features at presentation predict children with acute lymphoblastic leukemia at low risk for tumor...
TRANSCRIPT
Features at Presentation Predict Children With AcuteLymphoblastic Leukemia at Low Risk for TumorLysis Syndrome
Tony H. Truong, MD1
Joseph Beyene, PhD2
Johann Hitzler, MD1,3
Oussama Abla, MD1
Anne Marie Maloney, RN, MSN, ACNP1
Sheila Weitzman, MB, BCh1
Lillian Sung, MD, PhD1,2
1 Division of Hematology/Oncology, the Hospitalfor Sick Children, University of Toronto, Toronto,Ontario, Canada.
2 Program in Child Health Evaluative Sciences,the Hospital for Sick Children, University ofToronto, Toronto, Ontario, Canada.
3 Program in Developmental Biology, the Hospitalfor Sick Children, University of Toronto, Toronto,Ontario, Canada.
BACKGROUND. Tumor lysis syndrome (TLS) is a well-recognized complication of
acute lymphoblastic leukemia (ALL). The ability to predict children at differing
risk of TLS would be an early step toward risk-based approaches. The objectives
of the current study were 1) to describe the prevalence and predictors of TLS in
childhood ALL and 2) to develop a sensitive prediction rule to identify patients at
lower risk of TLS.
METHODS. Health records of children aged �18 years who were diagnosed with
ALL between 1998 and 2004 were reviewed. TLS was defined by the presence of
�2 laboratory abnormalities occurring in the time frame of interest. Predictors of
TLS were determined using univariate and multiple logistic regression analyses.
RESULTS. Among 328 patients, 23% met criteria for TLS. Factors predictive of TLS
were male sex (odds ratio [OR], 1.8; P 5 .041), age �10 years (OR, 4.5; P < .0001),
splenomegaly (OR, 3.3; P < .0001), mediastinal mass (OR, 12.2; P < .0001), T-cell
phenotype (OR, 8.2; P < .0001), central nervous system involvement (OR, 2.8;
P 5 .026), lactate dehydrogenase �2000 U/L (OR, 7.6; P < .0001), and white blood
count (WBC) �20 3 109/L (OR, 4.7; P < .0001). Among variables that were avail-
able at presentation, multiple regression analysis identified age �10 years,
splenomegaly, mediastinal mass, and initial WBC �20 3 109/L as independent
predictors of TLS. When all 4 of those predictors were absent at presentation
(n 5 114 patients), the negative predictive value of developing TLS was 97%, with
a sensitivity of 95%.
CONCLUSIONS. Clinical and laboratory features at the time of presentation identi-
fied a group of children with ALL at low risk for TLS that may benefit from a
risk-stratified approach directed at reduced TLS monitoring and prophylaxis.
Cancer 2007;110:1832–9. � 2007 American Cancer Society.
KEYWORDS: predictive, low risk, tumor lysis syndrome, children, cancer,leukemia.
T umor lysis syndrome (TLS) consists of hyperuricemia, hyperkale-
mia, hyperphosphatemia, and hypocalcemia and may result in
renal failure. It is well recognized that TLS occurs before or after the
initiation of chemotherapy for malignancies, such as childhood
acute lymphoblastic leukemia (ALL) and Burkitt lymphoma.1,2
Standard preventative approaches to minimize this complication
include hyperhydration, urine alkalization, xanthine oxidase inhibi-
tors (allopurinol), and, more recently, recombinant urate oxidase.3–5
Previous studies focused primarily on identifying patients at
increased risk of TLS for the purpose of selecting those who may
benefit from increased laboratory monitoring or urate oxidase ther-
apy.6–9 Risk factors have included presentation with a high initial
L.S. is supported by a Career DevelopmentAward with the Canadian Child Health ClinicianScientist Training Program, a strategic trainingprogram of the Canadian Institutes of HealthResearch.
We thank Carol Winter, Olena Shatokhina, andLoreto Lecce for their assistance with data man-agement and Camille Flynn for her assistancewith data entry.
Address for reprints: Lillian Sung, MD, PhD, Divi-sion of Hematology/Oncology, Hospital for SickChildren, 555 University Avenue, Toronto, Ontario,M5G 1X8, Canada; Fax: (416) 813-5327; E-mail:[email protected]
Received March 5, 2007; revision received June14, 2007; accepted June 18, 2007.
ª 2007 American Cancer SocietyDOI 10.1002/cncr.22990Published online 27 August 2007 in Wiley InterScience (www.interscience.wiley.com).
1832
white blood cell (WBC) count; evidence of large tu-
mor burden (bulky disease, hepatosplenomegaly);
high blood lactate dehydrogenase (LDH)10 or uric
acid levels; pre-existing dehydration, oliguria, or re-
nal failure9,11,12; and malignancies with high chemo-
sensitivity.13,14
However, the majority of children with newly
diagnosed ALL who are treated with standard TLS
prophylactic measures do not experience clinically
significant laboratory abnormalities either before or
shortly after chemotherapy.1 Yet patients without
high-risk features may be subjected to prophylactic
measures and monitoring similar to those used in
patients with high-risk features.
With the long-term aim of a risk-stratified
approach to the prevention of TLS, the objectives of
the current study were 1) to describe the prevalence
and predictors of TLS in childhood ALL and 2) to de-
velop a sensitive prediction rule to identify patients
who are at low risk for TLS.
MATERIALS AND METHODSMedical records from all children aged �18 years
who were diagnosed with ALL between 1998 and
2004 at the Hospital for Sick Children in Toronto,
Canada, were reviewed. We included all patients with
ALL but excluded those with 1) French-American-
British (FAB) classification L3 ALL, 2) patients who
were treated initially at another institution, 3) pa-
tients who were transferred to another institution
within the time frame of interest (from the date of
presentation to the seventh day after initiation of
chemotherapy), and 4) patients who did not receive
initial ALL therapy.
There were 342 children diagnosed with ALL dur-
ing the study period. Fourteen patients were
excluded for the following reasons: 6 patients had
FAB L3 morphology, 5 patients were diagnosed at
another center before arrival at our institution, 1
patient was transferred to another institution during
the time frame of interest, 1 patient received up-
front palliative care because of an unrelated underly-
ing medical condition, and medical records were
missing for 1 child. In total, 328 patients met inclu-
sion criteria and were reviewed. This study was
approved by the Research Ethics Board at the Hospi-
tal for Sick Children.
Outcomes AssessedThe primary outcome was the development of labo-
ratory TLS, which was defined as the occurrence of
any 2 or more of the following 5 laboratory abnorm-
alities during the time frame of interest: hyperkale-
mia (potassium �5.5 mmol/L), hyperphosphatemia
(phosphate �2.26 mmol/L), hypocalcemia (calcium
�2.0 mmol/L), hyperuricemia (uric acid�475 lmol/L),
and azotemia (creatinine �1.5 times the age-defined
upper limit of normal). Our institutionally defined
upper limit of normal of creatinine for both sexes,
by specific age groups, were: ages 7 to 60 days,
66 lmol/L; ages 2 months to 5 years, 44 lmol/L;
ages 6 to 9 years, 62 lmol/L; ages 10 to 13 years,
90 lmol/L; aged >14 years, 100 lmol/L. Laboratory
data were collected during the time frame of interest,
starting from the date of presentation, through to the
day of chemotherapy initiation (Day 0), and for each
of the following 7 days (Day 17), a time frame that
previously defined a higher risk for TLS.3 This labora-
tory definition was modified from previously pub-
lished definitions of TLS3,11,15,16 to be more inclusive
for the purpose of the current study. Because our
focus was on identifying a low-risk subset, we purpo-
sefully wanted to ensure that our definition maximized
sensitivity and minimized false-negative results. We
did not examine clinical TLS (seizures, arrhythmia,
dialysis, or death) as a separate endpoint.
If multiple measurements for a given electrolyte
were obtained on the same day, then the highest
daily value was recorded for serum potassium, phos-
phate, creatinine, and uric acid; whereas the lowest
daily value was recorded for serum calcium. To main-
tain consistency with previous studies, serum cal-
cium was not corrected for hypoalbuminemia.
Other outcomes of interest were measures used
in the prophylaxis or treatment of TLS, namely, the
initial intravenous fluid hydration rate on admission
to hospital; the duration of urine alkalinization; the
administration of allopurinol, urate oxidase, phos-
phate binders (aluminum hydroxide, sevelamer
hydrochloride), antihyperkalemic treatments (sodium
polystyrene sulfonate, insulin, salbutamol), and intra-
venous calcium; and the need for leukopheresis and/
or dialysis. The number of peripheral venipunctures
before the insertion of a central venous line was
recorded as a measure of the impact of TLS labora-
tory monitoring on each patient.
Potential Predictors EvaluatedThe data collected at presentation on potential pre-
dictors of TLS included laboratory features, such as
WBC and LDH, and clinical indicators of bulk dis-
ease, such as the presence of a mediastinal mass on
chest radiographs, hepatomegaly (defined as a palpa-
ble liver �3 cm below the right costal margin), and
splenomegaly (defined as a palpable spleen �2 cm
below the left costal margin) as assessed by the phys-
ical examination on admission. These cutoff values
Predicting Low Risk of TLS in ALL/Truong et al. 1833
were chosen a priori based on what were considered
clinically reasonable limits for deciding whether it
was clear that the liver or spleen was larger than nor-
mal. Other potential predictors examined were cen-
tral nervous system (CNS) status at diagnosis and
renal involvement by leukemia as inferred by renal
enlargement on abdominal imaging studies, when
available. A cutoff value for LDH of �2000 U/L was
chosen, because that level represents an elevation at
least 2 times the upper limit of normal for any age
and sex and has been used in previous publications.3
The degrees of derangement of the initial serum po-
tassium, phosphate, creatinine, uric acid, and cal-
cium levels at presentation were not examined as
potential predictors, because such an analysis would
have been incorrect methodologically, in that those
values would contribute toward the definition of the
outcome (development of TLS).
The following induction chemotherapy protocols
were in use during the study period for patients with
precursor B-cell ALL: from 1998 to 1999, either Pedi-
atric Oncology Group (POG) Protocol 9201, or 9605,
or 9406 or our standard, institutional 3-drug (Proto-
col AB) or 4-drug (Protocol C) induction regimen17;
and, from 2000 to 2004, the POG 9900 protocol se-
ries, which is divided into a 3-drug induction and a
4-drug induction. Patients with T-cell ALL received
either Protocol C (4-drug induction) from 1998 to
1999 or Children’s Oncology Group (COG) Protocol
A5971, after August 2000. Patients with Infant ALL
were treated according to POG 9407 (1998–2004).
Protocols that contained a 3-drug induction during
the first week of chemotherapy included POG 9201
and POG 9605 (daily prednisone; vincristine on Days
0 and 7; and L-asparaginase on Days 1, 4, and 7),
Protocol AB (daily prednisone; vincristine on Days 0
and 7; and L-asparaginase on Days 1, 3, and 5), and
the POG 9900 3-drug induction (daily dexametha-
sone; vincristine on Days 0 and 7; and pegylated as-
paraginase on Day 4, 5, or 6). Protocols that
contained a 4-drug induction in the first week of
chemotherapy included POG 9406 (daily prednisone;
vincristine on Days 0 and 7; L-asparaginase on Days
1, 4, and 7; and daunomycin on Day 7), Protocol C
(daily prednisone; vincristine on Days 0 and 7; dau-
nomycin on Days 0 and 7; and L-asparaginase on
Days 0, 3, 5, and 7), the POG 9900 4-drug induction
(daily prednisone; vincristine on Days 0 and 7; dau-
nomycin on Day 7; and L-asparaginase on Days 2, 4,
and 7), COG A5971 (Regimen B1: daily prednisone;
vincristine on Days 0 and 7; daunomycin on Days 0
and 7; and L-asparaginase on Days 3, 5, and 7), and
POG 9407 (daily prednisone; vincristine on Day 0;
daunomycin on Days 0 and 1; L-asparaginase on
Days 3, 5, and 7; and cyclophosphamide on Days 2
and 3).
Statistical AnalysisBaseline characteristic and demographic data were
described using frequencies and percentages for cat-
egorical variables and means � standard deviation or
interquartile range (IQR) for continuous variables.
Potential predictors of TLS were determined using
univariate logistic regression analyses. However, a
clinically useful prediction rule to identify those at
lower risk of TLS would incorporate factors available
at presentation. Therefore, only this subset of factors
was considered for the multiple logistic regression
model. Factors that were associated with TLS at P < .1
were entered into a forward selection model.
All statistical analyses were performed using the
SAS statistical program (SAS-PC, version 9.1; SAS
Institute Inc., Cary, NC). All tests of significance were
2-sided, and statistical significance was defined as
P < .05.
RESULTSIn total, 328 patients were included, and their demo-
graphics, clinical features, and induction chemother-
apy protocols are shown in Table 1. TLS, which was
defined as the presence of at least 2 laboratory
abnormalities during the time frame of interest,
occurred in 74 of 328 children (22.6%). The single
laboratory abnormality encountered most often was
TABLE 1Demographics of the Study Population
Characteristic
No. of patients (%),
N 5 328
Male sex 206 (62.8)
Acute lymphoblastic leukemia immunophenotype
Precursor B-cell 285 (86.9)
T-cell 38 (11.6)
Other (biphenotypic) 5 (1.5)
CNS-positive disease status 21 (6.4)
Mediastinal mass 26 (7.9)
Ward of admission
Inpatient ward 307 (93.6)
Intensive care unit 21 (6.4)
Induction chemotherapy protocol
3-Drug induction 171 (52.1)
4-Drug induction 136 (41.5)
Children’s Oncology Group protocol A5971* 16 (4.9)
Pediatric Oncology Group Protocol 9407* 5 (1.5)
Prednisone cytoreductive prophase 9 (2.7)
CNS indicates central nervous system.
* Protocols that contained a 4-drug induction regimen.
1834 CANCER October 15, 2007 / Volume 110 / Number 8
hypocalcemia (148 of 328 patients; 45.1%), whereas
the least frequent abnormality was azotemia (14 of
328 patients; 4.3%). The most common laboratory
abnormality pair for TLS was hypocalcemia and
hyperuricemia (40 of 328 patients; 12%), followed by
concurrent abnormalities of calcium and phosphate
(11%) (Table 2). The peak laboratory values of potas-
sium, phosphate, uric acid, and creatinine as well as
the nadir of calcium are shown in Table 3, which
compares those laboratory values between patients
with and without TLS. The day on which these
peaks/nadirs occurred is shown relative to the day of
chemotherapy initiation (Day 0) for both groups.
Factors that were associated with TLS in univari-
ate logistic regression analyses are shown in Table 4.
Mediastinal mass was the strongest predictor of TLS
(odds ratio [OR], 12.2; 95% confidence interval; [95%
CI], 4.9–30.4; P < .0001) and was identified in 58% of
children with T-cell ALL (22 of 38 patients) compared
with 1.4% of children with precursor B-cell ALL (4 of
290 patients). Of these potential predictors, only
those risk factors that were available immediately at
the time of presentation were entered into a multiple
regression analysis. Thus, CNS status and status of
leukemic renal involvement were not entered,
because these potential predictors generally are not
known until several hours or days after presentation.
The initial LDH value was a strong predictor of TLS
(OR, 7.6; P < .0001); however, because only 33 LDH
samples were determined on the day of presentation
to hospital, this variable could not be included in the
multiple regression analysis. Of the remaining 7 vari-
ables (sex, age, WBC, mediastinal mass, hepatome-
galy, splenomegaly, and T-cell immunophenotype), 4
variables were identified in multiple regression as in-
dependent predictors of TLS: age �10 years (adjusted
OR, 5.1; 95% confidence interval, 2.6–10; P < .0001),
splenomegaly (adjusted OR, 2.5; 95% CI, 1.3–4.6;
P 5 .005), mediastinal mass (adjusted OR, 6; 95% CI,
2.2–16.6; P 5 .0005), and initial WBC �20 3 109/L
(adjusted OR, 3.7; 95% CI, 2–7.1; P < .0001). Two-
thirds of all patients (214 of 328 patients; 65%) had 1
TABLE 2Prevalence of Laboratory Abnormalities in Childhood AcuteLymphoblastic Leukemia From the Date of Presentation to 7 DaysAfter Treatment
Laboratory parameter*No. of patients (%),N 5 328
Hypocalcemia 148 (45.1)
Hyperuricemia 54 (16.5)
Hyperphosphatemia 52 (15.9)
Hyperkalemia 33 (10.1)
Azotemia 14 (4.3)
Hypocalcemia and hyperuricemia 40 (12.2)
Hypocalcemia and hyperphosphatemia 35 (10.7)
Hyperphosphatemia and hyperuricemia 26 (7.9)
Hyperkalemia and hypocalcemia 22 (6.7)
Hyperkalemia and hyperuricemia 13 (4)
Hyperkalemia and hyperphosphatemia 12 (3.7)
Hypocalcemia and azotemia 10 (3)
Hyperphosphatemia and azotemia 8 (2.4)
Hyperuricemia and azotemia 5 (1.5)
Hyperkalemia and azotemia 2 (0.6)
* Abnormal laboratory parameters were defined as follows: hyperkalemia, serum potassium
�5.5 mmol/L; hypocalcemia, serum calcium �2.0 mmol/L; hyperphosphatemia, serum phosphate
�2.26 mmol/L; hyperuricemia, serum uric acid �475 lmol/L; azotemia, serum creatinine �1.5 times
the age-defined upper limit of normal.
TABLE 3Peak or Nadir of Laboratory Abnormality and Time Relative to Chemotherapy Initiation: Comparison ofPatients With and Without Tumor Lysis Syndrome
Laboratory parameter
TLS absent, N 5 254 TLS present, N 5 74
PMean 95% CI Mean 95% CI
Potassium peak, mmol/L 4.76 4.72–4.80 5.28 5.15–5.41 <.0001
Phosphate peak, mmol/L 1.92 1.89–1.94 2.49 2.33–2.64 <.0001
Calcium nadir, mmol/L 2.04 2.02–2.06 1.77 1.71–1.84 <.0001
Uric acid peak, lmol/L 288 275.6–300.4 533.5 464.6–602.5 <.0001
Creatinine peak, lmol/L 51.6 49.7–53.4 90.1 65.6–114.8 .003
Mean d to potassium peak* 2.45 2.12–2.79 2.27 1.73–2.81 .6
Mean d to phosphate peak* 0.66 0.42–0.91 1.14 0.79–1.48 .03
Mean d to calcium nadir* 2.48 2.15–2.82 2.2 1.75–2.66 .33
Mean d to uric acid peak* 20.98 21.35 to 20.62 21.58 21.92 to 21.25 .02
Mean d to creatinine peak* 20.13 20.44 to 0.19 0 20.55 to 0.55 .71
TLS indicates tumor lysis syndrome; 95% CI, 95% confidence interval.
* Mean days are expressed relative to chemotherapy initiation (Day 0). A negative value refers to day(s) prior to chemotherapy initiation.
Predicting Low Risk of TLS in ALL/Truong et al. 1835
or more of these 4 independent predictors of TLS at
presentation; and of these, 70 of 214 patients (33%)
developed TLS.
The absence of all 4 predictors of TLS was used
to define a group at low risk of developing TLS (the
low-risk TLS group). Of those who fulfilled low-risk
TLS criteria, 110 of 114 patients did not develop TLS,
resulting in a negative predictive value of 96.5% (95%
CI, 91.3–98.6%) and a sensitivity of 94.6% (95% CI,
87–98%). However, within this low-risk group, 4 of
114 patients (3.5%) also met our definition for TLS.
One of these 4 patients met criteria for TLS, because
this child presented with septic shock related to
streptococcal bacteremia and subsequently devel-
oped renal failure and required dialysis. The remaining
3 patients had only mild perturbations in potassium,
phosphate, and/or calcium that did not require sig-
nificant interventions beyond the prophylactic use of
phosphate lowering agents, increased hydration, and
increased laboratory monitoring.
A further analysis was done to refine our primary
definition of TLS to include only those patients who
had TLS laboratory abnormalities occurring within
any 48-hour time frame. In total, 54 patients (16.5%)
met this stricter definition of TLS. Multiple regression
analysis indicated that the same 4 factors remained
independent predictors of the stricter definition of
TLS: age �10 years (adjusted OR, 3.4; 95% CI, 1.6–
7.2; P 5 .002), splenomegaly (adjusted OR, 2.8; 95%
CI, 1.4–5.6; P 5 .003), mediastinal mass (adjusted OR,
3.7; 95% CI, 1.4–9.7; P 5 .0001), and initial WBC
�20 3 109/L (adjusted OR, 5.1; 95% CI, 2.4–10.8;
P < .0001). Of those who fulfilled the low risk of TLS
criteria, 112 of 114 patients did not develop TLS
according to the more strict definition, resulting in a
slightly improved negative predictive value of 98.2%
(95% CI, 93.8–99.5) and a sensitivity of 96.3% (95%
CI, 87.5–99).
The extremes of laboratory abnormalities and
the day of the extreme value relative to chemother-
apy initiation are shown in Table 5 according to
those at low risk (n 5 114 patients) and those not at
low risk (n 5 214 patients) for TLS. Overall, most
laboratory abnormalities occurred within 3 days after
the initiation of chemotherapy. Those in the low-risk
TLS group had milder laboratory abnormalities com-
pared with the nonlow-risk group. Generally, the lab-
oratory abnormalities occurred later in the low-risk
group compared with those in the nonlow-risk
group.
Hyperkalemia �6.0 mmol/L occurred in 11 of
328 patients (3.4%) during the time frame of interest,
all of whom also met both the conventional and
more strict definitions of TLS. Reassuringly, none of
these patients satisfied our prediction rule criteria for
the low-risk TLS group.
Measures taken to prevent TLS are presented in
Table 6. Leukopheresis was used at diagnosis for an
extremely high initial WBC in 11 patients. Three
patients required renal dialysis; 2 for acute renal fail-
ure secondary to TLS and 1 because of overwhelming
sepsis.
The median number of times blood was drawn
on the first, second, and third full day of hospitaliza-
tion was 3 times (IQR, 2–3 times), 2 times (IQR, 1–3
times), and 2 times (IQR, 2–3 times), respectively.
TABLE 4Predictors of Tumor Lysis Syndrome by Univariate Analysis
Variable
No. of patients (%)
OR 95% CI P*TLS present (N 5 74) TLS absent (N 5 254)
Male sex 54 (73) 152 (59.8) 1.8 1–3.2 .041
Age �10 y 32 (43.2) 37 (14.6) 4.5 2.5–8 <.0001
Splenomegaly 48 (64.9) 91 (35.8) 3.3 1.9–5.7 <.0001
Hepatomegaly 38 (51.4) 95 (37.4) 1.8 1–3 .033
Mediastinal mass 19 (25.7) 7 (2.8) 12.2 4.9–30.4 <.0001
Initial WBC �203109/L 49 (66.2) 75 (29.5) 4.7 2.7–8.1 <.0001
Initial LDH �2000 U/Ly 43 (58.1) 44 (17.3) 7.6 4–14.7 <.0001
T-cell immunophenotype 24 (32.4) 14 (5.5) 8.2 4–17 <.0001
CNS-positive disease 9 (12.2) 12 (4.7) 2.8 1.1–6.9 .026
Renal involvement{ 11 (14.9) 4 (1.6) 10.9 3.4–35.4 <.0001
TLS indicates tumor lysis syndrome; OR, odds ratio; 95% CI, 95% confidence interval; WBC, white blood count; LDH, lactate dehydrogenase; CNS, central nerv-
ous system.
* P value from univariate logistic regression analyses.y Initial LDH was defined as the first level obtained within 3 days of admission (N 5 237).{ Inferred from abdominal ultrasound as enlargement of the kidneys (ultrasound studies were obtained only when clinically indicated; N 5 37).
1836 CANCER October 15, 2007 / Volume 110 / Number 8
Central venous lines were placed an average of
7.2 days (IQR, 4–9 days) from the date of presenta-
tion (data not shown).
DISCUSSIONBy using a very inclusive definition of TLS, we
observed that the prevalence of TLS in children with
ALL before and within 1 week of chemotherapy
initiation was 23%. We used the absence of 4 inde-
pendent risk factors at presentation (age �10 years,
splenomegaly, mediastinal mass, and initial WBC
�20 3 109/L) to develop a prediction rule for identi-
fying those at low risk of TLS. In the absence of all 4
factors, there was a 97% probability that TLS would
not occur; and, in our series, those cases that did
occur (n 5 4) were relatively mild, were identified
early, and did not require significant interventions.
Although many studies have attempted to iden-
tify a group of children at high risk for TLS, we
believe that the current study is important, because
TABLE 6Use of Prophylactic Measures and Interventions for Tumor Lysis Syndrome: Comparison of Patients at LowRisk Versus Patients Not at Low Risk for Tumor Lysis Syndrome
TLS prophylactic or interventional measure
No. of patients (%)
P
At low risk of tls,
N 5 114
Not at low risk of TLS,
N 5 214
Initial intravenous fluid hydration rate, cc/m2/h, mean � SD 94 � 27.8 122.1 � 50.2 <.0001
D of urine alkalinization, mean � SD 5.9 � 2.3 6.2 � 2.6 .41
D of allopurinol, mean � SD 7 � 1.9 7.3 � 2.5 .25
Use of urine alkalinization 112 (98.2) 207 (96.7) .50
Allopurinol 114 (100) 208 (97.2) .1
Urate oxidase 0 (0) 20 (9.3) .002
Aluminum hydroxide 16 (14) 49 (22.9) .08
Sevelamer hydrochloride 0 11 (5.1) .01
Sodium polystyrene sulfonate 0 14 (6.5) .003
Insulin 0 1 (0.5) 1.0
Intravenous calcium 0 7 (3.3) .1
Leukopheresis* 0 11 (5.1) .01
Dialysisy 1 (0.9) 2 (0.9) 1.0
TLS indicates tumor lysis syndrome; SD, standard deviation.
* Patients who required leukopheresis included 5 patients with precursor B-cell acute lymphoblastic leukemia (ALL), 5 patients with T-cell ALL, and 1 patient
with mixed-lineage ALL.y Patients who required dialysis included 2 patients with T-cell ALL and 1 patient with precursor B-cell ALL.
TABLE 5Peak or Nadir of Laboratory Abnormality and Time Relative to Chemotherapy Initiation: Comparison ofPatients at Low Risk Versus Patients Not at Low Risk of Tumor Lysis Syndrome
Laboratory parameter
At low risk of TLS, N 5 114 Not at low risk of TLS, N 5 214
PMean 95% CI Mean 95% CI
Potassium peak, mmol/L 4.74 4.68–4.81 4.95 4.88–5.01 <.0001
Phosphate peak, mmol/L 1.95 1.90–2 2.1 2.03–2.16 .001
Calcium nadir, mmol/L 2.07 2.04–2.10 1.93 1.90–1.96 <.0001
Uric acid peak, lmol/L 267.8 247.0–288.6 383.7 354.4–412.9 <.0001
Creatinine peak, lmol/L 49.1 46.2–52 66.2 57.4–75.1 .0003
Mean d to potassium peak* 2.47 1.91–3.04 2.38 2.06–2.70 .77
Mean d to phosphate peak* 0.39 0.03–0.75 0.97 0.73–1.22 .007
Mean d to calcium nadir* 2.71 2.16–3.26 2.27 1.95–2.58 .17
Mean d to uric acid peak* 20.82 21.45 to 20.18 21.3 21.58 to 20.98 .19
Mean d to creatinine peak* 0.14 20.38 to 0.66 20.22 20.54 to 0.09 .24
TLS indicates tumor lysis syndrome; 95% CI, confidence interval.
* Mean days are expressed relative to chemotherapy initiation (Day 0). A negative value refers to day(s) prior to chemotherapy initiation.
Predicting Low Risk of TLS in ALL/Truong et al. 1837
it is the first attempt to our knowledge to delineate a
low-risk population. Our prediction rule is designed
to be applied at the time of initial hospital presenta-
tion, thus enabling the early identification of a group
of children at low risk for developing TLS who may
be candidates for less intensive TLS monitoring and
prophylactic interventions.
Generally, peripheral venipuncture is the only
means of drawing blood until a central venous cathe-
ter can be inserted, which, at our institution, is
accomplished on average 1 week from the date of
presentation. Reducing the frequency of unnecessary
laboratory monitoring would minimize trauma to
young patients but should be considered only as
long as reduced monitoring would not compromise
the ability to detect TLS early enough to upgrade
prophylactic measures or institute treatments. The
use of urine alkalinization in an attempt to increase
uric acid solubility remains controversial. Titration of
sodium bicarbonate infusions to maintain a urine
pH between 6.5 and 7.5 is a burden to nursing staff,
whereas calcium-phosphate precipitation and subse-
quent nephrocalcinosis is more likely in alkali set-
tings.4,18 Furthermore, over-alkalinization may lead
to precipitation of uric acid precursors, such as hy-
poxanthine or xanthine.5,19,20 Although urine alkali-
zation still is considered the standard of care in
many institutions and treatment protocols,21 the
ability to stop this maneuver in a low-risk group of
children would be beneficial. In addition, although it
is demonstrably effective at lowering uric acid levels
and eliminating the need for alkalinization, urate oxi-
dase is very expensive; and the definition of a low-
risk group would be valuable to help avoid that
unnecessary expense and the rare but real risk of he-
molysis in glucose-6-phosphate dehydrogenase-defi-
cient patients. Our data indicate that clinicians
indeed are identifying correctly those children at low
risk of TLS, because none of these children received
urate oxidase (Table 6), and the intensity of their
prophylaxis/intervention was far less compared with
the intensity for the group that was not at low risk of
TLS. However, our findings may help to standardize
this clinical gestalt and further reduce TLS preventa-
tive measures (such as alkalinization) and limit labo-
ratory monitoring in the low-risk population.
Although our prediction model had a 97% nega-
tive predictive value, further predictive capability by
the addition of the initial LDH value at presentation
is conceivable. In our study, a small minority of
patients had an LDH value obtained on the day of
presentation (n 5 33 patients); therefore, we could
not incorporate this factor into a model that was
intended for use at initial presentation. However,
results from 237 LDH samples collected over the first
3 days after presentation indicated that LDH eleva-
tion is a very significant risk factor for TLS (OR, 7.6).
Thus, future research may be focused on determin-
ing the additive value of this potential predictor of
TLS.
The current study was limited, because the low-
risk factors that we identified were demonstrated in
the setting of standard TLS preventive measures.
Although there is no guarantee that these same chil-
dren would remain at low risk of TLS in the absence
of measures like urine alkalinization, it seems to be a
reasonable assumption, because some institutions
have ceased using this intervention for children with
ALL. Nonetheless, the current study provides a base-
line estimate for TLS in a low-risk cohort that may
be used as a comparison group in future research.
We conclude that a group of children with ALL at
who are at low risk for TLS can be identified at the
time of hospital presentation and may benefit from
reduced intensity of laboratory monitoring and lim-
ited TLS prophylactic measures.
REFERENCES1. Kedar A, Grow W, Neiberger RE. Clinical versus laboratory
tumor lysis syndrome in children with acute leukemia.
Pediatr Hematol Oncol. 1995;12:129–134.
2. Cohen LF, Balow JE, Magrath IT, Poplack DG, Ziegler JL.
Acute tumor lysis syndrome. A review of 37 patients with
Burkitt’s lymphoma. Am J Med. 1980;68:486–491.
3. Cairo MS, Bishop M. Tumour lysis syndrome: new thera-
peutic strategies and classification. Br J Haematol. 2004;
127:3–11.
4. Davidson MB, Thakkar S, Hix JK, Bhandarkar ND, Wong A,
Schreiber MJ. Pathophysiology, clinical consequences, and
treatment of tumor lysis syndrome. Am J Med. 2004;116:
546–554.
5. Jones DP, Mahmoud H, Chesney RW. Tumor lysis syn-
drome: pathogenesis and management. Pediatr Nephrol.
1995;9:206–212.
6. Coiffier B, Mounier N, Bologna S, et al. Efficacy and safety
of rasburicase (recombinant urate oxidase) for the preven-
tion and treatment of hyperuricemia during induction
chemotherapy of aggressive non-Hodgkin’s lymphoma:
results of the GRAAL1 (Groupe d’Etude des Lymphomes de
l’Adulte Trial on Rasburicase Activity in Adult Lymphoma)
study. J Clin Oncol. 2003;21:4402–4406.
7. Goldman SC. Rasburicase: potential role in managing tu-
mor lysis in patients with hematological malignancies.
Expert Rev Anticancer Ther. 2003;3:429–433.
8. Goldman SC, Holcenberg JS, Finklestein JZ, et al. A rando-
mized comparison between rasburicase and allopurinol in
children with lymphoma or leukemia at high risk for tumor
lysis. Blood. 2001;97:2998–3003.
9. Wossmann W, Schrappe M, Meyer U, Zimmermann M,
Reiter A. Incidence of tumor lysis syndrome in children
with advanced stage Burkitt’s lymphoma/leukemia before
and after introduction of prophylactic use of urate oxidase.
Ann Hematol. 2003;82:160–165.
1838 CANCER October 15, 2007 / Volume 110 / Number 8
10. Csako G, Magrath IT, Elin RJ. Serum total and isoenzyme
lactate dehydrogenase activity in American Burkitt’s lym-
phoma patients. Am J Clin Pathol. 1982;78:712–717.
11. Hande KR, Garrow GC. Acute tumor lysis syndrome in
patients with high-grade non-Hodgkin’s lymphoma. Am J
Med. 1993;94:133–139.
12. Kopecna L, Dolezel Z, Osvaldova Z, Starha J, Hrstkova H.
The analysis of the risks for the development of tumour
lysis syndrome in children. Bratisl Lek Listy. 2002;103:206–
209.
13. Rajagopal S, Lipton JH, Messner HA. Corticosteroid in-
duced tumor lysis syndrome in acute lymphoblastic leuke-
mia. Am J Hematol. 1992;41:66–67.
14. Sparano J, Ramirez M, Wiernik PH. Increasing recognition
of corticosteroid-induced tumor lysis syndrome in non-
Hodgkin’s lymphoma. Cancer. 1990;65:1072–1073.
15. Seidemann K, Meyer U, Jansen P, et al. Impaired renal
function and tumor lysis syndrome in pediatric patients
with non-Hodgkin’s lymphoma and B-ALL. Observations
from the BFM-trials. Klin Padiatr. 1998;210:279–284.
16. Bunin NJ, Pui CH. Differing complications of hyperleuko-
cytosis in children with acute lymphoblastic or acute
nonlymphoblastic leukemia. J Clin Oncol. 1985;3:1590–
1595.
17. Al-Kasim FA, Thornley I, Rolland M, et al. Single-centre
experience with allogeneic bone marrow transplantation
for acute lymphoblastic leukaemia in childhood: similar
survival after matched-related and matched-unrelated do-
nor transplants. Br J Haematol. 2002;116:483–490.
18. Brereton HD, Anderson T, Johnson RE, Schein PS. Hyper-
phosphatemia and hypocalcemia in Burkitt lymphoma.
Complications of chemotherapy. Arch Intern Med. 1975;
135:307–309.
19. Stapleton FB, Strother DR, Roy S 3rd, Wyatt RJ, McKay CP,
Murphy SB. Acute renal failure at onset of therapy for
advanced stage Burkitt lymphoma and B cell acute lym-
phoblastic lymphoma. Pediatrics. 1988;82:863–869.
20. Berg C, Tiselius HG. The effect of pH on the risk of calcium
oxalate crystallization in urine. Eur Urol. 1986;12:59–61.
21. Albano EA, Sandler E. Oncological emergencies. In: Altman
AJ, ed. Supportive Care of Children with Cancer: Current
Therapy and Guidelines from the Children’s Oncology
Group, 3rd ed. Baltimore, Md: Johns Hopkins University
Press; 2004:221–242.
Predicting Low Risk of TLS in ALL/Truong et al. 1839