis mass spectrometry. Letters in Applied Microbiology 1991; 12: 133-136. Magee JT, Hindmarch JM, Burnett IA, Pease A. Epidemiological typing of Streptococcus pyogenes by pyrolysis mass spectrometry. r Med Microbial 1989; 30: 273-278. Freeman R, Goodfellow M, Gould FK, Hudson SJ, Lightfoot NF. Pyrolysis mass spectrometry (PyMS) for the rapid epidemiological typing of clinically significant bacterial pathogens. J Med Microbial 1990; 32: 283-286.
48
F. K. Gould
et al.
10. Orr K, Gould FK, Sisson PR, Lightfoot NF, Freeman R, Burdess D. Rapid inter-strain comparison by pyrolysis mass spectrometry in nosocomial infection with Xunthomonas maltophilia. J Hosp Infect 1991; 17: 187-195. 11. Freeman R, Gould FK, Sisson PR, Lightfoot NF. Strain differentiation of capsule type 23 penicillin-resistant Streptococcuspneumoniae from nosocomial infections by pyrolysis mass spectrometry. Letters in Applied Microbiology 1991; 13: 28-3 1.
of Hospital
Infection
(1991)
19, 41-48
Inter-strain comparison by pyrolysis mass spectrometry in the investigation of Staphylococcus aurezzs nosocomial infection F. K. Gould*,
R. Freeman?, P. R. Sisson$, N. F. Lightfoot
B. D. Cookson§
and
*Department of Microbiology, Freeman Hospital, f-Medical School and $Regional Public Health Laboratory, Newcastle upon Tyne and §Division of Hospital Infection, Central Public Health Laboratory, Colindale Avenue, London Accepted for publication
23 July
1991
Summary:
Pyrolysis mass spectrometry (PyMS) was used to examine isolates of Staphylococcus aureus from an outbreak of wound infections on a cardiothoracic surgical unit, some of which were thought to have been related to a point-source in the operating theatre. The PyMS results were compared with the results of phage typing. Both methods suggested that a single strain of S. am-em, of phage pattern 29/52/52A/79/80/81, was responsible for some of the wound infections. but PvMS also identified two natients with nhaee non-typable isolates. Ph’age typing indicated four staff members as p&siGe carriers of the epidemic strain, but PyMS indicated only two. Epidemiological enauirv confirmed that one of the two members of staff identified bv both methods was likely to have been the source of the theatre-based infection. PyMS is a rapid and relatively inexpensive technique for the investigation of nosocomial S. aureus infection and was more discriminatory than phage typing in this instance. Keywords: infection.
Pyrolysis
mass spectrometry;
Staphylococcus
aureus;
nosocomial
Introduction of outbreaks of wound infection due to Staphylococcus a common feature of hospital cross-infection control. A particularly important aspect is that of tracing a common source when epidemiological evidence suggests that a member of staff may be carrying and shedding an epidemic strain. In such investigations samples from apparently affected patients, other patients, staff and the environment will inevitably yield many isolates of S. aweus. Accurate characterization of these isolates is necessary to identify the putative epidemic strain. Phage The
investigation
aureus remains
Correspondence Newcastle upon
to: Dr R. Freeman, Tyne NE2 4HH.
01954701/91/090041
f08
Microbiology
Department,
$03.00/O
Medical
School,
0
41
Framlington
1991 The Hospital
Infection
Place,
Society
42
F. K. Gould
et al.
typing is the usual method employed since isolates of different phage types are unlikely to be related. Conversely, those of similar or even identical phage type are not necessarily a single strain and under such circumstances the method may provide insufficient discrimination. Phage typing is usually only performed in reference centres and the results may not be available immediately. The detection of a common source outbreak of nosocomial S. aUreUS infection and the institution of appropriate control measures may, therefore, be delayed. Pyrolysis mass spectrometry (PyMS) gives useful and rapid inter-strain comparisons for many bacterial species, including staphylococci.’ The method is based on the principle that pyrolysis of pure bacterial cultures generates mass ion spectra which are characteristic of the individual pyrolysed bacteria. Computerized analysis of the data is used to select those mass ions showing the best discrimination between the various spectra. These form the basis of a mathematically-derived analysis of similarities and differences which is then used for inter-strain comparison. In essence, the system produces a similarity matrix, using the selected mass ions as unweighted characteristics. At present the system remains comparative and reproducibility is restricted to individual batch examination.2 We now report the application of PyMS to the investigation of a presumed outbreak of nosocomial S. aUreUS infection. Materials
and methods
The outbreak In November 1990 an increase in S. aUreUS sternotomy wound infections was noted in the Cardiothoracic Centre. Surveillance records from the preceding several months indicated that the increase was recent. The affected patients were distributed across several convalescent wards and included at least one patient on the Cardiothoracic Transplant Unit as well as patients undergoing conventional cardiac surgery. Initial epidemiological investigation failed to establish whether the increase was due to a failure in postoperative wound technique or due to a point source, but several patients developed their infection soon after operation, suggesting the possibility of intra-operative contamination. In anticipation of the delay expected before phage typing results of the original isolates, it was decided immediately to organize nasal swabs from medical and nursing staff in the wards and the theatre suite. Any S. aureus isolates of similar sensitivity (penicillin resistant, otherwise fully sensitive) to those present in the postoperative wounds could then be examined at the same time. In all, 46 isolates (31 patient isolates, including two isolates from patients on an unrelated orthopaedic ward, and 15 isolates from the nares of members of staff) were collected and sent for phage typing. All 46 isolates were also examined by PyMS. Sample preparation and PyMS analysis The 46 cultures were encoded and examined
by two of the investigators,
Pyrolysis
of nosocomial
S. auf-em
43
who were unaware of the epidemiological data or of the subsequent results of phage typing until analysis was completed. Each slope was subcultured in duplicate onto plates of a single batch of diagnostic sensitivity test agar (DST; Oxoid) and incubated overnight at 37°C. Smears of each subculture in triplicate were made on pyrolysis foils (Horizon Instruments Ltd., Heathfield, Kent). The foils were inserted into pyrolysis tubes (Horizon Instruments) and dried in a hot-air oven at 80°C for 5 mins. They were then pyrolysed at 530°C as a single batch on a Horizon Instruments PYMS 200X pyrolysis mass spectrometer, as previously described.3 Integrated ion counts at mass intervals of 11 to 200 were recorded on floppy disk with the pyrolysis sequence number and total ion count for each sample. The duplicate subcultures were given consecutive PyMS numbers such that isolate 1 was pyrolysed as l/2; isolate 2 as 314, and so on.
Data analysis As a first step, differences in spectra solely attributable to variations in the amount of material pyrolysed per foil were identified and the data were corrected (normalization). The normalized spectral data for each subculture were analysed for within-group and between-group variation. Discriminant analysis was then performed on two separate datasets, one comprising the data from all the even-numbered subcultures and the second containing the data from the odd-numbered subcultures. In each set the 30 most discriminatory mass ions were subjected to principal component (PC) and canonical variate (CV) analysis. The meaned results of the triplicate samples for each subculture were then represented on an ordination diagram of PCCV 1 versus PCCV 2. The two resultant ordination diagrams (one for even-numbered and one for odd-numbered subcultures) were examined for clusters of apparently closely related isolates. Editing of the database to produce datasets containing only those isolates contained in such a cluster allowed further analysis for similarity and difference within the cluster. When the number of isolates to be thus analysed was less than 10 a final analysis was performed combining data from both sets of subcultures. The end-point of this analysis was a number of isolates in which the means of the triplicate analyses for each subculture of the same isolate were shown to differ as least as much from each other as from different isolates. Such isolates were regarded as indistinguishable by PyMS. The principles of the analysis have been described previously.3
Phage typing Phage typing was carried out at the Division of Hospital Infection, Central Public Health Laboratory, Colindale, London, by standard methods,4 using the International Standard Set of phages.
44
F. K. Gould
ef al.
Results
The ordination diagram produced from the analysis of the odd-numbered subcultures is seen in Figure 1. Two apparent clusters of isolates are seen. The larger diffuse cluster contains data from 25 patient isolates and 13 isolates from staff samples, and the smaller cohesive cluster contains data from six patient isolates and two staff samples. A similar ordination diagram was produced from the analysis of the even-numbered subcultures (not shown), and the distribution of isolates between the two clusters was the same. Analysis of the combined data from both subcultures of the isolates in each separate cluster showed that the isolates within the larger diffuse cluster were distinct since the differences between different isolates were always greater than the differences between two subcultures of the same isolate. In contrast, the analysis of the smaller cluster showed that these eight isolates were indistinguishable by PyMS since the differences demonstrated between means of triplicates of two subcultures of the same
;----\ , I’ ’ /’ ; ,’ ::
I’
n
,’ ’ ,’ ;
CPm
q+ n
:: ! 25 patients
:’ *\
: 113 staff
‘:
I’ : :
-10.2 t
-16.0 h__, -16
-8
0 Combined
8 PCCV group
16 score
24
:
I
Figure 1. Ordination diagram of the first two principal component canonical variates (PCCV 1 and 2) for odd-numbered subcultures of 46 isolates of Staphylococcus aweus examined by pyrolysis mass spectrometry. The distinct cluster seen on the right contains six patient isolates (solid squares) and two staff isolates (open squares).
Pyrolysis
of nosocomial
45
S. aureus
isolate were at least as great as any differences shown between the means of triplicates of any two separate isolates. The six patient isolates thus identified comprised five isolates from sternotomy wound infections on the affected wards in the cardiothoracic centre and one isolate from an orthopaedic ward. The two staff isolates in this same closely related cluster came from nasal swabs of, respectively, a theatre nurse and a cardiopulmonary bypass technician. The results of phage typing showed that 43 of the 46 isolates were typable, either at RTD or RTD x 100. Only one phage pattern (susceptibility to phages 29/S2/52A/79/80/81, either as a complete pattern or with less than two strong differences) was found in both staff samples and in some of the patient isolates and was, therefore, potentially that of an epidemic strain with a point source in the staff. On this basis four staff members and three cardiothoracic patients, together with one orthopaedic patient, were potentially linked. All four members of staff were treated with mupirocin nasal ointment, which eliminated the organism in each case. The results of both investigations are summarized in Table I, from which it can also be seen that six of the eight isolates found to be indistinguishable by PyMS were of phage pattern 29/52/52A/79/80/81 but two were phage non-typable. The other 38 isolates, shown by PyMS to be not only distinct from the clustered isolates but also from each other, included two isolates phage typed as 29/52/52A/79/80/81 and a varied collection of other phage groups and types. Both methods identified the same isolate from an orthopaedic patient epidemiologically unrelated to the cardiothoracic infection incident as being indistinguishable from the epidemic strain.
Table
Isolate
I. Comparison of results of phage typing and pyrolysis mass spectrometry analysis isolates of Staphylococcus aureus shown to be similar by either OY both methods Phage type 29/52/52A/79/80/81
source
Patients Cardiothoracic Orthopaedic Cardiothoracic Cardiothoracic Cardiothoracic Cardiothoracic
patient wound patient wound patient wound patient wound patient wound patient wound
Staff A. Surgeon B. Theatre nurse C. Theatre nurse D. Cardiopulmonary bypass technician
PyMS
analysis
Non-typable Yes Yes Non-typable Yes Yes
Clustered Clustered Clustered Clustered Clustered Clustered
Yes Yes Yes Yes
Not clustered Clustered Not clustered Clustered
on ten
result
46
F. K. Gould
et
al.
When both sets of results were compared with the epidemiological evidence it was found that staff members A, B and C were unable to be patients, but D (a circumstantially linked to all the identified cardiopulmonary bypass technician) had been present in the theatre suite on all relevant occasions, albeit briefly for some. A particularly important finding was that A, a surgeon, had not been present at any of the relevant operations. As part of the resultant enquiries it also emerged that traffic in and out of the cardiothoracic theatre suites had been unusually, and to some extent, unnecessarily, busy during the studied period and this may have contributed to the general increase in postoperative infections in the cases which both methods demonstrated to be unrelated to the common source. Discussion
Both methods (phage typing and PyMS) suggested that there was a small common-source outbreak of S. LZUY~ZLSwound infection that was contributing to the overall increase in staphylococcal infection in the cardiothoracic surgery patients. The results of phage typing showed that three cardiothoracic patients were infected with S. UUY~US type and that staphylococci of the same phage type were 29/52/52A/79/80/81 present in the nasal swabs of four members of staff. Since it was impossible further to distinguish between the isolates by this method all four staff members were treated with mupirocin, although it was clear from enquiries that three of them were probably not the point source of the outbreak and may themselves have become passively colonized by an outbreak strain. Follow-up studies have shown no evidence of further point-source infections due to S. uureus in sternotomy wounds. Comparison of the two methods shows that PyMS identified the same patients as phage typing but added two more patient isolates which were phage non-typable. PyMS also identified the same staff member as the point source, but with greater discrimination, the relevant member of staff being one of only two possibilities rather than four. The inclusion of phage non-typable isolates of S. UUY~USin a single strain identified by other methods, other members of which are phage typable, is not unexpected. Similar observations have resulted when S. aureus isolates have been examined by other methods, including restriction endonuclease analysis.’ Conversely, the possibility that isolates of S. aUreUS belonging to an identical phage susceptibility pattern are not necessarily a homogeneous strain has recently been shown in studies comparing the results of phage typing and electrophoretic typing methods in methicillin-resistant S. uweza (MRSA).‘j Both phenomena were also demonstrated in a PyMS study of an epidemiologically well-characterized outbreak of Listeriu monocytogenes infection.7 PyMS is not a typing method since it does not assign a permanent characterization to the examined organisms. Pyrolysis mass spectrograms
Pyrolysis
of nosocomial
S. aureus
47
vary with the duration and conditions of incubation of the organisms, leading to variability between batches. When, however, the organisms are prepared under identical conditions and analysed in a single machine batch, comparisons between isolates are objective and highly discriminatory. PyMS inter-strain comparison allows the identification of closely-related groups of isolates from within a larger collection of similar organisms and provides the essential information necessary to detect an outbreak, indicate the likely source and, therefore, institute control measures. PyMS is rapid and consumable costs are cheap in comparison to conventional techniques. Although the method is currently a research tool and the capital costs are c. Es0 000, the consumable costs are less than El per sample. The same apparatus and reagents can be used for a wide range of different bacterial species and the same method has been successfully applied in outbreaks due coagulase-negative staphylococci,’ to Streptococcus pyogenes,8 Salmonella,’ maltophilia’O penicillin-resistant and Listeria,7 Xanthomonas pneumococci. l1 This small study suggests that PyMS can also be used for the investigation of nosocomial S. aureus infection. The relative merits of PyMS and other new methods of strain characterization such as restriction fragment length polymorphism (RFLP) have yet to be established. More extensive application will be necessary before the value of PyMS in S. aweus epidemiology is determined. However, the speed, comparatively low cost and versatility of PyMS make it an ideal method for the initial screening of suspected outbreaks of allowing more expensive and time-consuming nosocomial infection, methods to be used more cost-effectively when formal typing is still necessary. References 1.
2. 3. 4. 5. 6. 7. 8. 9.
Freeman R, Gould FK, Wilkinson R, Ward AC, Lightfoot NF, Sisson PR. Rapid mass spectrometry of coagulase-negative inter-strain comparison by pyrolysis staphylococci from persistent CAPD peritonitis. Epidemiology and Infection 1991; 106: 239-246. Pitt TL. Editorial: Burnt offerings or typing system? J Hasp Infect 1991; 17: 157-158. Aries RE, Gutteridge CS, Ottley TW. Evaluation of a low-cost, automated pyrolysis mass spectrometer. J Anal Appl Pyrol 1986; 9: 81-98. Blair JE, Williams REO. Phage typing of staphylococci. Bull Wld Hlth Ovg 1961; 24: 771-784. Rhinehart E, Shlaes DM, Keys TF et al. Nosocomial clonal dissemination of methicillin-resistant Staphylococcus uureus: Elucidation by plasmid analysis. Arch Int Med 1987; 147: 521-524. Gaston MA, Duff P’S, Naidoo J et al. Evaluation of electrophoretic methods for typing methicillin-resistant StaPhvlococcus aweus. 7 Med Microbial 1988: 26: 189-l 97. Freeman R, Sisson PR, Lightfoot NF, McLauchlin J. Analysis of epidemic and sporadic strains of Listeria monocytogenes by pyrolysis mass spectrometry. Letters in Applied Microbiology 1991; 12: 133-136. Magee JT, Hindmarch JM, Burnett IA, Pease A. Epidemiological typing of Streptococcus pyogenes by pyrolysis mass spectrometry. r Med Microbial 1989; 30: 273-278. Freeman R, Goodfellow M, Gould FK, Hudson SJ, Lightfoot NF. Pyrolysis mass spectrometry (PyMS) for the rapid epidemiological typing of clinically significant bacterial pathogens. J Med Microbial 1990; 32: 283-286.
48
F. K. Gould
et al.
10. Orr K, Gould FK, Sisson PR, Lightfoot NF, Freeman R, Burdess D. Rapid inter-strain comparison by pyrolysis mass spectrometry in nosocomial infection with Xunthomonas maltophilia. J Hosp Infect 1991; 17: 187-195. 11. Freeman R, Gould FK, Sisson PR, Lightfoot NF. Strain differentiation of capsule type 23 penicillin-resistant Streptococcuspneumoniae from nosocomial infections by pyrolysis mass spectrometry. Letters in Applied Microbiology 1991; 13: 28-3 1.