Transient in vivo selection of a constitutively cephalosporin resistant Enterobacter cloacae causing ventriculitis

Presented in part at the Royal College of Physicians and Surgeons Annual Meeting, Ottawa, Ontario, September 11 to 14, 1992 as abstract N-676 Alberta Children’s Hospital and The University of Calgary, Calgary, Alberta Correspondence and reprints: Dr T Jadavji, Alberta Children’s Hospital, 1820 Richmond Road SW, Calgary, Alberta T2T 5C7. Telephone (403) 229-7813, Fax (403) 229-7221 Received for publication February 14, 1994. Accepted June 14, 1994 A JOFFE, A KABANI, K RAMO TAR, W KRU LICKI, G CAD RAIN, T JADAVJI. Tran sient in vivo se lec tion of a con sti tu tively cepha lo sporin re sis tant En tero bac ter cloa cae caus ing ven tricu li tis. Can J In fect Dis 1995;6(1):4448. A case of neo na tal ven tricu li tis com pli cat ing a ven tricu loperi to neal shunt and caused by one strain of En tero bac ter cloa cae (as shown on pulsed field gel elec tro pho re sis) is pre sented. Daily ven tricu lar fluid cul tures from day 1 to 9 re vealed in duci ble cepha lo sporin re sis tance in all iso lates ex cept on days 3, 4 and 5 of ther apy when iso lates were con sti tu tively re sis tant. This emer gence of re sis tance due to con sti tu tive Bush class 1 betalactamase pro duc tion is an ex cel lent ex am ple of the rapid emer gence of a pre domi nant strain of bac te ria de pend ing on an ti bi otic se lec tion pres sures in vivo. The tran sient na ture of the pre domi nant re sis tant phe no type may have been due to miss ing a dose of ce fo taxime on day 5 or in vivo fac tors al low ing per sis tence of sen si tive or gan isms in an ti bi otic pro tected sites. Cau tion is ad vised in the use of cepha lo sporins alone for se ri ous En tero bac ter spe cies in fec tions. Re peat cul ture and sen si tiv ity should be done in se vere in fec tions that are slow to re spond to cepha lo sporin ther apy.

B AC TE RIA EX PRESS BUSH CLASS 1 CHRO MO SOME MEDIated beta-lactamases in two ways: low 'b asal' level pro duction (eg, Escheri chia coli, Sal mo nella spe cies and Shigella spe cies) and 'i nduc ible' ex pres sion (1).In ducible ex pres sion oc curs tran siently in the pres ence of an in ducer an ti bi otic (eg, ce fox itin, imipe nem) in a pre dict able group of bac te ria, includ ing Pseu do mo nas ae rugi nosa, En tero bac ter spe cies, Citro bac ter spe cies, indole-positive Pro teus spe cies, Providen cia spe cies, Mor gan ella mor ganii and Ser ra tia spe cies (1,2).These bac te ria with in duci ble ex pres sion also have a sig nificant number (10 -6 to 10 -7 wild-type cells) of mu tant 'stable dere pressed' cells that pro duce con sti tu tively large amounts of beta-lactamase (1,2).These re sis tant vari ants can be quickly se lected as the pre domi nant bac te ria by la bile weak in ducer an ti bi ot ics both in vi tro (3,4) and in vivo (1,2,(5)(6)(7)(8)(9)(10)(11)(12).The emer gence of re sis tance by this mecha nism dur ing ther apy has been well de scribed and has re sulted in a high rate of treat ment fail ures (1).
We de scribe a case of neo na tal men in gi tis due to En terobac ter cloa cae that de vel oped re sis tance to beta-lactam an tibi ot ics dur ing ther apy and again be came sus cep ti ble while the pa tient was still on ther apy.This is the first case of which we are aware of tran sient se lec tion of a dere pressed mu tant en tero bac ter de scribed in the lit era ture.

CASE PRES EN TA TION
This eight-day-old Cau ca sian fe male was seen by the infec tious dis ease con sult ant for ir ri ta bil ity and fe ver of one day's du ra tion.She was de liv ered by cae sar ean sec tion (for fail ure to prog ress) with birth weight 3800 g to a well mother after a nor mal full-term preg nancy; mem branes were rup tured for 5 h with out evi dence of cho rioam nioni tis, and Ap gars were 9 and 9 at 1 and 5 mins, re spec tively.At birth she was found to have a sac ral mye lo men in go cele and hy dro cepha lus; sur gical re pair and ven tricu loperi to neal shunt in ser tion were done on day 2 of life.She was trans ferred out of the in ten sive care unit to the neo na tal ward on day 4 of life.
On ex ami na tion the neo nate was very ir ri ta ble.Vi tal signs showed a heart rate of 160/min, res pi ra tory rate 45/min, blood pres sure 84/42 mmHg and tem pera ture 38.6°C tym panic.Head cir cum fer ence was 38.7 cm (greater than 90th per centile) and weight 3740 g.The an te rior and pos te rior fon tan elles were large and full, and the skin over the en tire ven tricu loperito neal shunt res er voir and tub ing was ery the ma tous, tender and swol len.She was he mo dy nami cally sta ble, and the spinal wound was heal ing well.The rest of the ex ami na tion was non con tribu tory.Labo ra tory in ves ti ga tions re vealed a white blood cell count of 7.9x10 9 /L with 14% poly mor pho nu clear leu ko cytes, 52% band forms and 27% lym pho cytes, he moglo bin 105 g/L and plate let count 444x10 9 /L.Elec tro lytes, urea, cre ati nine, glu cose, liver en zymes and uri naly sis were all nor mal.Af ter blood, urine and cere bro spi nal fluid (CSF) cultures were drawn she was started on in tra ve nous ce fo taxime (150 mg/kg/day di vided every 8 h) and van co my cin (45 mg/kg/day di vided every 12 h), and op er ated for re moval of the ven tricu loperi to neal shunt and in ser tion of an ex ter nal ized ven tricu lar drain.CSF from the shunt res er voir showed a white blood cell count of 45,900/mm 3 , red blood cell count 50,000/mm 3 , glu cose less than 0.62 mmol/L and pro tein 1400 mg/L with Gram-negative ba cilli on Gram stain.Cul ture of CSF and blood (both sets) grew E cloa cae, urine was ster ile, abdomi nal ul tra sound was nor mal, and com puted to mo gra phy (CT) scan of the head showed evi dence of hy dro cepha lus and ven tricu li tis.
The fol low ing day the pa tient looked well, was afeb rile, and the skin along the pre vi ous shunt tub ing ap peared normal and non tender.Ce fo taxime was con tin ued and gen tami cin was added (7.5 mg/kg/day di vided every 8 h).Van co my cin was dis con tin ued.Daily CSF cul tures were done (twice on day 7), which be came ster ile on day 10 of ther apy.Blood cul ture was ster ile when first re peated on day 5. CT scan of the head with con trast on day 6 and day 10 of ther apy showed no change.In view of the re sis tant iso late (see be low), ther apy was changed to in tra ve nous gen ta mi cin and trimethoprim-sulfamethoxazole (TMP-SMX) (20 mg/kg/ day di vided every 6 h) on day 7 and con tin ued un til day 40.In traven tricu lar gen ta mi cin (1 mg every 24 h) was given from day 12 to 16 (CSF gen ta mi cin lev els 3 h post-dose were 19.2 mg/L and 12 h post-dose were 14.0 mg/L).The CSF bac te ri cidal ti tre on day 25 was 1:64.
The pa ti ent's re cov ery was un event ful apart from gen er alized sei zures on day 6.In ret ro spect, it was found that on day 5 the in tra ve nous line was out and ther apy held for 13 h 20 mins un til one dose of in tra mus cu lar cef tri ax one (50 mg/kg) was given.Ce fo taxime was re started in tra ve nously 18 h af ter the last dose.

MI CRO BI OL OGY
The sus cep ti bili ties of the E cloa cae iso lates in CSF are shown in Ta ble 1; the day 3, 4 and 5 iso lates were re sis tant to sec ond-and third-generation ce pahlo sporins and ti car cil lin, whereas all other initial and later isolates were susceptible.Each day at least four colonies were 'picked' from the culture plate to use for sensitivity testing; however, from day 6 to 9 there were very few colonies on the culture plate to 'pick' (day 7: one colony; day 8: one colony; day 9: three colonies) for sensitivity testing.

An ti bi otic
The results of the inducible beta-lactamase disk approximation test are shown in Figure 1.As described by others (8), the ceftriaxone and imipenem disks were placed on the Meuler-Hinton agar such that they were separated by exactly one half the sum of the zone size around each disk when used alone.The day 3, 4 and 5 isolates had constitutive resistance to ceftriaxone, whereas all other isolates (including the initial blood cultures) had only inducible (with imipenem and cefoxitin) resistance.The nitrocefin assay was positive on all isolates.Clavulanic acid did not change the sensitivity to ampicillin (Table 1) and thus did not inhibit beta-lactamase.Disk-diffusion sensitivity testing showed sensitivity of all isolates to imipenem (diameter 25 mm) and resistance of all isolates to cefoxitin (diameter 6 mm).Pulsed field gel electrophoresis: Genomic DNA in agar plugs from one blood isolate and the nine CSF isolates from day 2 to 9 were prepared essentially as described by Haertl and Bandlow (13).Restriction endonuclease digestion of genomic DNA was done with Xba I (New England Biolabs, Massachusetts) using a 4 mm slice of the agarose plug incubated with 20 units of enzyme in a total volume of 225 µL at 25⊃C overnight.Digested DNA was electrophoresed in a 1% agarose gel (Fastlane agarose, FMC Bioproducts, Maine) in 0.5 x TBE buffer (0.05 M Tris, 0.05 M boric acid, 1 mM EDTA) with the use of the contour-clamped homogeneous electric field apparatus (CHEF Mapper, BioRad, California).Lambda concatemers (New England Biolabs) were run as size controls.Electrophoresis was carried out for 27 h using ramped pulse-time beginning with 3 s and ending with 40 s at an applied voltage of 6 V/cm.The gels were stained with ethidium bromide and destained with distilled water for up to 12 h and photographed under ultraviolet irradiation.Identical banding patterns were found for all the isolates, with 17 bands ranging from approximately 50 to 700 kb size (Figure 2).

DISCUSSION
Selection of Gram-negative rods with constitutive Bush class 1 beta-lactamase production has resulted in emergence of resistance on therapy (with labile weak inducer antibiotics) in 14 to 56% of cases, with treatment failure or relapse in 25 to 75% of these (1).Emergence of resistance is thus recognized as a significant problem.The potentially causative bacteria accounted for 22% of all nosocomial isolates and 11% of all nosocomial bloodstream isolates in the United States in 1986-1989 (14).There have been reports of nosocomial outbreaks of resistance in a cystic fibrosis centre (15) and in neonatal intensive care units (5,11) shortly after introduction of broad spectrum cephalosporins.There are also increasing reports correlating use of broad spectrum cephalosporins to decreased susceptibility of P aeruginosa, Enterobacter species and Citrobacter species to ceftazidime, cefotaxime and piperacillin in the United States (16), in individual hosptals (17) and in individual patients (18).It has also been reported that cefotaxime use in a neonatal intensive care unit resulted in increased colonization with enterobacter strains and decreased susceptibility of these strains to cephalosporins (19).Re-emergence of sensitive strains of enterobacter has been shown hospital-wide with changes in antibiotic use (16,17).
Emergence of resistance while on therapy in neonates with E cloacae meningitis and ventriculitis has been recognized in the past (8,11).The present case adds to the number of documented cases of emergence of resistance while on ther apy with ce fo taxime in neo nates with se ri ous in fec tions due to E cloa cae.It clearly dem on strates the ra pid ity with which the re sis tant phenotype can be se lected as the pre domi nant or gan ism in vivo.This case is un usual in that the re sis tant phe no type was pre domi nant for only three days.
The iso lates from each day were tested by: the Vitek Auto-Microbic Sys tem (Vitek Sys tems, BioMer ieux, Mis souri); the Scep tor Mi cro di lu tion method (Becton-Dickinson Di agnos tic In stru ment Sys tems, Mary land); and the agar di lu tion method for cef tazidime, ce fo taxime and gen ta mi cin (Da lynn Labo ra tory Prod ucts).These were re peated for all the CSF isolates.Fail ure to rec og nize re sis tance in the ini tial or later isolates is thus a very un likely ex pla na tion for the tran sient na ture of the re sis tant phe no type (20).It is also un likely that from day 3 to 5 iso la tion of the re sis tant phe no type was a 'chance' in vi tro find ing.A chance oc cur rence of a re sis tant mu tant is ex pected in about 1x10 -6 colo nies (1,2); thus, with four colo nies 'picked' for sen si tiv ity test ing each day, the chance of de tect ing this for three days is: (4 colo nies) x (1x10 -6 ) x (3 days) = 9x10 -6 .Simi larly, since from day 6 to 9 (for five sepa rate cul tures) vir tu ally all colo nies grow ing were 'picked' for sen si tiv ity test ing, it is rea son able to as sume that the sen si tive phe no type was truly pre domi nant again.
There are very strong data to show that the mecha nism of re sis tance in these iso lates was Bush class 1 betalactamase.The ni tro ce fin test showed that beta-lactamase was pro duced.Cla vu lanic acid did not in hibit the en zyme (Table 1), ce fox itin was a strong in ducer of and sus cep ti ble to the en zyme, and imi pe nem was a strong in ducer of and re sis tant to the en zyme (Fig ure 1).Thus, se lec tion by ce fo taxime of a sta ble dere pressed mu tant on day 3 is the most likely ex plana tion for the emer gence of re sis tance.
An other po ten tial ex pla na tion for the re sis tant pheno type is that there were two dif fer ent strains of E cloa cae with dif fer ent sen si tivi ties.Al though iden ti cal mor pho types and bio types (on Vitek iden ti fi ca tion) for all iso lates are not ade quate to prove strain iden tity, pulsed field gel elec tro pho re sis (PFGE) (Fig ure 2) does ar gue very strongly that the in fec tion was caused by only one strain of E cloa cae.PFGE us ing Xba I to digest ge nomic DNA has been used suc cess fully to dis crimi nate clearly ge netic re lat ed ness of E cloa cae strains (13).Be cause of the Xba I low G-C rare rec og ni tion site TCTAGA, a PFGE gel with up to 20 bands was ex pected (21), pro vid ing a dis criminat ing easy to read gel.This was in deed the case, and the iden ti cal band ing pat tern on PFGE for all the E cloa cae iso lates very strongly sug gests a sin gle clone (13).There fore, the reemergence of the sen si tive phe no type on day 6 must be explained.
One po ten tial ex pla na tion is that the lack of a se lec tive advan tage for the re sis tant mu tant when one dose of ce fo taxime was missed for 18 h on day 5 al lowed re-emergence of the sen si tive phe no type.How ever, the cef tri ax one dose given at 13 h 20 mins makes this ex pla na tion less likely; cef tri ax one (like ce fo taxime) is quite la bile to beta-lactamase and of fers a se lec tive ad van tage to re sis tant mu tants.Per haps a more likely ex pla na tion is that in vivo fac tors al lowed some antibiotic sen si tive or gan isms, es tab lished with the ini tial in fec -tion, to per sist in a rela tively an ti bi otic pro tected site and be cul tured from day 6 to 9.An ti bi otic pro tected sites such as in ven tricu li tis or shunt tub ing would of fer lit tle ad van tage for resis tant mu tants, and would be ex pected to take longer to eradi cate.Thus, the re sis tant mu tant may have been eradicated with gen ta mi cin by day 5, and then the com bi na tion of a missed ce fo taxime dose on day 5 and in vivo pro tected sites con trib uted to the iso la tion of the sen si tive phe no type from day 6 to 9. Fi nally, the change in ther apy to TMP-SMX and genta mi cin on day 7 may have con trib uted to the lack of reemergence of a re sis tant phe no type a sec ond time.
The pres ent case was not an ex am ple of treat ment fail ure.In simi lar pa tients with out re sis tant strains CSF has also been culture-positive for 7.2±5.0days (22).Al though ce fo taxime for Gram-negative men in gi tis has not been evalu ated in a prospec tive fash ion, avail able data sug gest its safety and ef ficacy to be very good (22)(23)(24).How ever, cau tion must be ex er cised in the use of third-generation cepha lo sporins to treat se ri ous in fec tions due to or gan isms such as E cloa cae where emer gence of re sis tance has been a prob lem; this is true both in in di vid ual pa tients and hospital-wide.In pa tients slow to re spond, or in treat ment fail ure, emer gence of re sistance must be looked for with re peat cul ture and sen si tiv ity test ing.Con comi tant use of gen ta mi cin has not been shown to re duce emer gence of re sis tance (1) -in this case re sistance to cepha lo sporins de vel oped while on com bined gen tami cin and ce fo taxime ther apy.

CON CLU SIONS
We re port a case of neo na tal ven tricu li tis due to E cloacae that had rapid but tran sient emer gence of a sta ble derepressed re sis tant mu tant on ther apy as a re sult of chang ing an ti bi otic se lec tion pres sures and in vivo fac tors.This is an ex cel lent in vivo ex am ple of the mecha nism of emerg ing resis tance in Gram-negative rods that is be com ing an in creasing prob lem (25).We ad vise cau tion in the use of third-generation cepha lo sporins alone for these in fec tions.Re peat cul ture and sen si tiv ity test ing should be done in severe in fec tions with these bac te ria that are slow to re spond to cephalo sporin ther apy.

Figure 2 )Figure 1 )
Figure 2) Xba I digested Enterobacter cloacae genomic DNA after pulsed field gel electrophoresis and ethidium bromide staining.Lane 1: blood isolate day 1; lanes 2 to 10: cerebrospinal fluid isolates from day 2 to 9. Identical banding is shown for all isolates, indicating a single clone