Evaluation of infectious etiology and prognostic riskfactors of febrile episodes in neutropenic cancerpatients

Serap Gencera,*, Taflan Salepcib, Serdar Ozera

aDepartment of Infectious Diseases and Clinical Microbiology, Kartal Dr Lutfi Kirdar Training and ResearchHospital, Istanbul, TurkeybDivision of Medical Oncology, Department of Internal Medicine, Kartal Dr Lutfi Kirdar Training andResearch Hospital, Istanbul, Turkey

KEYWORDSFebrile neutropenia;

Cancer; Infection;

Bacteria; Risk factors

Summary Objectives. Febrile neutropenic cancer patients are at risk for developmentof serious infections, morbidity and mortality. The aim of this study was to determinethe type and frequency of infections and to evaluate some prognostic risk factors.

Methods. 220 episodes of neutropenic fever in 177 cancer patients have beenreviewed.

Results. Infections could be documented microbiologically in 38 (17.3%) episodesand suspected clinically in 29 (13.2%). The most common focus of infection was thelower respiratory tract (11.4%) followed by the urinary tract (6.4%). The mostfrequently isolated pathogen was Escherichia coli (31%) followed by Klebsiellapneumoniae (18%), Pseudomonas aeruginosa (13%) and Streptococcus pneumoniae(13%). The median durations of neutropenia and fever were 4 and 3 days, respectively.Mortality was seen in 25 patients (11.4%). Its rate was higher in documented infectionsexcept for non-bacteremic microbiologic infections in which no death was seen.Hypotension and shock were the most significant determinants of poor prognosis.

Conclusions. The management of these special patients should be given adequateattention and be considered important since the success of therapy depends onrevealing of etiologic agents.Q 2003 The British Infection Society. Published by Elsevier Science Ltd. All rightsreserved.

Introduction

While the success of treatment in cancer patients isincreasing with the use of high dose chemotherapy inrecent years, immunosuppression, especially pro-longed and severe neutropenia, leads in the patientsto infections with serious atypical progress. There is arisk of infection in over 60% of neutropenic patients

who become febrile, but, according to recent data,infections have been documented in only 30–50% ofthose patients.1 –5 Infections originate from thegastrointestinal tract, respiratory tract and skin,respectively.5,6 It is reported that gram-positivebacteria have become more common isolates asetiologic agents and the incidence of fungal infectionhas been increasing.1,5,7,8 However, there are recentreports suggesting that gram-negative infections arealso increasing.9–12

In this study, we investigate the neutropenicfebrile episodes of cancer patients treated with

0163-4453/03/$ - see front matter Q 2003 The British Infection Society. Published by Elsevier Science Ltd. All rights reserved.doi:10.1016/S0163-4453(03)00044-6

Journal of Infection (2003) 47, 65–72

www.elsevierhealth.com/journals/jinf

*Corresponding author. Selimiye, Eczane sok. Hamle ap. No:15/17 81170, Uskudar-Istanbul, Turkey. Tel.: þ90-542-2451702.

E-mail address: segencer@turk.net

chemotherapy in our hospital in order to determineinfections causing fever and to evaluate possiblerisk factors associated with these infections andmortality.

Patients and methods

Of adult cancer patients followed and treated in theMedical Oncology Department of Kartal Trainingand Research Hospital, only those who have fever (asingle oral temperature of $38.3 8C or $38 8C for$1 h) and whose absolute neutrophil count is#500/mm3 are hospitalized in the Infectious Dis-eases Department and included in the neutropenicfever protocol of our hospital. According to ourprotocol, fever related to transfusion of blood orblood products is excluded. In this retrospectivestudy, we investigated the patients includedbetween May 1997 and December 2000.

All patients were evaluated by physical examin-ation, sampling for at least 2 blood cultures, urineculture and, if appropriate, cultures from othersuspected body sites, chest radiography, a com-plete blood count, erythrocyte sedimentation rate,C-reactive protein, a complete urine analysis andstandard blood chemistry (including urea, creati-nine, electrolytes, ALT, AST, alkaline phospha-tase). Blood cultures were performed in anautomated system (BacTec). Additional blood cul-tures were done if the patient’s fever continued. Allorganisms isolated from clinical specimens wereidentified by standard criteria in our clinicalmicrobiology laboratory. Antimicrobial suscepti-bility testing of bacterial isolates was performedby the disk diffusion method according to the NCCLSrecommendations. Soon after obtaining cultures, abeta-lactam antibiotic (ceftazidime 2 g three timesa day IV or cefepime 2 g twice a day IV orpiperacillin/tazobactam 4.5 g three times a dayIV) was administered initially as empirical treat-ment if the patient has lower risk characteristics.The choice of initial empirical antibiotics was leftto the discretion of the admitting physician. Whenthe duration of fever was more than two days andinfection was not documented, amikacin was addedto the treatment regimen. If they had initiallyhigher risk characteristics, a combination of ami-kacin (15 mg/kg once a day IV) with one of theabove antibiotics was administered. In patientswith microbiologically documented infection,treatment was modified according to the sus-ceptibility of the pathogen. In patients with feverpersisting for 5 –7 days or with evidence offungal infection such as mucosal candidiasis or

esophagitis, fluconazole (100 mg every 12 h) wasadded to the regimen. All antibiotics were discon-tinued after the patient had been afebrile for 5days, at the time of neutrophil recovery (granulo-cyte count .500/mm3) or after disappearance ofsigns of infection.

Those patients who had a hematological malig-nancy (acute leukemia, lymphoma, etc.), who hadsevere mucositis, diarrhea, pneumonia, cardiopul-monary insufficiency, hypotension or any prominentorgan dysfunction, whose neutropenia duration was$10 days and most of the time neutrophil countwas #100/mm3, whose general appearance waspoor, were defined as high-risk group patients. Onthe other hand, those patients whose neutropeniaduration was ,10 days, neutrophil count.100/mm3, platelet count .75 000/mm3 andwhose general appearance was good were definedas low-risk group patients.13 –15

Neutropenic febrile episodes were classified intoone of three groups: (1) microbiologically docu-mented infection (MI), (a) with bacteremia (definedas a positive result of cultures of blood samples),(b) without bacteremia (defined as a positive resultof cultures of samples other than blood which couldexplain fever of the patient); (2) clinically sus-pected infection without microbiological evidence(CI) (defined as presence of physical or radiologicalfindings compatible with an underlying infection,but without any microbiological proof, includingpneumonia, skin and soft tissue infection, sepsissyndrome) and (3) fever of unknown origin (FUO)(without a clinically suspected site of infection or amicrobiologically documented cause of infection).

We assessed potential risk factors for documen-ted infections by comparing the characteristics ofthe patients who had documented infections withthose who had none. The following potential riskfactors were assessed: age, sex, duration ofneutropenia, duration of fever, previous febrileneutropenic episode, previous surgical intervention(up to previous 8 weeks), previous hospitalization,initial high fever ($39 8C), hypotension (systolicblood pressure ,90 mmHg), low PNL count(#100/mm3), low platelet count (#75 000/mm3),low hemoglobin value (#8 g/dl), presence ofmucositis, esophagitis, metastasis, properties ofthe high-risk group, requirement of blood products,median CRP values at the beginning of the episodeand at the end of the antimicrobial therapy. None ofthe patients had received any prophylaxis preced-ing the episode and none had used central venouscatheter.

The same potential risk factors were also assessedfor mortality by comparing the characteristics ofpatients with mortality and those with no mortality.

S. Gencer et al.66

Comparisons of the proportion of risk factorsbetween groups were performed using chi-squaretest and Fisher’s exact test when necessary.Comparisons of medians between groups wereperformed using Mann–Whitney U test. Associationof significant risk factors with the occurrence ofinfection or mortality was tested by a multivariatelogistic regression analysis. Mean CRP values at thebeginning of the episodes were compared withthose at the end of treatment by using Wilcoxonrank test. Statistical significance was assigned totwo-sided p values less than 0.05.

Results

This series includes 220 neutropenic febrile epi-sodes in 177 cancer patients. The median age ofpatients was 55 years, ranging from 14 to 93. Of thetotal 177 patients, 98 (55%) were men, while 79(45%) were women. The types of malignancies aresummarized in Table I.

The median time of the neutropenia underobservation was 4 days and neutropenia lasted for10 or more days in only 16 episodes. Severeneutropenia, defined as a neutrophil count#100/mm3, was seen at the beginning of 116(53%) episodes and during 33 (15%) episodes. Severeneutropenia was not seen in only 71 (32%) episodes.Also in 167 (76%) episodes there was at least one ofthe characteristics of the high-risk group. Durationof hospital stay ranged from 1 to 43 days (median 10days). The median duration of fever was 3 days.

Of the total 220 episodes, 67 (30.5%) hadclinically or microbiologically documented infec-tion (Table II). Seventy-four infections were diag-nosed in 67 episodes, seven were due to mixedinfections. Characteristics of these 74 infectionsare shown in Table III. The most common foci of

infection were the lower respiratory tract andurinary tract.

Twenty-five patients (11.4%) had an infection ofthe lower respiratory tract. In 15 of those withcough, dyspnea and typical features of consolida-tion observed on the chest radiography, no bac-terial or fungal pathogen was isolated. In all these15 patients, in whom further investigation could notbe done, pulmonary infection resolved with empiri-cal broad-spectrum antibiotic therapy. Six otherpatients had bacteremia, caused by Streptococcuspneumoniae in four, Enterococcus faecium in oneand Citrobacter spp. in one. In the remaining fourpatients, pathogens were isolated from the sputum.

Urinary infections were documented in 14 (6.4%)patients. Escherichia coli were isolated in sevencases, one of whom was bacteremic.

Wound infections, esophagitis, peritonitis andperianal abscess were other microbiologicallydocumented infections. Sepsis syndrome, cellulitisand gluteal abscess were clinically suspectedinfections other than pneumonia. Any microbiolo-gical infection could not be documented in episodeswith sepsis syndrome. Since it occurs mostly inresponse to an infection, these episodes wereaccepted as clinically suspected infection.

Bacteremia was diagnosed in 17 (7.7%) episodes.Gram-negative bacteria were isolated in 10 epi-sodes. E. coli caused bacteremia in five of them.Gram-positive pathogens causing bacteremia wereisolated in 7 episodes, five of which were caused byS. pneumoniae (Table IV). Six patients with bacter-emia had pulmonary infection, one had urinaryinfection and another one had wound infection(Table III).

Thirty-nine different bacterial pathogens wereisolated from all episodes (Table IV). All bacteriawere isolated from cultures obtained at the time ofadmission. Gram-negative organisms represented74.4% of all isolates. E. coli was the most common

Table I Types of malignancies of patients

Malignancy No. of patients ðn ¼ 177Þ No. of episodes ðn ¼ 220Þ

Lung cancer 72 87Lymphomas 33 46Breast cancer 22 26GIS malignancy 12 16Ovarian cancer 6 7Testicular cancer 3 6Endometrium and cervical cancer 4 4Osteosarcoma 3 4Multiple myeloma 3 3Nasopharyngeal cancer 3 3Acute myeloblastic leukemia 1 3Other solid tumors 15 15

Infections in febrile neutropenia 67

isolate and represented 31% of all isolates. Kleb-siella pneumoniae (18%), Pseudomonas aeruginosa(13%) and S. pneumoniae (13%) followed E. coli asthe most frequent isolates. Four of 29 gram-negative isolates produced extented-spectrumbeta-lactamase and the other two isolates wereresistant to most of beta-lactam antibiotics exceptcarbapenems. There were no carbapenem resist-ance among gram-negative isolates and no glyco-peptide resistance among gram-positive isolates.

All isolates of S. pneumoniae were penicillin-susceptible.

As the initial antimicrobial therapy, ceftazidimewas administered to 92 patients, and combinedwith amikacin in 69 cases; cefepime was adminis-tered to 84 patients, combined with amikacin in 31;piperacillin/tazobactam was administered to 35patients, combined with amikacin in 5 and carba-penems (either imipenem or meropenem) wereadministered to 9 patients. The initial treatmentwas modified in 20 (22%) patients of the ceftazidimegroup, 12 (14%) patients of the cefepime group and5 (14%) of the piperacillin/tazobactam group. In six(16%) of the total 38 patients who had microbiolo-gically documented infection, treatment was modi-fied to a carbapenem with regard to the antibioticsusceptibilities of isolates. In 25 patients treatmentwas modified to a carbapenem because of persist-ent fever and in 14 of them fever resolved on thefollowing day. In six patients, monotherapy with acarbapenem was preferred because of renal dys-function contraindicating use of aminoglycosides.

Forty-seven patients had oral mucositis and/oresophagitis. Candida spp. was isolated in 15 ofthem. Fluconazole was given to 39 patients due togram staining for treating oral mucositis, in 27 atthe beginning of and in 12 during the antibiotictherapy. Twelve of these patients also had esopha-gitis but candida was shown in only seven andisolated in only two of these. Eight patients withoral mucositis were treated with local nistatinsolution only. In 13 cases fluconazole and in twoamphotericin B were added to the empiricalantibiotic therapy according to persistence offever on the 7th day.

The median number of days with fever was 3 in195 episodes in which antimicrobial therapy wascompleted successfully. The median number ofdays with neutropenia was 4. Among risk factorsevaluated hypotension ðp ¼ 0:001Þ; requirement ofblood products ðp ¼ 0:041Þ and CRP values ðp ,

0:001Þ were significantly associated with infection(Table V).

We observed mortality in 25 patients (11.4%), 13of whom were due to infections (8 with sepsissyndrome, 3 with bacteremia, 2 with pneumonia).In 3 of them pathogens were isolated (P. aerugi-nosa, E. faecium, S. pneumoniae). The risk ofmortality was significantly associated with hypo-tension ðp , 0:001Þ (Table VI). The rate of mortalitywas significantly higher among patients with infec-tion (20.9 vs. 7.2%, x2 ¼ 8:690; p ¼ 0:003) althoughno death occurred in the group of microbiologicallydocumented infection without bacteremia. How-ever, multivariate analysis found no significantassociation between infection and mortality. The

Table II The classification of febrile neutropenic episodes

Classification n %

Microbiologically documented infection (MI) 38 17.3With bacteremia 17 7.7Without bacteremia 21 9.6

Clinically suspected infection (CI) 29 13.2FUO 153 69.6

Table III Characteristics of infections

Infections n ¼ 74

Pneumonia 25S. pneumoniae 4(4)K. pneumoniae 2E. faecium 2(1)Enterobacter spp. 1Citrobacter spp. 1(1)Othera 15

Urinary infection 14E. coli 7(1)K. pneumoniae 4P. aeruginosa 1Enterobacter spp. 1Proteus spp. 1

Sepsis syndromea 10Primary bacteremia without a known focus 9E. coli 4P. aeruginosa 2K. pneumoniae 1S. pneumoniae 1a-Hemolytic streptococcus 1

Esophagitis 7Candida spp. (isolated) 2Otherb 5

Wound infection 4P. aeruginosa 2(1)Staphylococcus aureus 2

Cellulitisa 2Peritonitis 1Enterobacter spp.

Perianal abscess 1E. coli

Gluteal abscessa 1

The numbers in parentheses denote number of bacteremicisolates.a Clinically suspected.b Not isolated, but shown with gram stain.

S. Gencer et al.68

rates of mortality were 14, 10 and 3% amongcefepime, ceftazidime and piperacillin/tazobac-tam groups, respectively ðp ¼ 0:486Þ:

The median CRP value in the first 48 h wassignificantly higher in the infection group (TableV). The median CRP value at the end of

antibiotic treatment was higher in the mortalitygroup, but this has not been taken intoconsideration because mortalities were seenearly. When comparing the first CRP valueswith the last values the CRP values at thebeginning of the episode were significantly

Table IV Bacterial pathogens isolated from febrile neutropenic patients during episodes

Blood ðn ¼ 17Þ Urine ðn ¼ 13Þ Sputum ðn ¼ 4Þ Wound ðn ¼ 3Þ Othera ðn ¼ 2Þ Total ðn ¼ 39Þ

Gram-negativesE. coli 5b 6 – – 1 12(31%)K. pneumoniae 1 4 2 – – 7(18%)P. aeruginosa 3c 1 – 1 – 5(13%)Enterobacter spp. – 1 1 – 1 3(8%)Citrobacter spp. 1 – – – – 1(3%)Proteus spp. – 1 – – – 1(3%)

Gram–positivesS. pneumoniae 5 – – – – 5(13%)E. faecium 1 – 1 – – 2(5%)Staphylococcus aureus – – – 2 – 2(5%)a-Hemolytic streptococcus 1 – – – – 1(3%)

The numbers in parentheses denote percent of isolated pathogens.a Peritoneal fluid and perianal abscess, each in one case.b Urinary infection with bacteremia, in one case.c Wound infection with bacteremia, in one case.

Table V Comparison of risk factors between episodes with clinically or microbiologically documented infection and episodes withFUO

MI and CI (n ¼ 67) FUO (n ¼ 153) Univariate pa Multivariate pb

Median age 57 53 0.042c 0.051Female/male 30/37 66/87 0.822Median febrile days 3 3 0.570c

Median neutropenic days 4 4 0.209c

Median days on antibiotics 10 10 0.906c

Previous historyFebrile neutropenic episode 12 (17.9) 31 (20.3) 0.684Surgical intervention 3 (4.5) 8 (5.2) 0.812Hospitalization 14 (20.9) 49 (32) 0.087

At the beginningFever $39 8C 14 (20.9) 34 (22.2) 0.826Hypotension (systolic BP ,90 mmHg) 13 (19.4) 1 (0.7) ,0.001 0.001PMNL ,100/mm3 38 (56.7) 78 (50.9) 0.432Platelets ,75 000/mm3 35 (52.2) 80 (52.3) 0.995Hemoglobin ,8 g/dl 25 (37.3) 58 (37.9) 0.933

Mucositis 14 (20.9) 29 (19.0) 0.739Esophagitis 6 (9.0) 6 (3.9) 0.146Candida isolation 5 (7.5) 10 (6.5) 0.803Properties of high-risk group 55 (82.1) 112 (73.2) 0.148Requirement of blood products 22 (32.8) 29 (19.0) 0.028 0.041Mortality 14 (20.9) 11 (7.2) 0.005 0.895Metastasis 6 (9.0) 31 (20.3) 0.030 0.068CRP (mg/l) (median value)In the first 48 h of episode 85 72 0.001c ,0.001At the end of antimicrobial therapy 42 44 0.986c

The numbers in parentheses denote percent of episodes.a Chi-square test used for all variables except median values.b Logistic regression analysis.c Mann–Whitney U test used for median values.

Infections in febrile neutropenia 69

higher than the values at the end of antimicro-bial therapy ðp , 0:01Þ:

Discussion

The most serious complication in cancer patientsduring chemotherapy is neutropenia and accompa-nying fever, of which infections constitute 50–80%of the causes.3,13 Since infection is the mostimportant cause of mortality, fever should beconsidered as a sign of a life-threatening infectionuntil the contrary is proven. For patients withneutropenia, a specific site of infection is generallylacking because of a diminished inflammatoryresponse. In the last 20 years, the initial evaluationdoes not identify a focus of infection in nearly twothirds of cases.2 –4,13,14 In our study, 30.5% ofpatients had clinically suspected or microbiologi-cally documented infection.

Most series report an incidence of bacteremiabetween 10 and 24%.16 Our incidence of bacteremiawas 7.7%. Eight of 17 patients with bacteremia hada focus of infection but in the other nine patientswith bacteremia, the focus was not found.

We found the lower respiratory tract the mostcommon site of infection (11.4%). It has been

reported that pneumonia complicates 40–60% ofall bone marrow transplantations. However, pneu-monia in the neutropenic population is compara-tively low and is estimated at 0.5–10%.17 –20

The spectrum of organisms responsible forinfectious complications in neutropenic patientshas changed during the last two decades. Gram-positive organisms, generally considered to be lessvirulent, especially the coagulase negative staphy-lococci, have emerged as the leading cause of acutebacterial infections, which represent 55–60% of allepisodes of bacteremia.7,13 However, severalrecent reports suggest that gram-negative organ-isms are increasing in frequency again and are nowrepresenting approximately 50% of all isolates.9 –12

This increase may be partly due to more intensivechemotherapy regimens.

In our country, the incidence of organisms variesfrom center to center. In a multicenter studyconducted between 1995 and 1996, the incidenceof gram-positive bacteremia was 69%.21 However,in another multicenter study conducted between1997 and 1998, the incidence of gram-positivebacteremia has been found equal to that of gram-positive bacteremia.22 In some centers, gram-negative organisms are predominant. In our study,gram-negative organisms are responsible for 74.4%

Table VI Comparison of risk factors between episodes of patients who had died and survived

Died (n ¼ 25) Survived (n ¼ 195) Univariate pa Multivariate pb

Median age 59 55 0.670c

Female/male 8/17 88/107 0.207Microbiological infection 3 (12) 35 (18) 0.458 0.635Clinical infection 11 (44) 18 (9.2) ,0.001 0.458FUO 11 (44) 142 (72.8) 0.003 0.585Previous historyFebrile neutropenic episode 4 (16.0) 39 (20.0) 0.628Surgical intervention 1 (4.0) 10 (5.1) 0.802Hospitalization 4 (16.0) 59 (30.3) 0.118

At the beginningFever $39 8C 3 (12.0) 45 (23.1) 0.180Hypotension (systolic BP ,90 mmHg) 11 (44.0) 3 (1.5) ,0.001 ,0.001PMNL ,100/mm3 11 (44.0) 105 (53.8) 0.353Platelets ,75 000/mm3 12 (48.0) 103 (52.8) 0.650Hemoglobin ,8 g/dl 11 (44.0) 72 (36.9) 0.495

Mucositis 3 (12.0) 40 (20.6) 0.287Esophagitis 1 (4.0) 11 (5.6) 0.723Candida isolation 1 (4.0) 14 (7.2) 0.526Properties of high-risk group 22 (88.0) 145 (74.4) 0.109Requirement of blood products 8 (32.0) 43 (22.1) 0.283Metastasis 4 (16.0) 33 (16.9) 0.907CRP (mg/l) (median value)In the first 48 h of episode 77 76 0.595c

At the end of antimicrobial therapy 73.5 42.5 0.014c 0.013

The numbers in parentheses denote percent of episodes.a Chi-square test used for all variables except median values.b Logistic regression analysis.c Mann–Whitney U test used for median values.

S. Gencer et al.70

of all isolates and 58.8% of bacteremic isolates. E.coli was the most common isolate among all isolates(31%) and bacteremic isolates (29%). This situationmay be associated with the characteristics of ourpatients, most of whom have been followed outsidethe hospital, have not received any prophylacticantibiotics and have had no long-lasting hospitaliz-ation and use of central venous catheters.

Standard treatment for cancer patients withfever and neutropenia has traditionally beenhospital admission for administration of broad-spectrum intravenous antibiotics which cover boththe most common and most dangerous pathogens.This approach has resulted in an infection-relatedmortality rate of 4–6% in adults.16 An aminoglyco-side in combination with an antipseudomonal beta-lactam has become the general standard at manycenters for the empirical treatment of febrileneutropenia. The choice of an initial empiricalregimen must be made with the knowledge of thesusceptibility patterns of recent isolates at thecenter where the patient is being treated.18

Infectious death occurs in 21% of episodes with afocus of infection, in comparison with 4% forepisodes without a focus.23 The rate of mortalityin our study was also 21% of episodes withdocumented infection and 7% for episodes ofunknown origin. However, no death was seen inmicrobiologically documented infections withoutbacteremia. This may be due to the opportunity tochange the initial antibiotics according to suscep-tibility patterns of pathogens isolated except forthose causing bacteremia. Therefore, isolation ofetiologic agents is very important. Changes in theempirical regimen should be guided by alterationsin the clinical state of a patient or evidence ofantibiotic resistant organisms.

In recent years, the concept of risk assessmentduring the initial phase of a febrile episode has beenintroduced and evaluated and outpatient antibiotictherapy has become an alternative to hospitaliz-ation in low-risk patients with febrile neutrope-nia.8,12,15,16 Patients with neutropenia can bedivided into low and high-risk groups on the basisof the projected duration of neutropenia. FirstBodey et al., then Pizzo et al. defined patients aslow-risk if the neutropenia was resolved within 7days and the underlying malignancy was in remis-sion.24,25 However, with the use of more intensivechemotherapy, periods of neutropenia lasting for 2weeks or longer are frequently seen.12 Talcott et al.analysed the risk profiles of febrile neutropenicpatients, stratifying them into four groups accord-ing to the main risk factors at clinical presen-tation.26,27 Although the severity and duration ofneutropenia is the main determinant in developing

infectious complications, a number of factors areassociated with the development of infection andmortality.28 –30 A recent international study con-ducted by the Study Section on Infections ofMultinational Association of Supportive Care inCancer (MASCC) established and validated a scoringsystem to identify low-risk patients.29 This scoringsystem was suggested by the recent guideline of theInfectious Diseases Society of America (IDSA) Feverand Neutropenia Guidelines Panel.8 According toour data, the presence of hypotension and high CRPvalues are important determinants of infection, andhypotension is also significantly associated withmortality. Higher CRP levels may suggest infection,as shown also in other studies.31,32 Serial determi-nation of CRP values help to determine the successof the antibiotic regimen. Although the status ofunderlying malignancy was not evaluated in thisstudy, we thought that underlying malignancy andlate admission to hospital contributed to our highincidence of mortality and low incidence of micro-biologically documented infection. This circum-stance necessitates more deliberately approach tooutpatient therapy in our patient population.

Risk assessment must be more comprehensive fordecision-making regarding initial antibiotic treat-ment in patients with neutropenia. This approachwill significantly reduce medical complications andcosts. Some parameters, such as CRP, whichdetermine the relative risk of infection and theprognosis, may be useful in deciding on modifi-cation of initial treatment. In conclusion, webelieve that evaluation of risk factors as much ascareful clinical and microbiological approach todetermine infections is necessary for successfultherapy.

References

1. Hathom JW, Lyke K. Empirical treatment of febrile neu-tropenia: evolution of current therapeutic approaches. ClinInfect Dis 1997;24(Suppl 2):S256—S265.

2. Freifeld AG, Walsh TJ, Pizzo PA. Infections in the cancerpatient. In: De Vita Jr., Hellman S, Rosenberg SA, editors.Cancer: Principles and Practice of Oncology, 5th edn.Philadelphia: Lippincott-Raven; 1997. p. 2659—2704.

3. Donowitz GR. Fever in the compromised host. Infect Dis ClinNorth Am 1996;10:129—148.

4. Chanock SJ, Pizzo PA. Fever in the neutropenic host. InfectDis Clin North Am 1996;10:777—795.

5. Pizzo PA. Management of fever in patients with cancer andtreatment-induced neutropenia. N Engl J Med 1993;328:1323—1332.

6. De Pauw BE, Donnelly JP. Infections in the immunocompro-mised host: general principles. In: Mandell GL, Bennett JE,Dolin R, editors. Principles and Practice of Infectious

Infections in febrile neutropenia 71

Diseases, 5th edn. Philadelphia: Churchill Livingstone; 2000.p. 3079—3090.

7. Kibler CC, Prentice HG. Pathogen ‘shift’ in febrile neutro-penia. Curr Opin Infect Dis 1999;12:351—354.

8. Hughes WT, Armstrong D, Bodey GP, et al. Guidelines for theuse of antimicrobial agents in neutropenic patients withcancer. Clin Infect Dis 2002;34:730—751.

9. Akova M. Etiology of bacterial infections in cancer patients inEurope: an ever changing scenario (12th ECCMID, Milan-Italy,21—24 April Abstract no: S344). Clin Microbiol Infect 2002;8(Suppl 1):50.

10. Rolston KVI, Tarrand JJ. Pseudomonas aeruginosa—still afrequent pathogen in patients with cancer: 11-year experi-ence at a comprehensive cancer center. Clin Infect Dis 1999;29:463—464.

11. Viscoli C. Planned progressive antimicrobial therapy inneutropenic patients. Br J Haematol 1998;102:879—888.

12. Aquino VM, Pappo A, Buchanan GR, et al. The changingepidemiology of bacteremia in neutropenic children withcancer. Pediatr Infect Dis J 1995;14:140—143.

13. Pizzo PA. Fever in immunocompromised patients. N Engl JMed 1999;341:893—900.

14. Karthaus M, Carratal J, Jurgens H, Ganser A. New strategiesin the treatment of infectious complications in haematologyand oncology: is there a role for out-patient antibiotictreatment of febrile neutropenia? Chemotherapy 1998;44:424—435.

15. Rolston KVI, Rubenstein EB, Freifeld A. Early empiricantibiotic therapy for febrile neutropenia patients at lowrisk. Infect Dis Clin North Am 1996;10:223—237.

16. Klaassen RJ, Goodman TR, Pham B, Doyle JJ. “Low-risk”prediction rule for pediatric oncology patients presentingwith fever and neutropenia. J Clin Oncol 2000;18:1012—1019.

17. Bergen GA, Shelhamer JH. Pulmonary infiltrates in thecancer patient. Infect Dis Clin North Am 1996;10:297—325.

18. Chanock SJ, Pizzo PA. Infectious complications of patientsundergoing therapy for acute leukemia: current status andfuture prospects. Semin Oncol 1997;24:132—140.

19. Collin BA, Ramphal R. Pneumonia in the compromised hostincluding cancer patients and transplant patients. Infect DisClin North Am 1998;12:781—805.

20. Quadri TL, Brown AE. Infectious complications in thecritically ill patient with cancer. Semin Oncol 2000;27:335—346.

21. Akova M, Akan H, Korten V, et al. Comparison of meropenemwith amikacin plus ceftazidime in the empirical treatment offebrile neutropenia: a prospective multicentre trial in

patients without previous prophylactic antibiotics. Int JAntimicrob Agents 1999;13:15—19.

22. Erman M, Akova M, Akan H, et al. Comparison of cefepimeand ceftazidime in combination with amikacin in theempirical treatment of high-risk patients with febrileneutropenia: a prospective, randomised, multicenterstudy. Scand J Infect Dis 2001;33:827—831.

23. De Pauw BE, Raemarkers JM, Schattenberg T, et al.Empirical and subsequent use of antibacterial agents in thefebrile neutropenic patient. J Intern Med 1997;740(Suppl):69—77.

24. Bodey GP, Buckley M, Sathe YS, et al. Quantitive relation-ships between circulating leukocytes and infection inpatients with acute leukemia. Ann Intern Med 1966;64:328—340.

25. Pizzo PA, Robichaud KJ, Wesley R, et al. Fever in pediatricand young adult patient with cancer. A prospective study of1001 episodes. Medicine (Baltimore) 1982;61:153—165.

26. Talcott JA, Finberg R, Mayer RJ, Goldman L. The medicalcourse of cancer patients with fever and neutropenia.Clinical identification of a low-risk subgroup at presentation.Arch Intern Med 1988;48:2561—2568.

27. Talcott JA, Siegel RD, Finberg R, Goldman L. Risk assessmentin cancer patients with fever and neutropenia: a prospec-tive, two-center validation of a prediction rule. J Clin Oncol1992;10:316—322.

28. Celebi H, Akan H, Akcaglayan E, Ustun C, Arat M. Febrileneutropenia in allogeneic and autologous peripheral bloodstem cell transplantation and conventional chemotherapyfor malignancies. Bone Marrow Transplant 2000;26:211—214.

29. Klastersky J, Paesmans M, Rubenstein EB, et al. Themultinational association for supportive care in cancer riskindex: a multinational scoring system for identifying low-riskfebrile neutropenic cancer patients. J Clin Oncol 2000;18:3038—3051.

30. Seropian S, Nadkami R, Jillella AP, et al. Neutropenicinfections in 100 patients with non-Hodgkin’s lymphoma orHodgkin’s disease treated with high-dose BEAM chemother-apy and peripheral blood progenitor cell transplant: out-patient treatment is a viable option. Bone Marrow Trans-plant 1999;23:599—605.

31. Manian FA. A prospective study of daily measurement of C-reactive protein in serum of adults with neutropenia. ClinInfect Dis 1995;21:114—121.

32. Santolaya ME, Cofre J, Beresi V. C-reactive protein: avaluable aid for the management of febrile children withcancer and neutropenic. Clin Infect Dis 1994;18:589—595.

S. Gencer et al.72