Diagnosis of Invasive Candidiasis in Neutropenic Children with Cancer
by Determination of D-Arabinitol/L-Arabinitol Ratios in Urine
BERTIL CHRISTENSSON,1* THOMAS WIEBE,2 CHRISTINA PEHRSON,3
AND LENNART LARSSON3
Departments of Infectious Diseases,1 Pediatrics,2 and Medical Microbiology,3
Lund University Hospital, S-221 85, Lund, Sweden
Received 26 August 1996/Returned for modification 29 October 1996/Accepted 3 December 1996
Determination of D-arabinitol/L-arabinitol ratios (referred to as D/L-arabinitol ratios) in urine as a tool for
the diagnosis of invasive candidiasis was investigated in a prospective study comprising 100 children with
cancer. The analyses were made by gas chromatography. Positive D/L-arabinitol ratios were found for 10 of 10
children with confirmed invasive candidiasis, 12 of 23 patients undergoing empiric antifungal chemotherapy,
and 4 of 67 children not receiving antifungal treatment. D/L-Arabinitol ratios were positive 3 to 31 days
(median, 12 days) before the first culture-positive blood sample was drawn or empiric therapy was initiated.
The regular monitoring of D/L-arabinitol ratios in urine holds great promise as a sensitive method for
diagnosing invasive candidiasis in immunocompromised children with cancer. Moreover, it may be possible to
use an early rise in D/L-arabinitol ratios as a basis for the institution of antifungal chemotherapy and as a
means of avoiding unnecessary treatment with potentially toxic antifungal agents.
Invasive candidiasis is becoming increasingly common due
to the growing number of immunocompromised hosts. An
early and accurate diagnosis is of importance for improving
survival resulting from the institution of antifungal chemotherapy
(2, 15) and decreasing the unnecessary use of toxic antifungal
agents such as amphotericin B. Unfortunately, currently
available methods are not sufficiently sensitive and specific for
diagnosing invasive candidiasis (11, 19). Therefore, empiric
antifungal chemotherapy has been advocated for persistently
febrile granulocytopenic patients not responding to broadspectrum
antibiotic therapy (21). Although such empiric therapy
has been shown to decrease the frequency, morbidity, and
mortality of invasive fungal infection (21), inevitably, it also
leads to a potentially hazardous and extremely costly overtreatment
of some patients.
D-Arabinitol is a major metabolite of most Candida species,
and both D-arabinitol and L-arabinitol are present in normal
serum and urine. In an earlier study (9), we developed a gas
chromatographic method to determine the relative amounts of
D-arabinitol and L-arabinitol (referred to as the D/L-arabinitol
ratio) in urine. Elevated urine D/L-arabinitol ratios have previously
been found in a small number of patients with invasive
candidiasis (9, 16). Here, we report the results of a prospective
study of pediatric oncology patients designed to determine the
value of monitoring D/L-arabinitol ratios in urine for the early
diagnosis of invasive candidiasis.
MATERIALS AND METHODS
Patients. From March 1992 through October 1995, 100 children (age range, 1
to 17 years; mean age, 9 years) were prospectively studied at the Department of
Pediatrics, Division of Oncology, Lund University Hospital. All children were
considered to be at high risk for invasive candidiasis. Their malignant diagnoses
were acute leukemia for 47% of the patients, lymphomas for 13% of the patients,
and Wilms’ tumor for 14% of the patients; the remaining patients had various
solid tumors. All patients had central venous catheters and were receiving cytotoxic
chemotherapy. Broad-spectrum antibiotic therapy was instituted when patients
were febrile and granulocytopenic. All hospitalized patients were given
nystatin orally. During periods of hospitalization, the aim was to collect urine
samples at least twice weekly. The children who were too young to deliver urine
samples spontaneously and those who delivered only one sample were excluded
from the study. In all, 1,076 urine samples were collected.
A febrile (.38.38C), neutropenic (blood neutrophil counts, ,0.5 3 109/liter)
patient not responding to broad-spectrum antibiotic treatment was defined as
having invasive candidiasis when blood cultures were positive for Candida and/or
when Candida species were cultured or proven histopathologically in tissue
samples from normally sterile locations. Invasive candidiasis was clinically suspected
when a febrile, neutropenic patient did not respond to antibiotic treatment,
and in such cases, antifungal chemotherapy was empirically instituted even
when blood cultures were negative for Candida. Patients were defined as having
bacteremia when they responded to antibiotic treatment and blood cultures were
positive for bacteria but negative for Candida. Patients were also evaluated
during nonneutropenic periods, when all blood cultures were negative for Candida
and bacteria and no antifungal treatment was being given.
For patients with confirmed or suspected candidiasis, blood for culture was
always drawn at the time of urine collection. Additional blood cultures were also
performed once or twice daily for many of these patients. On the other hand,
during nonneutropenic periods without clinical suspicion of fungal infection, the
twice-weekly urine sampling was not always accompanied by blood culturing.
For comparison, single urine samples from 56 healthy nonhospitalized children
(age range, 1 to 15 years; mean age, 5 years) were also studied.
Samples. The urine samples were divided into 1-ml aliquots and were then
stored at 2208C before analysis for D/L-arabinitol ratios (see below). A biphasic
blood culture system (Septi-Check; Roche Products, Ska¨rholmen, Sweden) was
used to culture bacteria and fungi. For detection of fungi, blood was cultured
aerobically for 5 days at 378C, followed by 9 days at 308C; the cultures were
inspected daily for turbidity and were also shaken twice daily. Tissue samples
were cultured on Sabouraud agar and on agar with 4% horse erythrocytes for 7
days at 308C. Typing of the Candida organisms to the species level was done by
testing for production of chlamydospores and fermentation of glucose, galactose,
saccharose, maltose, lactose, and trehalose on rice agar (10).
D/L-Arabinitol analysis. Urine samples were filtered (0.45-mm-pore-size cellulose
acetate filters; Schleicher & Schuell, Dassel, Germany), and a 5- to 15-ml
portion of the filtrate was dried under a stream of nitrogen. Trifluoroacetic
anhydride and hexane (200 ml each) were added, and the preparations were
heated at 808C for 10 min and again dried. Thereafter, hexane (200 to 500 ml)
was added, and a 1- to 3-ml aliquot was used to obtain D/L-arabinitol ratios by gas
chromatography. As described previously (9), the method has an interassay
coefficient of variation of 7.7%. The total time for handling and analysis of each
sample was less than 1 h.
Study design. Samples were coded and were then submitted to the laboratory
and analyzed in blind batches. The results were not available to the pediatricians
treating the patients to ensure that the initiation of antifungal therapy would be
based solely on clinical observations and the results of blood or tissue cultures
and histopathology. The following was recorded for each patient: cytotoxic and
radiological therapy, antimicrobial therapy and prophylaxis, microbiological results,
and neutropenic and febrile events. Student’s two-tailed t test was used to
make comparisons between the D/L-arabinitol ratios for different groups. The
study was approved by the Ethics Committee at the Medical Faculty, University
* Corresponding author. Phone: 46 46 17 10 00. Fax: 46 46 13 74 14. of Lund.
636
RESULTS
The D/L-arabinitol ratios (mean 6 standard deviation) for
the 56 healthy nonhospitalized children were 2.0 6 0.6, and for
the 95 children with cancer, the ratios were 2.5 6 0.7 (P ,
0.01) during nonneutropenic periods. The latter group was
used to define the upper cutoff limit (mean ratio 1 3 standard
deviations, corresponding to a value of 4.6), and values of .4.6
were considered positive.
Ten patients with long-standing fever and neutropenia were
diagnosed with invasive candidiasis on the basis of blood cultures,
tissue cultures, and/or histopathology, and all these patients
had positive peak D/L-arabinitol results (Table 1; Fig. 1,
group A). By using multiple blood cultures, C. albicans was
isolated from five patients, C. parapsilosis was isolated from
three patients, C. tropicalis was isolated from one patient, and
C. glabrata was isolated from one patient. For the last patient,
postmortem tissue cultures were also positive for C. albicans.
Urine samples were available from 7 of 10 patients before
blood cultures were positive and showed increased D/L-arabinitol
ratios 3, 3, 4, 8, 9, 16, and 21 days, respectively, before the
first culture-positive blood sample was drawn. The time course
of the D/L-arabinitol ratios in one of these patients is presented
in Fig. 2. Five patients died from invasive candidiasis, and in
four of these patients the D/L-arabinitol ratio was positive 3, 8,
9, and 21 days, respectively, before the first culture-positive
blood sample was drawn (Table 1).
Antifungal chemotherapy (amphotericin B or fluconazole)
was instituted empirically (due to a lack of response after 3 to
5 days of broad-spectrum antibiotic treatment) for 23 neutropenic
and febrile patients with blood cultures negative for
Candida and bacteria. Twelve children had positive peak D/Larabinitol
ratio results (Table 2; Fig. 1, group B). Ten of 12
TABLE 1. Peak D/L-arabinitol ratios for febrile, neutropenic children with cancer and confirmed invasive candidiasis
Patient no. Primary disease Species
Peak D/Larabinitol
ratio
Time interval (days)
between positive
D/L-arabinitol
ratio and positive
blood culture
No. of positive
samples/no. of
all samplesa
Diagnosis based on: Patient
status
1 ALLb C. albicans 30 NAc 3/24 3 blood cultures Dead
2 NHLd C. albicans 30 8 7/24 7 blood cultures Dead
3 AMLe C. albicans 7.4 16 5/22 2 blood cultures, histopathology Alive
4 ALL C. parapsilosis 5.8 NA 2/12 7 blood cultures Alive
5 ALL C. parapsilosis 6.7 21 2/3 2 blood cultures, histopathology,
pre- and postmortem culture
Dead
6 ALL C. albicans 10.4 4 4/16 2 blood cultures Alive
7 ALL C. parapsilosis 30 3 3/18 2 blood cultures Alive
8 Aplastic anemia C. tropicalis 30 NA 5/13 17 blood cultures Alive
9 T-cell leukemia C. albicans 30 9 5/12 2 blood cultures, postmortem
cultures
Dead
10 ALL C. glabrata/C. albicans 5.1 3 2/2 2 blood cultures, postmortem
cultures
Dead
a Number of positive urine samples/total number of urine samples.
b ALL, acute lymphatic leukemia.
c NA, no adequately timed samples available.
d NHL, non-Hodgkin’s lymphoma.
e AML, acute myeloid leukemia.
FIG. 1. Peak D/L-arabinitol ratios obtained for the following: patients with
confirmed invasive candidiasis (group A; n 5 10), patients receiving empiric
antifungal treatment (group B; n 5 23), patients not undergoing antifungal
treatment or showing evidence of invasive candidiasis (group C; n 5 67), and
single samples from healthy nonhospitalized children (group D; n 5 56). The
broken line at the value of 4.6 corresponds to the upper cutoff limit.
FIG. 2. D/L-Arabinitol ratios for patient 2 (Table 1) with confirmed invasive
candidiasis. A, the time at which the first blood sample, culture positive for C.
albicans, was drawn. Amphotericin B was given from day 36 to day 62, when the
patient died (horizontal black bar). The broken line at the value of 4.6 corresponds
to the upper cutoff limit.
VOL. 35, 1997 D/L-ARABINITOL IN INVASIVE CANDIDIASIS 637
children showed positive D/L-arabinitol ratios 6, 10, 10, 12, 13,
14, 15, 21, 28, and 31 days, respectively, before empiric antifungal
chemotherapy was instituted, and Fig. 3 presents the
ratios over time for one of these patients. Six patients died;
four of them died during or shortly after the institution of
antifungal chemotherapy.
The remaining 67 patients received no antifungal treatment,
and the mean of their D/L-arabinitol ratios, determined by
using each patient’s peak value, was 3.2 6 1.1 (Fig. 1, group C).
Four patients had peak values above the cutoff limit, but for all
of these patients the values later returned to normal. Four
patients died due to their malignancies, and none of these had
a positive D/L-arabinitol ratio.
On the basis of the results for patients in groups A and C,
the sensitivity of the arabinitol test in diagnosing invasive candidiasis
is 100% and the specificity is 94%, resulting in a positive
predictive value of 71% and a negative predictive value of
100%. If it is assumed that all the empirically treated group B
patients did have invasive candidiasis, positive and negative
predictive values are both 85%.
Forty children were diagnosed with bacteremia. Of these
patients, 19 had at least two blood cultures positive for grampositive
bacteria, 15 patients had blood cultures positive for
gram-negative bacteria, and mixed infections were seen in 6
patients. The mean D/L-arabinitol ratio during bacteremia was
3.1 6 0.9, and one patient had a positive value of 5.2, which,
however, was recorded during a period of empiric antifungal
chemotherapy. One patient died from streptococcal bacteremia
and one died from a mixed infection (these two patients
had previously received empiric antifungal chemotherapy; see
above).
D/L-Arabinitol ratios were significantly higher during neutropenic
periods when antifungal therapy was not given (3.0 6
1.1) than during nonneutropenic periods (2.5 6 0.7) (P ,
0.01). Also, D/L-arabinitol ratios were significantly higher during
periods of bacteremia (3.1 6 0.9) than during nonneutropenic
periods (P , 0.01).
In all, a total of 1,039 blood samples for culture were drawn
during the study period. Forty-six cultures of blood from 10
patients were positive for Candida species, and 237 cultures of
blood from 40 patients were positive for bacteria. The mean
numbers of urine samples per patient delivered for D/L-ara-
TABLE 2. Peak D/L-arabinitol ratios in febrile, neutropenic children with cancer, empirically treated for invasive candidiasis
Patient no. Primary disease
Peak D/Larabinitol
ratio
Time interval (days)
between positive
D/L-arabinitol ratio
and empiric treatment
No. of positive
samples/no. of
all samplesa
Patient
status
Postmortem
culture and
histopathology
result
11 ALLb 3.2 NAc 0/21 Alive
12 AMLd 5.0 14 3/8 Alive
13 Wilms’ tumor 7.2 28 4/9 Alive
14 Non-Hodgkin’s lymphoma 7.7 NA 5/15 Alive
15 Rhabdomyosarcoma 3.7 NA 0/11 Alive
16 ALL 3.6 NA 0/36 Dead Negative
17 ALL 8.2 12 9/29 Alive
18 Megacaryocyteleukemia 4.5 NA 0/2 Alive
19 Malignant Schwannoma 4.6 10 2/18 Dead Negative
20 Rhabdomyosarcoma 2.6 NA 0/12 Dead Negative
21 B-cell lymphoma 3.4 NA 0/17 Alive
22 ALL 4.9 15 2/12 Alive
23 ALL 16.0 NA 2/5 Dead Not done
24 CMLe 3.2 NA 0/17 Alive
25 ALL 5.5 21 2/6 Alive
26 B-cell lymphoma 4.1 NA 0/8 Alive
27 B-cell lymphoma 4.3 NA 0/26 Dead Not done
28 B-cell leukemia 3.6 NA 0/9 Alive
29 AML 9.8 31 8/18 Alive
30 Neuroblastoma 4.9 13 2/5 Alive
31 AML 7.6 6 2/20 Dead Negative
32 AML 4.9 10 4/16 Alive
33 CML 2.5 NA 0/2 Alive
a Number of positive urine samples/total number of urine samples.
b ALL, acute lymphatic leukemia.
c NA, no adequately timed samples available.
d AML, acute myeloid leukemia.
e CML, chronic myeloid leukemia.
FIG. 3. D/L-Arabinitol ratios for patient 17 (Table 2), who was receiving
empiric antifungal therapy. Amphotericin B was given from day 35 to day 50
(horizontal black bar). The broken line at the value of 4.6 corresponds to the
upper cutoff limit.
638 CHRISTENSSON ET AL. J. CLIN. MICROBIOL.
binitol analysis were 15, 14, and 9 for groups A, B, and C,
respectively.
DISCUSSION
This is the first prospective study on the diagnostic value of
monitoring urine D/L-arabinitol ratios in immunocompromised
patients at high risk of invasive candidiasis. All 10 patients with
confirmed invasive candidiasis, based on multiple positive
blood cultures alone (6 patients) or such cultures together with
tissue biopsies or postmortem examination (4 patients),
showed positive D/L-arabinitol ratios (Table 1; Fig. 1). Positive
ratios were also found for 12 of 23 children who had negative
blood cultures and who were being empirically treated with
antifungal chemotherapy (Table 2; Fig. 1) and for 4 of 67
patients not receiving antifungal therapy. However, blood cultures
are usually positive for only about 50% of patients with
proven disseminated candidiasis, even when biphasic media or
lysis-centrifugation is used (1, 14, 19). It is thus tempting to
speculate that 12 of the 23 empirically treated patients with
increased D/L-arabinitol ratios had unconfirmed invasive candidiasis.
This study was not designed to monitor therapy by
using D/L-arabinitol ratios, but the data presented for two patients
(Fig. 2 and 3) indicate that the outcome of antifungal
treatment correlates with changes in D/L-arabinitol ratios.
C. glabrata has not been shown to produce D-arabinitol when
it is grown in vitro (9). The patient with both positive blood
cultures for C. glabrata and increased D/L-arabinitol ratios
probably had a mixed candidal infection, since all postmortem
tissue cultures were positive for C. albicans. Considering subjects
not included in the present study, we have also found
elevated D/L-arabinitol ratios in two nonimmunocompromised
patients with C. glabrata fungemia, but for these two patients
no tissue cultures were performed (unpublished data).
During nonneutropenic periods, children with cancer had
higher D/L-arabinitol ratios (2.5 6 0.7) than healthy nonhospitalized
children (2.0 6 0.6) (P , 0.01). The ratios further
increased (3.0 6 1.1) during neutropenic periods without empiric
antifungal chemotherapy. This could reflect an increased
fungal load in these patients at risk for fungal infection, although
fungal surveillance cultures were not included in our
study, and all patients were given nystatin orally. D/L-Arabinitol
ratios also increased (3.1 6 0.9) during bacteremia, and it has
previously been reported that bacteremia is a risk factor for
fungal infection in patients with hematologic disorders (7). We
do not believe, however, that bacteremia per se gives rise to
increased D/L-arabinitol ratios in urine, since we have found
normal D/L-arabinitol ratios in all of 25 adult nonimmunocompromised
patients with bacteremia whom we studied (unpublished
data).
Perhaps the most striking finding of this study is that D/Larabinitol
ratios often increased days to weeks before invasive
candidiasis was suspected or confirmed through blood cultures
or histopathology. Elevated D/L-arabinitol ratios were found 3
to 31 days (median, 12 days) before a culture-positive blood
sample was first drawn from 7 of 10 children with confirmed
invasive candidiasis and before empiric antifungal treatment
was initiated for 10 of 12 children. The time gain achieved with
D/L-arabinitol ratios was actually even greater considering the
time needed for culture. It has previously been reported that
the outcome for patients with invasive candidiasis depends
largely on the early institution of antifungal therapy (2, 15).
Renal dysfunction influences the absolute levels of D-arabinitol
in both serum and urine (18). The method of using
D/L-arabinitol ratios, which was first suggested by Roboz et al.
(18), has been found to overcome this problem, and, using
negative chemical ionization mass spectrometry, Roboz and
Katz (17) found increased D/L-arabinitol ratios in serum from
15 of 16 patients with confirmed candidiasis. The present study
indicates that monitoring of D/L-arabinitol ratios in the urine of
immunocompromised children with cancer is useful for diagnosing
invasive candidiasis. A considerable advantage of our
method is that the analysis can be performed on a standard gas
chromatograph, which expands the diagnostic applicability of
D/L-arabinitol ratio measurements. It should be emphasized
that the method has not yet been evaluated prospectively with
immunocompromised adults or multitrauma patients with
Candida superinfections due to long-term broad-spectrum antibiotic
treatment. Nevertheless, we have found increased D/Larabinitol
ratios in urine samples from several such patients
(unpublished data).
Another approach to solving the problem of increased concentrations
of D-arabinitol in the serum of patients with renal
dysfunction is to determine the D-arabinitol/creatinine ratio. In
a prospective study, Walsh et al. (22) found elevated D-arabinitol/
creatinine ratios in 74% of cancer patients with verified
candidemia and in 40% of patients with deep-tissue candidiasis
without fungemia. Although they used another method to detect
the in vivo production of D-arabinitol, the results of Walsh
et al. (22) support our unpublished observations that increased
D/L-arabinitol ratios in urine are seen also in adult cancer
patients with invasive candidiasis.
A number of circulating candidal antigens, e.g., the cytoplasmic
Candida enolase antigen, cell wall mannan, and heat-labile
glycoprotein antigens (Cand-Tec), have also been used for
diagnosis (5). Analysis of multiple serial serum samples may
improve the sensitivity of Candida detection, since antigens are
cleared rapidly from serum and/or immune complexes are
formed. Walsh et al. (20) found a diagnostic sensitivity of 75%
and a specificity of 96% in a prospective clinical trial by using
multiple sampling and an assay for Candida enolase in serum.
Studies on the detection of serum mannan have shown sensitivities
from 0 to 65%, as reviewed by de Repentigny (5).
Antibody tests are often negative for immunocompromised
patients, and it has been claimed that such tests are of more
prognostic than diagnostic significance (11). Nevertheless, Deventer
et al. (6) found anti-Candida enolase antibodies in 53%
of a group of immunodeficient patients with invasive candidiasis,
and Navarro et al. (12) reported a sensitivity of 89% when
analyzing antibodies to cell wall-bound and cytoplasmic antigens.
It should be pointed out, however, that neither of the
cited reports presented any data regarding antibody kinetics
during the course of infection. Newer techniques such as Candida
DNA amplification have produced encouraging results in
small studies, and sensitivities of between 60 and 79% for
culture-positive clinical specimens have been achieved (3, 4, 8).
Another approach involves a Limulus amoebocyte lysate assay
for the detection of elevated levels of 1,3-b-D-glucan in plasma,
which was found to have a sensitivity of 90% for patients with
confirmed fungal infection (13). To date, however, neither of
these two methods has been thoroughly evaluated in prospective
clinical studies.
The lack of a “gold standard” for the diagnosis of invasive
candidiasis complicates the evaluation of new diagnostic methods.
However, the results of the present study indicate that
elevated D/L-arabinitol ratios can be detected and used as a
valuable predictor at an early stage of invasive candidiasis in
children with cancer. Therefore, we suggest that, together with
routine blood and tissue cultures, the D/L-arabinitol ratios for
such patients should be monitored routinely, and for patients
with increased ratios, antifungal chemotherapy should be instituted.
VOL. 35, 1997 D/L-ARABINITOL IN INVASIVE CANDIDIASIS 639
ACKNOWLEDGMENTS
This study was supported by grants from the Children’s Cancer
Foundation of Sweden, the Royal Physiographic Society in Lund, and
the Medical Faculty of Lund University.
REFERENCES
No comments:
Post a Comment