Reduction in clinical response to empiric antimicrobial therapy of febrile granulocytopenic patients receiving TM P / SMX infection prophylaxis

In the course of a multicentre clinical trial evaluating two antibacterial regimens for the empiric treatment of suspected infection in febrile neutropenic cancer patients, a suboptimal response was noted among recipients of antibacterial prophylaxis with trimethoprim/sulphamethoxazole (TMP/SMX). Multivariate analysis identified TMP/SMX prophylaxis as a predictor of poor outcome independent of other variables such as classification of infection, marrow recovery, neutrophil count at first fever, indwelling central venous catheter use, and underlying disease. This effect appeared to be restricted to recipients of tobramycin plus ticarcillin (TT). TMP/SMX suppresses potentially pathogenic aerobic Gram-negative bacilli and allows colonization and subsequent infection by Gram-positive microorganisms against which TT-like regimens have limited activity. Recognition of this phenomenon may permit a more appropriate selection of antibacterial agents for the therapy of suspected infection in the neutropenic patient.

T RIMETHOPRIM /SULPHAMETHOXAZOLE (TMP/SMXJ HAS been used in many centres for infection prophylaxis in granulocytopenic cancer patients.Prophylactic success has been reported by some investigators but not by oU1ers (1).TMP/SMX may permit colonization by resistant organisms such as fungi (2).aerobic Gramnegative bacilli (3) and Gram-positive cocci such as Staphylococcus epidermidis (4.5).
During the course of a multicentre trial comparing U1e efficacy and toxicity of t\vo empiric systemic antibiotic regimens for suspected infection in febrile granulocytopenic patients.it was noted U1at patients who received antimicrobial prophylaxis wilh TMP/SMX had a poorer outcome compared wiU1 those who did not (6).This observation has not been previously reported.Multivariate analytical techn iques were used lo separate the effect ofTMP/SMX prophylaxis from other prognostic variables as precliclors of outcome.

PATIENTS AND METHODS
This was a randomized multicentre tri a l conducted under the direction of the Clinical Trials Group of U1e National Cancer Institute of Canada.The protocol and clinical results are reported elsewhere (6).Briefly, 195 granulocytopenic patients (less than 1.0x10 9 granulocytes/L) with their first episode of fever (defined as an oral temperature greater than 38°C noted on three occasions over a 12 h period) were randomized lo receive intravenous ticarcill in (3 g every 4 h) p lus either intravenous moxalactam (2 g every 6 h) or intravenous tobramycin (1.25 mg/kg every 6 h).There was no s ignificant clifference between Lhe study groups (Table 1).TMP/SMX was administered for infection prophylaxis in a nonrandom ized manner according lo local institutional policy.in doses of 160 mg TMP and 800 mg SMX orally every 12 h .Prophylaxis was discontinued when the patient became febrile and entered the trial.Patients were classified according to whether they had received more than five days of prophylaxis (Lhe minimum period required to suppress aerobic Gramnegative fecal micro flora) (2. 7).Analysis: The outcomes of the infectious episodes were categorized as 'complete response•.•temporary improvement' and •failure•.•complete response• was defined as complete clisappearance of all symptoms and signs of infection persisting for four days after cessation of antibiotics: •temporary improvemenl' as a fall in temperature of 1. 7°C or to nom1al accompanied by improve-ment in any clinical signs: and 'failu re• as the absence of improvement or improvement only after addition of alternative antibiotics.For the purpose of Lhis analysis t\vo outcomes were used: response .wh ich included complete response and those temporary improvements which were felt to represent.a beneficial effect of protocol treatment; failure, which included all other cases.
The statistical significance of relation ships between individual prognostic factors and response was assessed using the contingency x 2 test.Th e s im ul taneous effect on outcome of multiple prognostic factors was evaluated with a mu ltiple logistic model.This model identifies the contribu lion of single prognostic variables lo outcome taking into account.other facto rs in the model.Finally, a log-linear model was used to assess wheU1er the effect ofTMP /SMX on outcome vari ed with treatment assignment.U1at is .whether an interaction was present.All analyses were performed using programs available in BMDP (8).Associations were considered significant at P<0 .05.

RESULTS
A total of 244 febrile granulocytopen ic patients experiencin g their first episodes of fever were entered into the study.Of these.195 were evaluable for response ; 106 received moxalactam plus ticarcillin (MT); and 89 received tobran1ycin plus ticarcillin (TT).The reasons for the in evaluability of the rem ru ning 49 patients are detailed elsewhere (6).Forty-nvo MT recipien ts (40%) and 35 TT recipients (39%) received more tha n five days ofTMP I SMX prophylaxis.Patients receiving TMP I SMX were more likely to have acute le ul<emia as the underlying illness.more likely to have had a Hickma n central venous catheter, and more likely to have profound granulocytopenia than patien ts who h ad received no prophylaxis.These relationships were not unexpected.as prophylactic TMPISMX and Hickman catheters are more likely to be used in patients with acute leukemia.a group with more profound and longer duration granulocytopenia .
On multivariate analysis , diffe rences in underlying disease did not appear to account for adverse outcomes associated with TMPI SMX prophylaxis (Table 2).The classification of infection, pattern of granulocyte recovery.presence of an indwelling central venou s catheter.and use ofTMPISMX for more than five days correlated independently with response.
To exan1ine the effect of TMPISMX on response rates , the use of TMPISMX was evaluated by disease classifica tion and allocation (Ta b le 3).TMPISMX use had no influence on response rates in MT recipients .A negative influence on outcome was observed only in TT recipients.The effect was most evident in TMPISMX recipients who h ad acute leukemia.as there was a lower proportion of nonleukemics in the c5roup receiving more than five days ofTMPISMX.
Because TMPISMX pro phylaxis m ay influence the etiology of infection.outcome was evalu ated with respect to TMPISMX use.allocation and infectin g pathogen.Based on all the participating centres.a trend towards more Gram-positive infections was demonstrated among patients receiving m ore than five days of TMPISMX (12 of 19.63%) compared v.riU1 those who did not (18 of 37. 49%).However.data from one of the centres where TMP I SMX is used extensively (Winnipeg.Ma n itoba) show that TMPISMX recipients h ad significan tly m ore Gram-positive infection s th an those who d id not receive T MPISMX prophylaxis (nin e of 10 and five of 15. respectively.x 2 5.689.P~0 .0 2 ) .Of pa tients with microbiologically documen ted infection there were eight frulures am on g TMPI SMX recipients allocated to receive TT. Two of these wer e associated wi U1 slrains of Escherichia coli resistant to boU1 an ti -

DISCUSSION
Multivariate analyUcal techniques have proved useful for id entifying important variables for predic ting outcome of empiric th e ra py in febril e g ranu locytopenic cancer patients.Other inves tigators have s h own th e initial granulocyte co un t. gra nulocyte recovery patte rn.s ite of infection and bacteriologic con firm atio n of infection to be important prognostic va ri ab les (9.10).In U1e present multi centre study simil ar techn iqu es we re u sed to id entity TMP /SMX prophylaxis as a predictor of poor outcome ind ependent of other factors known to inOuence the outcome of e mpiric antibiotic therapy in this patient population.Althou gh the effec t o f TMP/SMX prophylaxis on outcome was in depe nde nt of underlying disease.use of indwe lling centra l venous cathete rs.granulocyte recovery p a tte rn .infectio n class ifi catio n. a nd differences be tween th e participating institutions.it was depend ent upon regimen a ll ocation .The poorer response rate was observed on ly in TT recipients who had received TMP/SMX prophylaxi .
The relationsh ip between TMP/SMX prophylaxis a nd poor outco me in TT recipi e nts was exam ined.First.TMP/SMX can a llow co lon iza tion and s ubsequ ent infection by microorganisms that m ay be cross-res is ta nt to TMP/SMX a nd to then' regim e n (3 .11).Th e present study.however.identified on ly t•wo E co li bacte remia s and one Strep bovis bacterem ia resistant ton' a mon g the TT fa ilures.The remaining five fa ilures were due to mi c roorga nism s susceptible to on e or both of th e agents in the regimen.su~gesli n g th a t other factors in ad dition to drug susceptibili ty were importa nt in determining outcome .
Second.TMP/SMX s uppresses aerobi c Gram -negative bacilli in the gastroin testinal tract.and a llows co lonization by Gram-pos itive cocci such as Staph epidermidis an d viridans streptococci (2.4.12).This fact.togeth er with th e increas ed in tegumental damage du e to cytotoxic ch emoth e ra peuti c agents used fo r cancer treatment.may acco unt for the obse rved in c rease in proportion of Gram-positive infections a mong TMP/ SMX recipie nts (4.5.12 .1 3 ).In contrast.other reports link the u e of indwelling central venous lines to an increased incidence of Gra m -positive infec tion s in immunocompromised pa ti e nts (1.l 4).Th e prese nt study s uggests that indwe lling centra l venous cath eters a nd 238 TMP/SMX are independent factors in c reasing U1e risk of Gram -positive infection.
Antipseudomonal penicillin plus aminoglycos ide or broad-spectrum ceph alosporin plus a minoglycos ide combinations a ppear to h ave limited activity aga inst Gram-pos itive infections in n eu tropen ic patients (15-17).Tobramycin is not particularly active in vitro against streptococci (18).a nd the comb ination of ticarcillin with an aminoglycoside h as not been successful in a series of single Gram-positive bacteremias.most of which were due to Staph epidem1idis (15).Superinfection \vith IT-resistant Staph epidennidis has been reported in neutropen ic patients receiving TMP/SMX and TT (19).In fact.the occurren ce of Gram-pos itive brea kthrough bacteremia appears to be rela tively common among recipients of anlipseudomonal peni cillin plu s aminoglycoside regimens (20-22).The addition to U1e regimen of a s pecifi c Gram-positive acrent such as vancomycin appears to reduce significantly Lhe morbidity associated with U1e febrile episode (20.21.23.24).
Th e interaction between TMP/SMX.TT a nd outco me in U1e present study is coupled to U1e predomin a n ce of Gram-positive infection observed in the trial a nd the suboptimal responses in IT recipi ents (6) .The a uthor feel that TMP/SMX in the present study.as in oU1ers (2.12.1 3).selected for Gram-positive pathogens aga in st which TT a lone m ay h ave limited activity.Reports from other centres suggest that the suboptimal activity observed for IT against Gran1-pos itive paUwgens is a phenomenon a ppli ca ble to other antipseud omonal penicillin or broad-spectrum cephalosporin plus a minoglycos ide combinations ( 15-17 .20.22).Accordingly.U1e a uU1ors suggest that febrile n eutropenic patients who receive empiri c th erapy \vith extended spectrum betalactams plus a minoglycoside regim ens following TMP/ SMX prophyl a.'Cis \¥ill likely require regime n modifica tion with a Gram -positive agent s u ch as vancomycin.
lrimelh opi-Lm -sulfamelhoxazole for infec ti on prophylaxis in neutr openic cance r patients: Relation ship of effi cacy lo anti microbia l spectrum and timing of administra tion.
H aU1orn JW.Hi em en z J. el al.A r andomized tri a l co mpa ring ceflazidime alone wi th co mbination antibiotic therapy in can ce r patients wiU1 fever and neui.J•openia.N Eng! J M ed 1986:3 15:552-8.23.Rubin M .Hathorn JW.M arshall D. G ress J. Stein berg SM.Pizzo PA.G ra m -positive infec tion s and t h e u se of vancomycin in 550 episodes of fever an d n eutropenia .Ann In tern M ed 1988 :108: 3 0 -5.24.Eu r opean Organization for Resea rch and Treatment of Ca n ce r .Internationa l Antim ic rob ia l Th er apy Cooper ative Group a nd th e Nation al Cance r Institute of Canada Cli nica l Trials Group.Vcu1comycin added to empirica l combination antibiotic th erapy for fever in g ra nulocy topenic can ce r patients.J Infect Dis 199 1:163 :95 1-8.

TAB l E 1 Patient characteristics in a randomized multicentre trial comparing the efficacy and toxicity of two empiric sys- temic antibiotic regimens for suspected infection in febrile granulocytopenic patients
TMP / SMX Trimethoprim/ sulphamethoxazole CAN j INFECT DIS VOL 3 No 5 SEPTEMBER/OCTOBER 1992

TABlE 3
Prophylaxis reduces clinical response to therapy UnivariateResponse rates according to allocation, the use of prophylaxis, and underlying illness in a randomized multicentre trial comparing the efficacy and toxicity of two empiric systemic antibiotic regimens for suspected infection in febrile granulocytopenic patients No 5 SEPTEMBER/OCTOBER 1992 biotic .a nd two we re associated \vith Gram-negative bacilli susceptible to boU1 antib iotics.The four remajnin failures ha d Gram-positive infections associated \vith resistance to both antibi otics in one cas e (Streptococcus bovis).re istance to tobramycin in two cases (a viridans group streptococcu s and Staph epidermidis).