Epidemiological studies o f noso c o m i a l infections with Pseudomonas aeruginosa using a D N A p r o b e

Epidemiological studies of nosocomial infections with Pseudomonas aeruginosa using a DNA probe. Can J Infect Dis 1992;3(6):299-306. A DNA probe encoding the Pseudomonas aeruginosa pilin gene has been developed in the authors· laboratory and has been shown to be a useful epidemiological tool. In the present study this technology. together wiLh other typing methods. has been used to clennc relatedness and po siblc transmission routes of P aeruginosa strains isolated in several hospital wards. Clusters of P aeruginosa infec!ions. suspected to be the result of no. ocomial transmission. clcvclopccl in a general intensive care unit (ICU) and a neurosurgical warci/ ICU. as well as in a burn unit. were tudiccl using antibiograms . lipopolysacchariclc-scrotyping. and gene probe analysis. Results of these tuclics demonstrated that each of the general and neurosurgical ICU isolates were different. making nosocomial transmission very unlikely. However. wiU1in the burn unit. patient isolates had identical profiles. suggesting that spread between patients was occurring or that a common source of infection was present. Changes in infection cont rol measures within the unit were introduced and may have contributed to eradication of the outbreak DNA probe studies were \·aluable in clarifying cpiclemiological relatedness of isolates that was not evident wilh the other typing strategies and idcnlined a po siblc bum-associated strain. RESUME: Une sonde d"ADN pour l'cncocla(jc du gene de Pseudomonas aeruginosa a etc mise au point clans cc Jaboraloire et s·csl revclee un outil cpiclcmiologiquc utile. Dans Ia prcsentc elude. cctle technologic. de memc que cl'autres methodes de lypage. ont etc utilisces pour clcnnir lcs interrelations et lcs voics de transmission possibles de souches de P. aeruginosa isolees clans diffcrcntes unites hospitalieres. Des infections a P. aeruginosa. presumemcnt attribuables a unc transmission nosocomiale, om de observees dans unc unite de soins intcnsifs generaux. clans unc unite de soins imcnsifs neurochirurgicaux et dans unc unite de brt:lles. ou ellcs ont etc etuclices a !'aide cl'anlibiogrammcs. de mCLhocles de typage serologique par lipopolysaccharides ct cl'analyscs par sonde gcnctiquc. Lcs rcsultats de ces etudes ont dcmontrc que chacun des isolats prelevc dans les unites de soins inlensifs gencraux ctncuro-chirurgicaux Clail diiTerent. ce qui a permis d'ccarter !'hypothese de Ia transmission nosocomialc. Cependanl. au sein de !'unite des brt:lles. isolats

P SEUDOMONAS AERUCINOS/\ IS A MAJOR OPPORTUN ISTI C pathogen in immunocompromised individuals. and patients with cystic fibrosis and lhermal trauma. Though recognized as a prevalent organ ism in U1e hospital environ m ent a nd a re latively frequent cause of nosocomial disease. Lhe reservo ir of U1e organ ism and U1e precise mode of transmission wilhin hospita ls are usually unclear.
Existing meti1ods for typ ing strains of P aeruginosa include plasmid analysis. bacteriocin (pyocin) typing. phage sensitivity. biotyping. colonial morpho logy. antibiotic sensitivity patterns and the use of DNA probes. Serotyping generally is accepted as the most practical and reliable for routine use in clinical microbiology laboratories and is based on an agglutination reaction between bacterial lipopolysaccharide (LPS) and typespec ific rabbit antisera (l ). While most clinical isol ates may be typed using Lhis method. strains from cystic fibrosis patients are unique in that the majority po lyagglutinate or are nontypab le in 0 -typ ing sera (2) . In addition. il has been reported previously that LPS serotype may not be a stable ep idemiological marker and may change over lime (3.4).
A DNA probe encoding the P aeruginosa strain PAK pilin gene has been developed (5) and is used to probe restriction enzyme digested P aeruginosa chromoso mal DNA yieldi ng strain-specilk restriction fragment lengti1 po lymorphism (RFLP) pallerns . It has been repo rted that serial P aeruginosa isolates from a patient with cystic fibrosis maintained a constant RFLP pattern and pilin gene sequence despite considerable phenotypic variability (4). In add ition. recent stud ies of over 300 P aeruginosa isolates from cystic fibros is patients and a variety of olher sou rces have demonstrated 21 RFLP patterns and have shown U1e p ilin gene probe to be a u efu l ep id emiological tool (6). The occurrence of clusters of P aeruginosa infections in hospilalizecl patients that were closely related ep idem iological ly provided U1 e unique opportuni ty lo evalu ate U1e pil in gen e probe in the study of nosocomial ep id em iology. The goa ls of t.his study were to determine if nosocomial spread of P aeruginosa was occw-ring and to compare the pi! in gene probe wilh LPS-serotyping as ep id em iological typing techniques in U1is selling.

Infections with Pseudomonas aeruginosa:
The University of Alberta Hosp itals is a 1250-becl tertiary ca re facility representing a ll areas of clinica l med icine. including long term ca re fac ili ties and separate intensive care units (ICUs) for neonates. pediatrics. neurosurgery. coronary care. ca rdi ovascular surgery. burns and a genera l surgical-med ical !CU. Clu sters of infections \-vilh P aeruginosa were observed over a s hort pci-iod of Lime \viti1in ti1ree of the hosp ita r s ICUs (Table  I). These three units-Lhe gene ral intens ive ca re unit (G ICU). neurosurgica l ward/ICU (NW ICU) and the burn 300 intens ive ca re unit (BICU) -are physically remote from each other within the hospital. Clusters included at least Lhree epidem iologically related patients s imultaneous ly or sequential ly hospitalized within one of U1e ICUs. from whom P aeruginosa was isolated from an infected s ite. Pneumon ia was defin ed as t h e presence of pu rulent sputum (grossly and microscopically) \vi.Lh or \vi.Lhoul fever associated wilh a chest x-ray infiltrate.
The GlCU is a 16-bed unit consisting of 10 open beds and six private rooms. Nosocomial P aeruginosa pneumonias developed in three mechanically ventilated patients over a 10-day period. with patient GICU-1 a lso having a sternal wound infection \vi.th this organism follo\.ving coronary artery bypass surgery (Table 1). These three patients occupied U1ree of four adjacent private rooms \vi.lh patients GICU -1 and GICU-2 sideby -s ide and \vi.th a noncolonized patient in Lh e room separating patients GICU-2 and GlCU -3. Patient GICU-X was transferred from anoti1er institution d uring Lhis outbreak and was colonized \viti1 P aeruginosa upon adm ission to t he GICU . The NWICU consists of a 24-bed ward \vi.Lh a n attached lO -bed !CU. This outbreak involved one venlilated an d two unvenlilated patients. all of whom developed P aeruginosa nosocomial pneumonias over a period of one week.
The BICU is a lO-bed self-contained unit in which all necessary care of Lhe burn patient. including venUlalion and hydroU1erapy. is carried out. The nature of Lhis outbreak is described unde r 'Results ·.
Bacterial strains: P aeruginosa strains isolated from patients involved in clusters of nosocomial infections. as well as concurrent isolates from h ospitalized patients nol involved in outbreaks. were obtained from Lhe clinica l microbiology laboratory of ti1e University of Alberta Hospitals. Clinical specim ens including sputum. trach eal secre tions. a nd wo und a nd burn swabs were p lated on sheep blood and MacConkey aga r plates. Blood cu ltures were innoc ul ated into broth a nd growth was detected by Lhe BACTEC NR660 system (Becton Dic kin so n. Maryland). Isolates were identified by conventional methods including standard tests for oxidase. growlh at42°C and fluorescence under ultraviolet light. One iso late from each of patients BICU 1. 2 and 3 was inclu ded in ti1e ep idem ioloO"ical survey repo rted previously (6). Antibiograms: Sensitivi ty testing of a ll isol a tes was performed in ti1 e clini cal microbiology laboratory of th e University of Alberta Hospita ls u si ng U1e automated Vitek micro broth dilution technique (Vitek Systems Inc. Missouri) . Table 2 lists the antib iotic susceptibility profil es observed for isolates in Lhis study. Lipopolysaccharide 0-serotyping: All strains were se rotyped using ti1e International Antigenic Typing System (DIFCO. Michigan) in e iU1er the Provincial Laborat01y for Public Health of No rthern Alberta o r Lhe Department of Pediatrics. University of British Co lumbia. Isola tes  6) whjch agglu tinated in more than one typing serum were designated as polytypeable. DNA probe studies: Ch romosomal DNA was prepared from late-log phase bacteria usinO' a modWcation of U1e procedure of Coleman et al (7). Purified DNA (1 to 2 pg) was digested wi.th th e restriction endonucleases PsU and Hindiii (GIBCO Laboratories. New York). electrophoresed in 0. 7% agarose and transferred lo ni Lrocellulose paper by the m e thod of Southern (8). A 1.2 kilobase Hindiii fragment. of PAK chromosomal DNA (5) was nick t.ranslaled and used to probe the nilrocellulose filters under high stri ngen cy conditions (9). washing a l 21 °C rather than 65°C. Au toradiographs we re U1en exposed at -70°C . Fragments hybridizing with U1e radiolabelled probe were s ized by comparison wi.lli m olecular weight standa rds run togeth er with U1e sarnp les in th e agarose gel. Twenty-one RFLP pallems were defined in screen ing large numbers of isolates ar1d were reported previously (6). Experience wiU1 lliis techniqu e has demonsLrated complete reproducibility ofRFLP patlems a long as ge nomic DNA is digested to completion and Southern lransfer is adequate (4.6). In lliis study. a techn ica lly satisfacto ry result was gene ra lly obtained on the first. attempt. wiili occasional isolates requiting a second or co nfirmatory testing.
Selecled s lrains were also typed with a second DNA probe to clarify a pparent iliscrepancies in resulls belween antibiograrn . LPS serotype and pilin gene probe. Probe U. a 741 base pair fragm ent en coiling U1e region upslream of U1e P aeruginosa exotoxin A structural gene (10) . vvas isolated from a 5% acrylam id e gel (9) following digestion of plasmid pBRTox wiili Psll and Nrul (G IBCO Laboralories) . SouU1ern blots were prepared of Bgiii-a nd Xhol-(G IBCO) digested c hromosoma l DNA a nd were probed. washed and exposed to x-ray film as described a bove except U1al ilie 741 base pair probe was rarno la belled using U1e random primer lechnique ( 11) . Results a re expressed as U1e s ize of probe reactive fragments in ldlobases for ilie Bglll/Xhol diges tion s in Table l. Table 1 s ummarizes the resu lts of ilie U1 ree clusters investigated in ili is s ludy. including the antibiotic sens ilivity profiles. serotypes and RFLP pallerns oblain ed in DNA probe slud ies . General ICU: DNA probe studies revealed U1at all four patients involved in this outbreak -including the three patients who sequentially became co lonized while in this unit as well as the patient who wa transferred from another institution -were co l onized w i th diffe r ent train of P aeruginosa. This cone! u sion is supported by the antib iocrram patterns and LPS-serotyping.

RESU LTS
Four isolates were taken over a three-week period and from two different sites from patient G lCU-1. All isolates were identical in DNA probe studies. suggesting that this patient remained colon ized with a single train of P aeruginosa througho u t. the study peri od . Slight. variab ili ty in U1e anlib i ogram pa tterns raised lhe posibilily of colonization wiU1 two different. str ain s while LPS-erotyping reveal ed the iso lates from palient. G lCU -1 to be polytypeab le. 1\vo colon ies of P aeruginosa \Vith different colon i al morphologies were isolated from a single sputum specimen for patient G lCU -2. While LPS-serotyp ing suggested that Lhe patient was co lonized wiU1 two different strains. DNA probe tuclie with bolh the pilin gene and probe U demonstrated Lhat U1is was mo t p r obab ly a single strain. Neurosurgical ward ICU: All three patients involved in Lhis cluster of infections were co lon ized by different. strain of P aeruginosa (indicated by each of the t hree t.udy techniques). 1\vo colonies wiU1 differing morphologies were isolated from the sputum specim en of patient NWICU -2 and. in contrast. to patient G ICU -2. LPS-erolyping and D A probe studies concurred that these two colony types represented a ingle strain. Burn unit: An outbreak of burn wound sepsis in Lhe burn unit was first recognized when palient B ICU -2 302 devel oped septic shock from bacteremia with a r esistant str ain of P aer-uginosa: wiU1in a few days anoU1er patient in Lhe burn unit (patient BICU-3) had al so become co loni zed with a simil arly resistant str ain (Table 1). LPS -se rotyping and DNA probe studi es using bolh probes demonstrated that U1ese two patients were pr obab ly colonized wiU1 U1e identical slrain . In retrospect. the r e had been a patient in the bum unit three monU1s earlier infected with a resistant strai n of P aemginosa (patient BI CU-1). Studi es of Lhis isolate revealed that. it. too. was U1e identical strain. Environ m enta l sampling was perfoi-med and ch anges were made in U1e routine care of patients in the burn unit to prevent further transmission (see D iscu ssion).
'[\vo other patients became colonized with P aeruginosa in the two monU1s s ubsequent to the recognition of lh i s outbreak. Patient BICU-4 was col on ized with a different strain as demonstrated by all three typi ng techniques. Patient Bl U-5 was colonized by a strain wilh the same serotype (0 11) as U1e ea rlier ou tbreak strain . However. DNA probe studies using boU1 probes in add ition to U1e antibiog r am patterns indicated U1at U1is represented a different and uniqu e str ain. Typing of P aeruginosa isolates from U1e first bacteremic patient following recognition of lliis outbreak (patient B ICU -6) also revealed a str ain which was different from the epidemic strain . Background strains: 1\vo spo rad ic cli n ica l isolates were avai l abl e from P aer-uginosa-infected patients hosp i tal ized during th i s period but unrelated to the identified outbreaks. One. an eye isolate from a patient. wiU1 conjunctivitis . was typed as RFLP pattern 6 with U1e pili n gene probe. while U1e second. a sputum isolate from a m echanically ventilated patient in the cardiovasc ul ar !CU . could not be typed with lh is probe despite rep ated attempts. This was unusual as less than 5% of isolates have been fo u nd to be nontypeable in U1is system (6). It does. however. suggest Lhat this isolate was different from others studi.ed in this report. Isolates obtained from a GICU patient one year prior to U1e recognition of the cluster were also analyzed and found to be type 5. No environmental isolates obtained from hospital sources during this pe1iod were available for typing.

DISCUSSION
P aeruginosa is a ubiquitous organism found in soil. plants. water and food . Indeed. 5 to 29% of nonnal individuals are colonized in their gastrointestinal tracts (12). A recent study (13) found that 22.6% of patients were colonized with P aeruginosa on admission to a general hospital. with an additional 10.8% acqu iring the organism during their hospitalization. Within U1e hospital. potential sources of paU1ogenic organisms are numerou and include U1e patient's own microflora. lhose in U1e hospital environment or canied by medical personnel. contaminated equipment and other patients. It generally is agreed that colonized and infected palients are the major reservoir of resistant Gram-neo"alive organisms within hospitals (14.15). There is also considerable evidence U1at spread of such resistant strains between hospitalized patients occurs on U1e hands of health care personnel. 30 to 40% of whom carry Gran1-negative organisms on their hands (some of them continuously) (16)(17)(18)(19)(20). Gastrointestinal acqu is ition and carriage of Gram-negatives appears to be an important intermediate step in certain situations (14). while some patients are self-infected from their own gastro intestinal microflora (21). A gastrointestinal reservoir was fe lt to be important in one prospective study in which transfer of P aeruginosa between surgical patients in a hospital ward was demonstrated (22) .
Previous studies of ICUs have demonstrated conflicting results related to cross-infection of P aeruginosa between patients. Large epidemics related to contaminated respiratory therapy equipm ent have been reported in U1e past (23). but are rarely a problem today. Cro s -infection of P aeruginosa between patients in one prospective ICU study was attributed to heaiU1 care personnel. as 24% of h a nd cultures and 30% of uniform cultures grew this organism (24). On lhe other hand. prospective studies in a spinal cord unit (25) and a surgical-medical ICU (26) concluded that cross-infection wilh P aeruginosa is not a major problem in these settings. Clearly. the validity of all such studies depends on U1e reliability of the strain typing systems employed.
The present authors have investigated clusters of nosocomial infections wiU1 P aeruginosa in three reus in which transmission between patients was strongly Nosocom ia l infections with P aeruginosa suspected on clinical grounds. The results demonstrated U1al the patients involved in U1e GTCU and NWlCU outbreaks vve re each infected with a different a nd unique strain. revea ling U1atlransmission between patients was not a factor in U1ese 1:\vo outbreaks. From an infection control standpoint it is important to ensure U1at pote ntially viru lent and multiply resistant Gramnegative organisms are not being spread between hospita lized patients. ll was not possible in this study to exam ine potential sources of the P aeruginosa colonizing the patients.
Of the currently available typing schemes for P aeruginosa. LPS -serotyping genera lly is considered to be U1e most practical. while colony morphology and antibiogram patterns are U1e least accurate (27.28). While reproducibility in testing serial isol ates has been questioned. antim icrobial susceptibil ity data assessment is likely lhe most readily available typing technique for most clinicians and was Ulerefore included in this investigation. In this study. multiple isolates from individual patients revealed s light variability in antibio-gran1 s. malting interpretation confusing. As such. it is difficult to establish firmly epidemiological relatedness between strains on U1is basis. FurU1ermore. rapid c hanges in antibiogram patterns frequently are seen in s ubpopulalions of P aeruginosa in response to antimicrobial therapy. furU1 er add ing to the difficulty in interpretation .
Isolates of P aeruginosa from noncystic fibro is patients can be reproducibly serotyped in 92 to 99% of ca es (2.29). In contr ast. 68 to 80% or more of isolates from cystic fibrosis patients are nontypeable or polytypeabl by LPS-serotyping (2.30.31). Development of 0 -antigen specific monoclonal antibodies (MAbs) for serotyping may circumvent U1is problem and provide improved discrimination between strains (32 -35). although reproducib ili ty for the MAbs was only 75% in a recent study (36). As noted earlier. changes in LPS serotype may occur in single strains over time (4.37). and some strains a re now known to synU1esize more Ulan one type of chemically and antigenically distinct LPS molecule (38). The reproducibility and discrimin -atOJy power of several typing methods has been compared. and use of more U1an one typing system has been advocated. at least for cystic fibrosis isolates (39. 40).
ll is unsual iliat P aeruginosa isolates from patient GICU-1 could not be serotyped while DNA probe studies demonstrate conclusively that this patient was co lonized wiU1 a single unique strain throughout the U1ree-week study period. Furthermore. serotyping isolates of different colony morphologies in patient GICU -2 suggested the coexistence of two different strains. while DNA probe studies indicated U1at U1is reflected phenotypic valiability in a single strain as has been demonstrated previously (1) . Rapid adaptation with a lteration in an organism's phenotype in response to environm ental stimuli Ulrough modulation of gene ex-pression is well -established (41). Rece ntly gen e tic rea rra ngem ent in P a eru.ginosa h as been described as a mechanism for a ltering exp ression of virul ence factors including LPS e rotype (42).
The burn unit outbrea k was ini tially recogn ized when patients BICU-2 and BI CU-3 we re sequential ly infected with a res ista nt strain of P aeruginosa. A.ntibiograms. DNA probe stud ies a nd serotyping demonstrated that a patient who h ad d ied of pse udomon as sepsis in the burn unit U1ree months earlier (patient BICU-1) was infected \vith t11e san1e strain . This s u ggested nosocomial transmission through cross-infection or. more likely. a common source of infection was present as there was no te mporal overlap between patients BlCU -1 and BICU -2 and -3. Mu ltiple enviro nmental cul tures from the burn uni t. including water sources and the hydrotherapy lank. failed to grow P aeruginosa except for a s ingle cu lture on one occasion from the scale used to monitor weights of all patients in U1e unit. Unfortunately. th is iso late was inadvertently discarded and cou ld not be typed to determine if tl1is wa s a potential reservoir of P aeru.ginosa witl1in the burn un it. However. use of U1 e scale was discontinued. as was t11e hydrotherapy routin e for burn victims. in the hope of preventing fu rth er sp read of U1is organism. 1\vo patients became colonized wiU1 P aeruginosa sho rtly after recognizing this cluster (BICU -4 a nd BICU-5): antibiogram and DNA probe stu dies of t11ese. a well as those from a s ubsequent bacteremic patient (BICU-6). were consistent wiU1 e radication of U1e outbreak strain. LPS -serotyping. however. cou ld n ot differentiate BlCU-5 isolates from th e ea rlie r outbreak stra in. Th is serotype (0 11) is known to be prevalent (23) and the discriminatory power of LPS-erotyping m ay not be su ffi cient fo r such hi.<rh frequency types (35). Monitoring of antibiograms and RFLP patterns in burn unit isolates has confirm ed the eradication oflh e multiresistant epid emic strain . while elimi nation of hydrotherapy i.n U1e routine care of the burn patient h as h ad a major in1pact on the frequen cy an d pattern of P aeru.ginosa colonization a nd the morbidi ty a n d mortality related to it.
While no definite association between RFLP types and disease states h ave been identified . U1e ep idemic strain detected in the bum unit cluster is of partic ular interest because of a possible predilection of tl1is RFLP pattern for burn patients. While other types have b een isolated from burn patients . RFLP type 7 appears to be a relatively infrequent type and has bee n isolated only from burn patients. Furthermore. U1is type has n ow been implicated in bum wound sepsis both in tl1is outbreak a s well as in a p rolonged outbreak involving a virulent multiresistant P aeruginosa isolate spanning 18 months in the burn unit of ano U1 er major city (6). While the possibility t11at RFLP type 7 may have an affin ity for bums or that it may h ave inc reased viru lence in this setting is intriguing . il rem ains specula -304 live a t present. In s upport of U1is possibility is U1e recognition U1al serious disease is caused by a small number of clones within a pailiogenic s pecies (43) .
Molecul a r biology has recently contribu ted novel techniqu es for U1e ide ntification of a wide variety of mic roorgan isms and the d ifferentiation of strains witl1in s pecies (44). Chromosomal DNA fingerprinting of P aeruginosa by res triction en zym e digestion witl1 ra re c u tung endonucleases a nd pulsed -field electrophoresis has been used to type polyagglutinable cystic fibrosis isolates (45) and to monitor U1e epidemiology of P aerug inosa in patients attending a cystic fibrosis clin ic (46). A DNA probe consisting of a 741 b ase pair Psll-Nru l fragment encoding t11e nucleotide sequ en ce upstream of U1e P aeruginosa exotoxin A gene h as been described ( l 0) and used in studying epid emiology in cystic fibrosis (3 .47.48). acute leukemia (49) and in distinguishing relapse from reinfection in a variety of infections due to P aeru.ginosa (37). Anoilier DNA probe en coding U1e N-lerminus of the exotoxin A structura l gene and its 5' flan king region has been described a nd used wiU1 a probe encoding ilie PAK pilin gen e to type a small number of isolates (50). Most recently. a combina tion of esterase electrophoretic typing and ribotyping has been applied to cystic fibrosis isolates h ighlighting U1e complexity of P aeru.ginosa colonization in this disease (51).
Few studies h ave used U1ese new and potentially more reliable techniques in hospital epidemiology to eval u ate reservoirs a nd m ech anisms of nosocomial tran s mi ss ion of P aeru.ginosa. Pulsed-field electrophoresis has been used to monitor ainvay colonization in a n ICU sellin g and h as demonstrated a lack of con tamination from environm ental sources or between patients (52) while. more recently. poss ible cross-contan1ination between immunocomprom ised children in a h ematological lCU with a gastrointestinal reservoir has been shown (53) . Multilocus enzyme electrop horesis (54) h as been used in a prospective study of a gen e ra l medicineon cology uni t which demonstra ted tl1al 10% of 283 patients a clmi tted over a six-month period acquired P aeru.ginosa in hospital (55). Of U1ese. 15% were likely cross-contamination from oU1er patients and 27% were linked to environm en tal sources. while the source for the remain ing 60% of nosocomi ally colonized patients was nol established.
In s ummary. by using DNA probe technology, pers isten ce of a common strain of P aeruginosa h as been dem onstrated in a bum unit infecting three patients over a U1ree-month period. suggesting possibile crossinfection or infection from a common source. While c ross-infection between patients in the GICU and NWlCU was felt to be likely by virtue of U1e proximity in lime and space of U1ose involved. the data h a ve shown that patients involved in these 1:\vo clusters were each infected with a different stra in. Wh en multiple isolates from individual patients were typed . the hybridization pallem remain ed constant. s u ggesting that each patient was co lonized with a s ing l e strain at a ll sites and U1roughout the study period . in spite of conside rabl e phenotypic vari ability including a ntibiotic suscept ibili ty a nd LPS serotype. Results of RFLP typing of sel ected i so l a t es with a second DNA probe a hvays concu rred with r es ults obtained using U1 e pLlin gen e probe .
In gene r a l. LPS-serotyping and a ntibiog r a ms in this study a ll owed for discrimina tion between unrelated i solates obtained from differ ent patients . but varia bility in sequential isolates from individual patients would make conclusions based on these techniques difficult. Nosocomial infections with P aeruginosa tial r o l e for DNA probe stud i es in establis hing r el ations hips between i sol a t es whi ch a r e epidemiologica lly r el a t ed b ut ap pear to b e different (based on antibiogram a n d se r otyping) . while a l so disting uishing between unr el ated i sol ates exhibiting highly preval ent serotypes such as type 011. Di scrimin ation b et'Neen strains with Lh e pilin gene probe mav b e l ess than ideal. howeve r . as 80% of s trains b elong Lo types 1 to 5 (6). Of furU1er con ce rn i s U1 e possib il ity U1at genom i c r earr a n gements may l ead to ins ta bility of RF'LP p atterns (56). allliough the use of two independent. probes in tilis study directed at d ifferent r egion s of U1e genome str engU1ens tile conclusions regarding r el atedness between strains.