Journal of Antimicrobial Chemotherapy (1998) 41, Suppl. D, 13–24
Control of Gram-negative bacillary infections
Fever and infection as a consequence of neutropenia,
mainly in acute leukaemia or agranulocytosis, were first
described about 100 years ago.1,2Because of its rarity, little
attention was devoted to the syndrome until neutropenic
fever and infection became a significant clinical problem
with the common use of cytostatic therapy for cancer.
It was Bodey et al.3who clearly demonstrated the close
relationship between the level and the duration of
neutropenia and the frequency of infection in patients
treated with chemotherapy for acute leukaemia. A t that
time, bacteraemia complicating severe neutropenia was
mainly caused by Gram-negative bacilli, namely Pseudo -
monas aeruginosa, and was associated with a disastrous
90% mortality rate,4in spite of the availability of anti-
biotics active in vitro.
Two major events played a basic role in the change of
the mortality of Gram-negative bacillary infections in
neutropenic patients: (i) the wide acceptance of empirical
therapy and (ii) the optimal use of antibiotic combina-
It has been recognized that infection with Gram-
negative organisms, especially when P. aeruginosa was
involved, was often fulminant in neutropenia and killed
more than half of the patients within 48 h of onset.5In
addition, it became clear that neutropenia minimized
many signs and symptoms of infection and that fever was
often the only early manifestation of sepsis.6Based on
these observations, the concept of empirical therapy for
fever in neutropenic patients was proposed by Schimpff
et al.7In the study that launched the concept, a com-
bination (carbenicillin gentamicin) active against P.
aeruginosa and other Gram-negative bacilli was used
empirically as soon as neutropenic patients became
febrile, without waiting for clinical and/or microbiological
evidence of infection. There was a dramatic reduction in
the mortality associated with retrospectively documented
infection due to P. aeruginosa. A lthough it was unclear
whether the empirical approach or the effective com-
bination played the major role, the concept of empirical
therapy was very widely accepted without any further
controlled investigation. Of course, the early (empirical)
use of antibiotics for fever in neutropenic patients makes
the microbiological documentation of infection more
difficult. A s shown in Table I, summarizing data from trials
V III and IX of the International A ntimicrobial Therapy
Cooperative Group–European Organization for Research
and Treatment of Cancer (IA TCG-EORTC) in 1290 cases
of febrile neutropenia, microbiological documentation
was obtained in about 25% of the febrile patients, usually
through positive blood cultures. However, as many as 40%
of patients remained without any microbiological and/or
clinical documentation of infection. The respective res-
ponse rates to empirical therapy in these two groups of
neutropenic patients were 38% and 57%.8,9
The value of synergic antibiotic combinations for the
treatment of Gram-negative bacillary bacteraemia in
neutropenic patients has been demonstrated by Klastersky
et al.10in a series of studies, culminating in the IA TCG-
EORTC clinical comparative trial IV , which showed that a
full course of ceftazidime
regimen in which amikacin was discontinued early.11In
that study, the overall mortality rate in patients with acute
leukaemia and Gram-negative bacillary bacteraemia who
received the optimal therapy was 17% (definite infectious
amikacin was superior to a
Science and pragmatism in the treatment and prevention of
J . Klastersky
Service de Médecine Interne et L aboratoire d’Investigation Clinique H. Tagnon, Institut Jules Bordet,
Centre des Tumeurs de l’Université L ibre de Bruxelles, 1 rue Héger-Bordet, 1000 Brussels, Belgium
The following aspects of the management of patients with granulocytopenia and fever are
reviewed in this article: adaptation of initial antibiotic regimens to the recent changes in the
most common causative pathogens (namely a change from Gram-negative bacteria to Gram-
positive bacteria and fungi); subsequent modifications of the empirically administered treat-
ments; improvement of the host’s defence by reducing the duration of neutropenia; and
indications for out patient therapy of febrile episodes.
© 1998 The British Society for A ntimicrobial Chemotherapy
J . Klastersky
mortality was 8%), a dramatic difference from the 91%
figure reported 25 years earlier.4
Besides providing the benefit of synergy, amino-
glycoside-containing combinations have the advantage of
not leaving untreated a patient whose pathogen would be
resistant to -lactams. In IA TCG-EORTC trial I, the
combination of carbenicillin
carbenicillin cephalothin because infections caused by
microorganisms resistant to both
poorly;12in fact, most patients infected with doubly
resistant strains died. In IA TCG-EORTC trial III, in-
fections caused by -lactam-resistant and aminoglycoside-
sensitive organisms did not have an increased mortality
compared with those due to fully sensitive pathogens,
although their response rate to empirical therapy was
significantly lower.13These observations suggest that the
gentamicin was superior to
inclusion of the aminoglycoside into the empirical com-
bination might ‘buy time’ and allow for antibiotic
adjustment according to microbiological documentation.
Of course, the more frequent use of an aminoglycoside
can lead to greater toxicity and requires the monitoring
of blood concentrations. However, the value of synergic
aminoglycoside-containing combinations has only been
demonstrated in patients with prolonged and severe
granulocytopenia and concomitant Gram-negative bacter-
aemia. These conditions were found in only 75 (3.2%) of
2356 patients included in the last three IA TCG-EORTC
trials, so empirical aminoglycosides can be discontinued
early in all other patients.11
Changing microbiological pattern
During the last decade, Gram-negative bacilli have gradu-
ally been replaced by Gram-positive cocci as the cause of
microbiologically documented infections complicating
A s indicated in Table II, summarizing the blood culture
data from seven consecutive IA TCG-EORTC trials,
Gram-positive microorganisms (mainly Staphylococcus
epidermidis and various strains of streptococci) now
represent 70% of the bacteraemic isolates. That trend has
been observed universally. In addition, new pathogens
have emerged in neutropenic patients, but at a slower pace
than in A IDS patients, and resistant pathogens have
become more common. The resistance is either intrinsic,
such as the resistance of Candida krusei to imidazoles, or
acquired, as exemplified by recent outbreaks of vanco-
mycin-resistant enterococci in patients treated previously
with cephalosporins or vancomycin. Constant epidemio-
Table I. Infection documentation in IA TCG-EORTC
trials V III nd IX
Fever not related to infection
Table II. Single-organism bacteraemia in IA TCG-EORTC trials
I II III IVVV IIIIX
(1973–8) (1978–80)(1980–83) (1983–6)(1986–8)(1989–91)(1992–4)
febrile episodes (%)
Gram-negative bacteraemia (%)
Gram-positive bacteraemia (%)
other Gram-positive organisms
logical surveillance is thus necessary in patients with
compromised defences in order to detect the emergence of
rare and/or resistant pathogens.
The question of whether the emerging Gram-positive
organisms should be fully covered by empirical therapy
is still unresolved. Most studies of the introduction of
vancomycin in Gram-positive infections14–16used vanco-
mycin or teicoplanin because only these antibiotics
are effective against all the potential Gram-positive
pathogens. These early studies showed the superiority of
vancomycin-containing regimens. However, other authors
claim that infections caused by S. epidermidis, which are,
at the present time, the only pathogens in neutropenic
patients for which vancomycin is necessary, are quite
indolent and carry a very low mortality rate; thus, these
infections would not require empirical therapy before
microbiological documentation.17On the other hand,
streptococcal infections in neutropenic patients can be
associated with a high morbidity and a significant mortality
rate (c. 20%);15,18therefore, they should be adequately
covered empirically, at least in hospitals where these
streptococcal infections in neutropenic patients are
In some centres, the problem of streptococcal infec-
tions, namely in bone marrow transplant recipients, has
been solved by the pre-emptive use of vancomycin.19,20
Excessive use of glycopeptides could lead to the emer-
gence of resistance to vancomycin in other bacterial
species, such as staphylococci, so other approaches have
been used. These include the administration, as a first-line
therapy, of piperacillin–tazobactam, in combination with
amikacin. This is very effective against most species
of streptococci and provides good cover of most Gram-
negative organisms.9With such an empirical regimen,
S. epidermidis is not covered but, as it causes rather
indolent infections, vancomycin or teicoplanin can be
added at the time of microbiological documentation. This
approach avoids the empirical use of vancomycin or
teicoplanin in most patients and might help in keeping the
emergence of vancomycin-resistant microorganisms at a
A flow chart for use in chasing antimicrobial therapy in
neutropenic patients is given in the Figure.21
Prophylaxis of infection
Better hospital hygiene, the exclusive administration of
cooked food, chemoprophylaxis and the early use of
effective antibiotics in neutropenic patients have markedly
reduced the incidence of Gram-negative organisms as a
cause of infection in neutropenic patients. On the other
hand, the common use of intravenous lines and chemo-
therapy-induced mucositis might have, among other factors,
increased the frequency of Gram-positive organisms.
Clearly, these epidemiological changes cannot be explained
solely by the use of chemoprophylaxis with quinolones.
When quinolones are used for prevention of infection
in neutropenic patients, the rate of Gram-negative bac-
teraemia is reduced to 1–2%. The prevention of Gram-
positive infections is more difficult; co-trimoxazole and
penicillin have been used and found to be effective.20,22,23
The main problem with chemoprophylaxis is the emer-
gence of resistance; however, since the antibiotics used for
chemoprophylaxis in cancer patients are widely used in the
community, it is unlikely that their use in neutropenic
patients will significantly aggravate the overall situation.
Therefore, since chemoprophylaxis with quinolones has to
date proved very efficacious in preventing Gram-negative
infection during neutropenia, it would appear unwise to
deny it to our patients.24However, close monitoring for the
emergence of resistance is mandatory in the prophylactic-
ally treated patients and in their environment. Moreover,
further research to develop new agents for the prevention
of infection is necessary since past experience with the use
of antibiotics has taught us that microbes always end up by
finding ways of making our preventive strategies useless.
Prophylaxis of fungal and viral infection is another
important issue, but it is largely limited to the patients who
undergo an allogeneic bone marrow transplant. Imida-
zoles (like fluconazole) can effectively prevent candidiasis,
both superficial and deep-seated; muco-cutaneous lesions
due to herpes simplex can be prevented by acyclovir.25,26
The prophylaxis of cytomegalovirus (CMV ) and
Aspergillusspp. infections is particularly difficult, although
active agents against these organisms are available. A sper-
gillosis has been successfully prevented by the nasal
nebulization of amphotericin B,27in patients whose upper
respiratory passages are colonized by Aspergillus. Pre-
emptive administration of low doses of amphotericin B has
been effective in preventing fungal infections in bone
marrow transplanted patients receiving high doses of
corticosteroids for graft-versus-host disease (G-V HD).28
High-dose acyclovir and ganciclovir have been used in
patients with bone marrow transplantation to prevent
CMV infections. In patients with suspected CMV infec-
tions, ganciclovir has been used as pre-emptive therapy to
be administered to high-risk patients, perhaps those with
positive surveillance cultures, who have not yet presented
clear-cut clinical signs of infection.29
One might ask whether chemoprophylaxis of infection
in neutropenic patients represents very early therapy? A ll
the agents commonly used for chemoprophylaxis today
are systemically absorbed and thus may be considered as
The concept of chemoprophylaxis as a pre-emptive
therapy is useful because it unifies the approach to febrile
neutropenia as a global therapeutic approach. The popula-
tion of neutropenic patients is heterogeneous. Therefore
very early specific approaches (prophylaxis or pre-emptive
therapy) might be useful in some circumstances to
J . Klastersky
minimize morbidity and mortality while, in others, more
conservative therapies (empirical or microbiologically
oriented therapies) are more appropriate.
Such pre-emptive therapies, as already stated, have
been used mainly in patients undergoing bone marrow
transplantation and have consisted of penicillin and/or
vancomycin to prevent and/or treat early streptococcal
infection,23,30low-dose amphotericin B in patients receiving
high doses of corticosteroids28and ganciclovir when a
positive bronchoalveolar lavage for CMV has been
Following this concept, quinolone prophylaxis might be
viewed as a pre-emptive therapy for Gram-negative
bacillary infection aimed at reducing the need for broad-
spectrum empirical antibiotics to treat febrile episodes.31
Endpoints for evaluating therapy of febrile
Death is the most straightforward endpoint for evaluating
the efficacy of therapy for any potentially lethal condition.
The causes of death in febrile neutropenia are shown in
Table III; it summarizes the experience, at the Institut
Bordet, with patients who were enrolled in consecutive
IA TCG-EORTC studies of empirical therapy. It can be
seen that the overall mortality during febrile neutropenia
has decreased, in our centre, from 24% to 12% over the
last 20 years. Of the patients who died during the episode
of febrile neutropenia or immediately after it, during a
single hospitalization, infection was responsible for the
death in 34% and 39% in our first and second study,
respectively; the overall rate of infectious deaths was
3–4%.32,33Bleeding was an important cause of death in the
first review, but progression of cancer was more frequently
responsible for the fatal outcome during the second
period. These observations from our centre are supported
by other recent studies. In the recently published IA TCG-
EORTC study, extensive neoplastic disease was res-
ponsible for 32 (50%) of the 64 deaths during or after an
episode of febrile neutropenia.9A s shown in Table IV ,
recent IA TCG-EORTC trials (V II, IX and X I) of
empirical management of febrile neutropenia all show8,9,34
that the overall mortality ranges between 5% and 12%,
Figure. Guidelines for the diagnostic and therapeutic approach to febrile episodes in granulocytopenic patients (aA s per the latest
IA TCG-EORTC study; other regimens are acceptable. bContinue A B, without aminoglycoside and/or glycopeptide for 7 days).
with an infectious mortality rate between 1% and 3%.
Clearly, mortality cannot serve as a valid endpoint for
the evaluation of empirical therapy: the overall mortality
is largely related to factors other than infection, such
as bleeding or extensive neoplasia, and the infectious
mortality rate itself, including deaths from initial and
subsequent infections, is now extremely low.
Most authors, have used the response rate to evaluate the
efficacy of empirical therapy. Table IV summarizes the data
from three recently published IA TCG-EORTC trials, each
with a large number of patients, a similar degree of
neutropenia, an equal proportion of acute leukaemias and
a comparable incidence of bacteraemias.8,9,34It can be seen
that the overall response rates range from 52% to 74%;
since the ultimate outcome for most patients is excellent—
fewer than 10% fatalities—successful changes of antimicro-
bial therapy must have contributed to this improvement.
In the latest IA TCG-EORTC study,34which compared
meropenem monotherapy with ceftazidime
38% of the patients had a glycopeptide and 25% had an
antifungal agent added after the initial regimen had been
considered ineffective. Only the response to the initial
empirical regimen is indicative of its intrinsic efficacy and
should be the parameter for comparing different regimens.
The IA TCG-EORTC group considers any modification of
the initial regimen as a failure, even if the ultimate outcome
is satisfactory; others may call it ‘success with modification’.
Of course, in non-blinded studies, therapeutic modifi-
cations can be biased by the clinical impression of the
physician in charge. Future prospective studies should be
blinded more often and should adopt appropriate methods
to evaluate the modifications made during therapy of
febrile neutropenia; at the same time, new endpoints should
be sought: time to defervescence, duration of hospital-
ization and quality of life.
The concept of modifying the initial regimen to secure
Table III. Causes of death in febrile neutropenia
Causes of death
progression of cancer
a Sculier et al.32
b Rossi & Klastersky33
J . Klastersky
optimal success, the so-called ‘success with modification’,
has been introduced by Pizzo et al.35In this early study, the
success of initial empirical therapy was 64% and modi-
fications produced success in another 31%, resulting in an
overall success rate of 95%. In a more recent study, from the
same institution, comparing ceftazidime with imipenem,
the initial response rate was only 39%; however, after
modification the overall response rate improved to
A nother problem with the response rate, as an endpoint
to evaluate empirical therapy, is the heterogeneity of the
study population (or groups) in different investigations.
This makes it difficult to interpret the results towards a
change in practice. A s already mentioned, a recent trial36
that compared ceftazidime with imipenem claimed a 100%
response rate with virtually no mortality; however, the
mean duration of neutropenia was 8 days, the incidence of
bacteraemia was only 12% (two Gram-positive infections
for every Gram-negative) and the proportion of leukaemic
patients was not stated (44% of leukaemia and lymphoma
patients). In the IA TCG-EORTC studies (trials V III, IX
and X I) which are summarized in Table IV , the overall
success rate (as judged by our criteria) was 53% and the
mortality rate was 5–12%. However, in these latter studies
there were more patients with leukaemia, severe neutro-
penia and bacteraemia; it is interesting that of the patients
undergoing allogeneic bone marrow transplantation the
response rate was still 52% and the mortality 12%.
In another study, patients with leukaemia and bone
marrow transplants were compared with patients with
lymphoma and solid tumours randomized to receive
pipercillin tobramycin or ceftazidime; no difference was
observed between the two regimens, but the overall res-
ponse rate to unmodified therapy was 182/593 (31%)
in the patients with leukaemia/transplantation and
62/129 (48%) in the lymphoma/solid tumour group.37
Future studies might usefully look at more careful
selection of study populations and stratification of patients
by additional prognostic factors.
The number of variable factors in these studies of
empirical therapy of febrile neutropenia probably
accounts for the difficulty in demonstrating significant
differences between the various regimens.
A benefit of synergic combination therapy has been
demonstrated only in leukaemic patients with Gram-
negative bacteraemia and severe protracted granulo-
cytopenia.11They represent a very small subset of the
patients treated for febrile neutropenia. However, recent
studies suggest that the initial use of combination therapy
leads to fewer modifications of the initial regimen and
results in shorter hospital stays.38Table V summarizes the
response rates in bacteraemic patients from recent
studies;8,9,34the response rate with the regimens employed
today is worse for Gram-positive bacteraemia (25–50%)
than for Gram-negative bacteraemia (54–70%).
The cumulated data presented here are useful in
providing a rough indication of the overall effectiveness of
standard regimens in patients whose general charac-
teristics are reasonably well defined; they do not demon-
strate the superiority of any particular regimen.
Improvement of the host’s mechanisms of
The major types of immune deficit that clinicians
encounter can be divided into six broad categories: (i)
granulocytopenia, (ii) deficiencies of cellular immunity,
(iii) impairment of humoral immunity, (iv) obstruction of
a normal lumen by tumour/fibrosis, (v) central nervous
system dysfunction and (vi) damage to normal anatomical
barriers, such as the skin or mucosal surface.
The prevention of infection varies with each type of
deficit but four general approaches can be applied to all
patients: attempts to improve the patient’s immune
deficiency, reducing rates of acquisition of potential
pathogens, suppressing these organisms already colonizing
the patient that are likely to result in later infection, and
avoiding procedures that disrupt normal anatomical
A s indicated in Table V I, the various defects of the
host’s defences are associated with infections by different
pathogens with different clinical expressions. However,
Table V . Response rates in bacteraemic patients from recent IA TCG-EORTC trials of empirical
therapy of febrile neutropenia
Response rate in bacteraemia (%)
the potential list of pathogens responsible for infection in
cancer patients is certainly much longer than this.
The availability of recombinant myeloid colony-
stimulating factors (G-CSF and GM-CSF) has led to new
approaches in chemotherapy-induced neutropenia. Based
upon the available data, the following recommendations
seem appropriate. These growth factors are probably
useful in the setting of bone marrow transplantation for
reducing the severity and the duration of granulo-
cytopenia. They might also be useful for patients receiving
aggressive therapy likely to cause significant (
granulocytes/mm3) granulocytopenia lasting more than a
week. However, recent controlled studies in elderly
patients with acute myeloblastic leukaemia failed to
demonstrate a significant advantage for either G-CSF or
These factors may also prevent recurrence of sepsis that
occurred during a preceding episode of neutropenia and
allow the administration of chemotherapy at full doses and
without delays caused by persisting neutropenia.41
A s far as cellular immune deficiencies are concerned,
there have been some successful attempts to transfer
immune cells for specific diseases such as CMV . The
results, although promising, are still at an early stage.42
The use of vaccines is another possible approach, but
active vaccination has not been very successful in cancer
patients because of the poor antibody response.43Revac-
cination has a special role in bone marrow transplanted
Passive transfer of antibody may be useful in some
instances, such as the prevention of infections caused by
varicella-zoster virus;45on the other hand, the infusion
of anti-endotoxin antibodies has not been successful
in neutropenic patients,46nor has the use of CMV -
specific intravenous immunoglobulin effectively prevented
primary CMV infection and disease after bone marrow
Correction of humoral immune dysfunction with high
doses of non-specific intravenous immunoglobulin re-
placement has decreased the number of bacterial infec-
tions in some studies,48but this technique has not been
widely accepted, except perhaps for patients having under-
gone allogeneic bone marrow transplantation.49
A review of recent publications concerning fungal infec-
tions in cancer patients clearly confirms that protracted
severe granulocytopenia is a major risk factor for the
development of such infections. Other risk factors, such as
Table V I. Factors predisposing to infection in patients with cancer
Host defence defect Pathogen Clinical infection
Candida spp. including
L isteria monocytogenes
L egionella pneumophila
Cellular immune deficiencyencephalitis
Humoral immune dysfunctionbacteraemia
J . Klastersky
the use of broad-spectrum antibacterial agents, cortico-
steroids and central venous catheters, as well as the type of
cytotoxic chemotherapy, the performance status of the
patient and the stage of the underlying disease, also
predispose to the development of invasive fungal infec-
Fungal infection can be documented initially in 5% of
patients with febrile neutropenia; this figure has not
changed much over the years. It is obvious that bacterial
and fungal sepsis can coexist and that the bacteraemia may
mask the (more difficult to document) fungal infection;
this later manifests itself as a persisting or recurring fever,
after the eradication of bacteraemia by empirically pre-
scribed antibiotics. A s neutropenia persists, the risk of
further fungal infection increases; many fevers in patients
with or after prolonged neutropenia are caused by
A s shown in Table V II, these fungal infections represent
a significant proportion of further infections which are
observed during the course of febrile neutropenia; more-
over, the mortality associated with these further fungal
infections is much higher than that due to further bacterial
It has been suggested that certain categories of granulo-
cytopenic patients may benefit from antifungal pro-
phylaxis and/or empirical therapy; conversely, there are
other neutropenic patients who might benefit only
marginally from such strategies.50
Imidazoles, namely fluconazole, for the prevention of
local and systemic infections caused by Candida spp.
have been effective in granulocytopenic patients;53oral
polyenes are also active, but less well tolerated. So far, the
development of resistance has not been a major problem,
but it is of potential concern, especially if prolonged
administration becomes routine.
Empirical therapy with antifungal agents in neutropenic
patients who do not respond to broad-spectrum antibiotics
has become an accepted practice;54it is justified by the
frequency of fungal infections in granulocytopenic
patients and by the difficulties in making an early specific
diagnosis. However, the precise indications for its use are
still unresolved; as already mentioned, in recent studies of
empirical antibacterial therapy, about 25% of patients are
eventually given antifungals.34
In patients at high risk of developing severe fungal
infection, such as those receiving high-dose corticosteroid
therapy for GV HD or after allogeneic bone marrow trans-
plantation, the early pre-emptive administration of a low
dose of amphotericin B was found to be effective in
reducing the development of systemic fungal infection.28
In terms of choice of therapy, amphotericin B is still the
standard approach. This particularly applies to empirical
treatment given before the recognition of a specific
offending pathogen, when both Candida spp. and Asper -
gillusspp. need to be covered. When the infection is known
to be caused by a sensitive variety of Candida spp.,
imidazoles might be as effective as amphotericin B; how-
ever, this has not been established in neutropenic patients.55
To what extent liposomal preparations of amphotericin
B will provide increased clinical efficacy remains to be
proven in controlled trials; nevertheless, it seems clear that
the side effects related to amphotericin B can be reduced
by the use of the liposomal preparations, allowing the
administration of increased dosages.56
Because severe and prolonged granulocytopenia plays
such a major predisposing role for fungal infection, it is
possible that the use of bone marrow stimulating factors
might prove particularly effective in decreasing the
frequency and severity of these infections.57
There is a definite need for new diagnostic methods and
Table V II. Further infections: incidence and mortality
De Pauw et al.37
Cometta et al.9
Total evaluable patients
Microbiologically demonstrated infections
Clinically demonstrated infections
active agents for the diagnosis and treatment of severe
fungal infections.58Until such progress is made, the
optimal approach to these common and severe com-
plications of cancer therapy will probably consist of early
administration of the least toxic and optimally effective
preparations to those patients known to be at highest risk.
A s already stated, febrile neutropenic patients do not
represent a homogenous population and all these patients
do not have the same risk of developing serious com-
plications or of death. Moreover, the incidence of rapidly
fatal infections, even in patients with severe neutropenia,
has markedly decreased over the last few years. Therefore,
the standard practice of hospitalizing all febrile neutro-
penic patients for antibiotic therapy has become an
important research issue, particularly in the current health
care environment, where cost is being increasingly
Following the work of Talcott et al.,59which led to the
recognition of well-defined risk groups among febrile
neutropenic patients, there have been several attempts to
treat cancer patients with fever and neutropenia in an
outpatient setting. Talcott et al.60reported the experience
of 30 patients without progressive neoplasia and/or non-
neoplastic serious morbidity after early discharge and
home treatment with intravenous antibiotics; however,
only 16 (53%) responded to the initial regimen (mezlo-
cillin gentamicin or ceftazidime) and nine had to be
readmitted for treatment of serious complications or
surveillance of persisting fever. Rubenstein et al.61selected
their febrile neutropenic patients for home therapy on
the basis of the lack of any comorbidity requiring
hospitalization. They compared an oral regimen (cipro-
floxacin clindamycin) with intravenous antibiotics
(aztreonam clindamycin); the intravenous regimen was
associated with a response of 88%. There is no clear
explanation for the difference in response rates in these
two pivotal studies, since the selection criteria were quite
similar, as were the underlying diseases in both groups;
perhaps subtle differences in patient selection and a better
coverage of Gram-positive pathogens in the second study
might have played a role. It is important to stress that no
patient died in either study.
Other strategies included sequential prophylactic anti-
biotics (rifampicin ciprofloxacin) followed, in the case of
fever, by parenteral vancomycin
and oral ciprofloxacin in patients who had undergone
autologous bone marrow transplantation;62the incidence
of documented infection was greatly reduced and
bacteraemias were virtually eliminated; patients could be
discharged early to the outpatient setting.
Y et another approach is auto-medication with
pefloxacin co-amoxiclav by the patients at home, to be
tobramycin once daily
started if fever and/or chills appear;63nine failures
were seen in 68 patients; two patients presented with a
staphylococcal bacteraemia and one of them died. A uto-
medication in febrile neutropenic patients is questionable,
because possible signs predicting a poor outcome (sepsis,
vasomotor instability, pneumonia) might go undetected
and, in addition, microbiological documentation is not
A ll these observations suggest that many patients with
solid tumours and with an expected duration of neutro-
penia of 7 days, can have their febrile episode managed
successfully in an out-patient setting; however, more con-
trolled studies are needed to evaluate the cost-effective-
ness and the benefits, in terms of efficacy and quality of
life, of this approach.
The use of prophylactic antibiotics and bone marrow
growth promoting factors might prove to be more
beneficial and cost-effective in low-risk patients allowed
home early than in those with more severe and/or
protracted neutropenia. The other patients, namely those
with haematological malignancies, should probably start
therapy for febrile neutropenia as in-patients; however, we
need to develop early discharge criteria and home thera-
peutic strategies for these patients as well. Perhaps
sequential strategies such as empirical intravenous anti-
biotics followed by an early switch to oral medication in
responders, as recently reported,65might prove not only to
be cost-effective but also to improve quality of life in
patients with moderately severe neutropenia.
Many problems with empirical therapy of febrile
neutropenia remain to be solved; these consist mainly of
the increase in the frequency of Gram-positive micro-
organisms, some of which are methicillin-resistant; the
increase in the number of -lactamases in Gram-negative
microorganisms; the emergence of new and/or resistant
pathogens depending on institutional practices; and the
appearance of specific clinical and/or microbiological
syndromes according to evolving cancer therapy.
Febrile neutropenia has been a changing syndrome over
the last 20 years. The perspectives and goals that we face
today are mainly: the constant adaptation of antibacterial
prophylactic and therapeutic regimens in response to the
emergence of resistant strains; the definition of prognostic
factors influencing the outcome of febrile neutropenia; the
introduction of pragmatic algorithms for adaptation of
therapy; the introduction of ambulatory and/or home
therapy for certain categories of patients; the definition of
indications for the use of cytokines to restore earlier bone
marrow function and to help with home management of
febrile neutropenia; and, the recognition of risk factors for
fungal infections and the improvement of our diagnostic as
well as therapeutic strategies.
J . Klastersky
1. Brown, P. K. (1992). In Histoire Illustrée de l’Hématologie de
l’Antiquitée à Nos Jours. (Bernard, J., Bessi, M. & Binet, J. L., Eds)
Editions R. Dacosta, Paris.
2. Schultz, W. (1992). In: (Bernard, J., Bessi, M. & Binet, J. L., Eds)
Histoire Illustrée de l’Hématologie de l’Antiquitée à Nos Jours.
Editions R. Dacosta, Paris.
3. Bodey, G. P., Buckley, M., Sathe, Y. S. & Freireich, F. J. (1966).
Quantitative relationships between circulating leukocytes and in-
fection in patients with acute leukemia. Annals of Internal Medicine
4. McCabe, W. R. & Jackson, G. G. (1962). Gram-negative
bacteremia. II. Clinical, laboratory, and therapeutic observations.
Archives of Internal Medicine110, 856–64.
5. Bodey, G. P., Middleman, E., Umsawadi, T. & Rodriguez, V.
(1972). Infections in cancer patients. Results with gentamicin
sulfate therapy. Cancer 29, 1697–701.
6. Sickles, E. A., Greene, W. H. & Wiernik, P. H. (1975). Clinical
presentation of infection in granulocytopenic patients. Archives of
Internal Medicine 135, 715–9.
7. Schimpff, S., Satterlee, W., Young, V. M. & Serpick, A. (1971).
Empiric therapy with carbenicillin and gentamicin for febrile patients
with cancer and granulocytopenia. New England Journal of
Medicine 284, 1061–5.
8. EORTC International Antimicrobial Therapy Cooperative Group.
(1993). Efficacy and toxicity of single daily doses of amikacin and
ceftriaxone versus multiple daily doses of amikacin and ceftazidime
for infection in patients with cancer and granulocytopenia. Annals
of Internal Medicine 119, 584–93.
9. Cometta, A., Zinner, S., de Bock, R., Calandra, T., Gaya, H.,
Klastersky, J. et al. (1995). Piperacillin–tazobactam plus amikacin
versus ceftazidime plus amikacin as empiric therapy for fever in
granulocytopenic patients with cancer. Antimicrobial Agents and
Chemotherapy 39, 445–52.
10. Klastersky, J., Daneau, D., Swings, G. & Weerts, D. (1974).
Antibacterial activity in serum and urine as a therapeutic guide in
bacterial infections. Journal of Infectious Diseases129, 187–93.
11. EORTC International Antimicrobial Therapy Cooperative
Group. (1987). Ceftazidime combined with a short or long course of
amikacin for empirical therapy of Gram-negative bacteremia in
cancer patients with granulocytopenia. New England Journal of
Medicine 317, 1692–8.
12. EORTC International Antimicrobial Therapy Project Group.
(1978). Three antibiotic regimens in the treatment of infection in
febrile granulocytopenic patients with cancer. Journal of Infectious
Diseases 137, 14–29.
13. Klastersky, J., Glauser, M. P., Schimpff, S. C., Zinner, S. H.,
Gaya, H. & the European Organization for Research on Treatment
of Cancer Antimicrobial Therapy Group. (1986). Prospective ran-
domized comparison of three antibiotic regimens for empirical
therapy of suspected bacteremic infection in febrile granulocytopenic
patients. Antimicrobial Agents and Chemotherapy 2 9, 263–70.
14. Karp, J. E., Dick, J. D., Angelopulos, C., Charache, P., Green,
L., Burke, P. J. et al. (1986). Empiric use of vancomycin during
prolonged treatment-induced granulocytopenia. Randomized,
double-blind, placebo-controlled clinical trial in patients with acute
leukemia. American Journal of Medicine81, 237–42.
15. Shenep, J. L., Hughes, W. T., Robertson, P. K., Blankenship,
K. R., Baker, D. K., Meyer, W. H. et al. (1988). Vancomycin,
ticarcillin, and amikacin compared with ticarcillin–clavulanate and
amikacin in the empirical treatment of febrile, neutropenic children
with cancer. New England Journal of Medicine319, 1053–8.
16. EORTC International Antimicrobial Therapy Cooperative
Group. (1991). Vancomycin added to empirical combination anti-
biotic therapy for fever in granulocytopenic cancer patients. Journal
of Infectious Diseases 163, 951–8.
17. Rubin, M., Hathorn, J. W., Marshall, D., Gress, J., Steinberg,
S. M. & Pizzo, P. A. (1988). Gram-positive infections and the use of
vancomycin in 550 episodes of fever and neutropenia. Annals of
Internal Medicine 108, 30–5.
18. Awada, A., Van der Auwera, P., Meunier, F., Daneau, D. &
Klastersky, J. (1992). Streptococcal and enterococcal bacteremia in
patients with cancer. Clinical Infectious Diseases 15, 33–48.
19. Attal, M., Schlaifer, D., Rubie, H., Huguet, F., Charlet, J.-P.,
Bloom, E. et al. (1991). Prevention of Gram-positive infections
after bone marrow transplantation by systemic vancomycin: a
prospective, randomized trial. Journal of Clinical Oncology 9,
20. Lamy, T., Michelet, C., Dauriac, C., Grulois, I., Donio, P.-Y. &
Le Prisé, P.-Y. (1993). Benefit of prophylaxis by intravenous sys-
temic vancomycin in granulocytopenic patients: a prospective,
randomized trial among 59 patients. Acta Haematologica 90,
21. Klastersky, J. (1993). Febrile neutropenia. Current Opinion in
Oncology 5, 625–32.
22. EORTC International Antimicrobial Therapy Cooperative
Group. (1984). Trimethoprim–sulfamethoxazole in the prevention of
infection in neutropenic patients. Journal of Infectious Diseases
23. EORTC International Antimicrobial Therapy Cooperative
Group. (1994). Reduction of fever and streptococcal bacteremia in
granulocytopenic patients with cancer. Journal of the American
Medical Association 272, 1183–9.
24. Ball, P. (1995). Editorial response: is resistant Escherichia
coli bacteremia an inevitable outcome for neutropenic patients
receiving a fluoroquinolone as prophylaxis? Clinical Infectious
Disease 20, 561–3.
25. Saral, R., Burns, W. H., Laskin, O. L., Santos, G. W. &
Lietman, P. S. (1981). Acyclovir prophylaxis of herpes-simplex-
virus infections. New England Journal of Medicine305, 63–7.
26. Goodman, J. L., Winston, D. J., Greenfield, R. A.,
Chandrasekar, P. H., Fox, B., Kaizer, H. et al. (1992). A controlled
trial of fluconazole to prevent fungal infections in patients under-
going bone marrow transplantation. New England Journal of
Medicine 326, 845–51.
27. Beyer, J., Barzen, G., Risse, G., Weyer, C., Miksits, K.,
Dullenkopf, K. et al. (1993). Aerosol amphotericin B for prevention
of invasive pulmonary aspergillosis. Antimicrobial Agents and
Chemotherapy 37, 1367–9.
28. O’Connell, M. R., Schmidt, G. M., Tegtmeier, B. R., Faucett, C.,
Fahey, J. L., Ito, J. et al. (1994). Prediction of systemic fungal
infection in allogeneic marrow recipients: impact of amphotericin
prophylaxis in high-risk patients. Journal of Clinical Oncology 12,
29. Goodrich, J. M., Mori, M., Gleaves, C. A., Du Mond, C., Cays,
M., Ebeling, D. F. et al. (1991). Early treatment with ganciclovir to
prevent cytomegalovirus disease after allogeneic bone marrow
transplantation. New England Journal of Medicine325, 1601–7.
30. Broun, E. R., Wheat, J. L., Kneebone, P. H., Sundblad, K.,
Hromas, R. A. & Tricot, G. (1994). Randomized trial of the addition
of Gram-positive prophylaxis to standard antimicrobial prophylaxis
for patients undergoing autologous bone marrow transplantation.
Antimicrobial Agents and Chemotherapy38, 576–9.
31. Karp, J. E., Merz, W. G., Hendricksen, C., Laughton, B.,
Redden, T., Bamberger, B. J. et al. (1987). Oral norfloxacin for
prevention of Gram-negative bacterial infections in patients with
acute leukemia and granulocytopenia. A randomized, double-blind,
placebo-controlled trial. Annals of Internal Medicine 106, 1–7.
32. Sculier, J. P., Weerts, D. & Klastersky, J. (1984). Causes of
death in febrile granulocytopenic cancer patients receiving empiric
antibiotic therapy. European Journal of Cancer and Clinical
Oncology 20, 55–60.
33. Rossi, C. & Klastersky, J. (1996). Initial empirical antibiotic
therapy for neutropenic fever: analysis of the causes of death.
Supportive Care in Cancer4, 207–12.
34. EORTC International Antimicrobial Therapy Cooperative
Group and the GIMEMA Infection Program. (1995). Meropem
versus ceftazidime plus amikacin as empirical therapy for fever in
cancer patients with granulocytopenia. Canadian Journal of Infec-
tious Diseases 6, Suppl. C, 308, Abstract 1083.
35. Pizzo, P. A., Hathorn, J. W., Hiemenz, J., Browne, M.,
Commers, J., Cotton, D. et al. (1986). A randomized trial comparing
ceftazidime alone with combination antibiotic therapy in cancer
patients with fever and neutropenia. New England Journal of
Medicine 315, 552–8.
36. Freifeld, A. G., Walsh, T., Marshall, D., Gress, J., Steinberg, S.
M., Hathorn, J. et al. (1995). Monotherapy for fever and neutropenia
in cancer patients: a randomized comparison of ceftazidime versus
imipenem. Journal of Clinical Oncology13, 165–76.
37. De Pauw, B. E., Deresinski, S. C., Feld, R., Lane-Allman, E. F.
& Donnelly, J. P. for the Intercontinental Antimicrobial Study Group.
(1994). Ceftazidime compared with piperacillin and tobramycin for
the empiric treatment of fever in neutropenic patients with cancer.
Annals of Internal Medicine120, 834–44.
38. Elting, L. S. & Bodey, G. P. (1995). Outcomes of bacteremia in
neutropenic cancer patients: influence of major organ or tissue
infection. Supportive Care in Cancer3, 331–3, Abstract no. 2.
39. Dombret, H., Chastang, C., Fenaux, P., Reiffers, J.,
Bordessoule, D., Bouabdallah, R. et al. (1995). A controlled study of
recombinant human granulocyte colony-stimulating factor in elderly
patients after treatment for acute myelogenous leukemia. New
England Journal of Medicine332, 1678–83.
40. Stone, R. M., Berg, D. T., George, S. L., Dodge, R. K.,
Paccuicci, P. A., Schulman, P. et al. (1995). Granulocyte–
macrophage colony-stimulating factor after initial chemotherapy for
elderly patients with primary acute myelogenous leukemia. New
England Journal of Medicine332, 1671–7.
41. Vose, J. M. & Armitage, J. O. (1995). Clinical applications of
hematopoietic growth factors. Journal of Clinical Oncology 13,
42. Riddell, S. R., Watanabe, K. S., Goodrich, J. M., Li, C. R.,
Agha, M. E. & Greenberg, P. D. (1992). Restoration of viral im-
munity in immunodeficient humans by the adoptive transfer of T cell
clones. Science 257, 238–41.
43. Klastersky, J., Mommen, P., Cantraine, F. & Safary, A. (1986).
Placebo controlled pneumococcal immunization in patients with
bronchogenic carcinoma. European Journal of Cancer and Clinical
Oncology 22, 807–13.
44. Bron, D. (1995). Bone marrow transplantation. In Handbook of
Supportive Care in Cancer (Klastersky, J., Schimpff, S. C. & Senn,
H. J., Eds), pp. 491–510. Marcel Dekker, New York, NY.
45. Hardy, I., Ghershon, A. A., Steinberg, S. P. & LaRussa, P.
(1991). The incidence of zoster after immunization with live
attenuated varicella vaccine. A study in children with leukemia.
New England Journal of Medicine325, 1545–50.
46. Ziegler, E. J., McCutchan, J. A., Fierer, J., Glauser, M. P.,
Sadoff, J. C., Douglas, H. et al. (1982). Treatment of Gram-negative
bacteremia and shock with human antiserum to a mutant
Escherichia coli. New England Journal of Medicine307, 1225–30.
47. Bowden, R. A., Fisher, L. D., Rogers, K., Cays, M. & Meyers, J.
D. (1991). Cytomegalovirus (CMV)-specific intravenous immuno-
globulin for the prevention of primary CMV infection and disease
after marrow transplant. Journal of Infectious Diseases164, 483–7.
48. Chapel, H. M., Lee, M., Hargreaves, R., Pamphilon, D. H. &
Prentice, A. G. for the UK Group for Immunoglobulin Replacement
Therapy in Multiple Myeloma. (1994). Randomised trial of intra-
venous immunoglobulin as prophylaxis against infection in plateau-
phase multiple myeloma. Lancet 343, 1059–63.
49. Sullivan, K. M., Kipecky, K. J., Jocom, J., Fisher, L., Buckner,
C. D., Meyers, J. D. et al. (1990). Immunomodulatory and anti-
microbial efficacy of intravenous immunoglobulin in bone marrow
transplantation. New England Journal of Medicine323, 705–12.
50. Guiot, H. F. L., Fibbe, W. E. & van’t Wout, J. W. (1994). Risk
factors for fungal infection in patients with malignant hematologic
disorders: implications for empirical therapy and prophylaxis.
Clinical Infectious Diseases18, 525–32.
51. Walsh, T. J., Hiemenz, J. & Pizzo, P. A. (1994). Evolving risk
factors for invasive fungal infections—all neutropenic patients are
not the same. Clinical Infectious Diseases18, 793–8.
52. Klastersky, J. (1995). Therapy of infections in cancer patients.
In Handbook of Supportive Care in Cancer (Klastersky, J.,
Schimpff, S. C. & Senn, H. J., Eds), pp. 1–44. Marcel Dekker, New
53. Slavin, M. A., Osborne, B., Adams, R., Levenstein, M. J.,
Schoch, H. G., Feldman, A. R. et al. (1995). Efficacy and safety of
fluconazole prophylaxis for fungal infections after marrow trans-
plantation—a prospective, randomized, double-blind study. Journal
of Infectious Diseases 171, 1545–52.
54. EORTC International Antimicrobial Therapy Cooperative
Group. (1989). Empiric antifungal therapy in febrile granulo-
cytopenic patients. American Journal of Medicine86, 668–72.
55. Rex, J. H., Bennett, J. E., Sugar, A. M., Pappas, P. G., van der
Horst, C. M., Edwards, J. E. et al. (1994). A randomized trial
comparing fluconazole with amphotericin B for the treatment of
candidemia in patients without neutropenia. New England Journal
of Medicine 331, 1325–30.
56. Tollemar, J., Ringden, O. & Tyden, G. (1990). Liposomal
amphotericin B (AmBisome)treatment in solid organ and bone
marrow transplant recipients. Efficacy and safety evaluation.
Clinical Transplantation 4, 167–75.
57. Bodey, G. P., Anaissie, E., Gutterman, J. & Vadhan-Raj, S.
(1993). Role of granulocyte–macrophage colony-stimulating factor
J . Klastersky
as adjuvant therapy for fungal infection in patients with cancer.
Clinical Infectious Diseases 17, 705–7.
58. Klastersky, J. (1995). Prevention and therapy of fungal infec-
tions in cancer patients. A review of recently published information.
Supportive Care in Cancer3, 393–401.
59. Talcott, J. A., Siegel, R. D., Finberg, R. & Goldman, L. (1992).
Risk assessment in cancer patients with fever and neutropenia: a
prospective, two-center validation of a prediction rule. Journal of
Clinical Oncology 10, 316–22.
60. Talcott, J. A., Whalen, A., Clark, J., Rieker, P. P. & Finberg, R.
(1994). Home antibiotic therapy for low-risk cancer patients with
fever and neutropenia: a plot study of 30 patients based on a
validated prediction rule. Journal of Clinical Oncology12, 107–14.
61. Rubenstein, E. B. & Rolston, K. V. I. (1995). Outpatient
treatment of febrile neutropenic patients with cancer. European
Journal of Cancer 31A, 2–4.
62. Gilbert, C., Meisenberg, B., Bredenburgh, J., Ross, M.,
Hussein, A., Perfect, J. et al. (1994). Sequential prophylactic oral
and empiric once-daily parenteral antibiotics for neutropenia and
fever after high-dose chemotherapy and autologous bone marrow
support. Journal of Clinical Oncology12, 1005–11.
63. Gardembas-Pain, M., Desablens, B., Sensebe, L., Lamy, T.,
Ghandour, C. & Boasson, M. (1991). Home treatment of febrile
neutropenia: an empirical oral antibiotic regimen. Annals of
Oncology 2, 485–7.
64. Klastersky, J. (1993). Febrile neutropenia. Current Opinion in
Oncology 5, 625–32.
65. Iomiak, A. T., Yau, J. C., Huan, S. D., Cripps, M. C., Goel, R.,
Perrault, D. J. et al. (1994). Duration of intravenous antibiotics for
patients with neutropenic fever. Annals of Oncology5, 441–5.