Keywords
Childhood acute myeloid leukemia - drug-resistant sepsis - hematopoietic stem cell
transplantation - relapse - sepsis-related mortality
Introduction
High-dose chemotherapy is the backbone of acute myeloid leukemia (AML) requiring significant
supportive care.[1],[2] Sepsis has been a considerable obstacle in providing optimal care for AML chemotherapy,
and the associated cost adds to the already strained finances of families.[3] Over the last few decades, hospital infection control policies (HIC) have had to
grapple with changing hospital flora and the spectrum of evolving bacterial species,
particularly multidrug-resistant organisms (MDROs). We had documented a sudden increase
in the incidence and isolation of drug-resistant gram-negative bacterial sepsis after
2012. We aimed to analyze the changing spectrum in bacteremia over 18 years and the
impact that emerging drug-resistant organisms have had on the outcome.
Patients and Methods
A retrospective observational study was performed in the department of pediatric hematooncology,
blood and marrow transplantation. We included all children up to 18 years of age diagnosed
to have AML at our center and who received chemotherapy from January 2002 to June
2019. Exclusion criteria included children who proceeded to upfront hematopoietic
stem cell transplantation (HSCT), those with infant AML, those diagnosed to have AML
M3 with t (15;17) or variants, and those diagnosed to have mixed phenotype acute leukemia.
The diagnosis of AML was confirmed through flow cytometry to determine leukemia-associated
immunophenotype. We performed fluorescence in situ hybridization and karyotyping in all diagnostic bone marrow samples. The treatment
was risk-stratified, based on molecular markers.
All children received treatment as per the UKMRC AML 15 protocol, including two cycles
of induction (cytarabine, daunorubicin, and etoposide) followed by two cycles of consolidation
(high-dose cytarabine). We performed a bone marrow assessment at the end of the first
induction for remission status and following the second consolidation for the end
of treatment assessment. We performed a repeat bone marrow assessment at the end of
the second induction for those with minimal residual disease (MRD) positive disease
after induction cycle 1. An evaluation was done through examination for morphological
remission alone up until 2009. From 2009, we performed MRD assessment by flow cytometry
for all children at all time points. As per the protocol, MRD of <0.1% was negative.
We offered HSCT if the child had high-risk cytogenetics, persistently positive MRD,
refractory disease, and relapsed disease.
We continued antifungal, antiviral, and pneumocystis carinii prophylaxis for all children
throughout chemotherapy. Antifungal prophylaxis included fluconazole up until 2012.
Since 2012, we have consistently used micafungin as antifungal prophylaxis. In the
children with proven fungal infection in the previous cycle, we used voriconazole
or amphotericin as prophylaxis. All children received granulocyte-colony stimulating
factor from day 7 after completion of chemotherapy.
Febrile neutropenia was treated with antibiotics as per the HIC policy after taking
adequate samples for cultures. We followed a policy of de-escalation of antibiotics
when the child was afebrile for 48 h, and the blood cultures were sterile. Antibiotic
policy was based on the HIC and the existing antibiogram. A combination of third generation
cephalosporin with beta-lactamase inhibitor was the most frequent first-line antibacterial
agent used. A carbapenem with or without teicoplanin for Gram-positive cover was used
in case of persistent febrile neutropenia. The decision was also based on the presence
or absence of any specific focus for infection and the anticipated duration of neutropenia.
All culture-positive bacteremia was treated with antibiotics based on the drug sensitivity
after consulting the infectious disease specialists.
We performed stool sample screening for the presence of carbapenem resistance through
a multiplex polymerase chain reaction. We introduced colistin early in the children
with resistant pathogens on stool screening. In children with suspected central line-associated
bacteremia, we performed an additional peripheral blood culture. In case of persistent
fever for over 72 h or if there were localizing signs, we performed early imaging.
We analyzed the data on the demography, subclassification of AML, remission status,
rate of relapse, treatment postrelapse, infectious complications, species isolated
and drug sensitivity, rates of fungal infections, and the associated morbidity and
mortality. The data on infections were analyzed pre-and post-2012. A septic event
was the detection of culture-proven bacteremia. Written informed consent was obtained
from the parents of all children, and the institutional ethics review board approved
the study.
Statistical methods
Continuous variables were represented as mean ± standard deviation or median, if they
were normally or nonnormally distributed. Categorical variables were expressed as
percentage. Comparison of normally distributed continuous variables was done by independent
sample t-test and of nonnormally distributed continuous variables by Mann–Whitney
test. Categorical variables were compared by either Chi-square test or Fisher's exact
test based on the number of observations. Kaplan–Meier curve was drawn to assess the
survival pattern with log rank test for comparison of survival pattern. Data entry
was done into MS excel sheet, with validation and analysis by SPSS version 25.0 (Armonk,
NY: IBM Corp.). All P < 0.05 were considered as statistically significant.
Results
We included a total of 48 children with an equal male: female ratio; median age of
8 years. The predominant subtype of AML was M2 in 34 children (70%), followed by M4
in 9 (18%), M1 in 2 (4%), M6, M3 with t (17;17) and myeloid sarcoma stage 4 in one
child each.
Culture-positive bacteremia
Thirty-one (64%) children developed culture-positive bacteremia and sepsis during
chemotherapy. The septic events were predominantly noted during induction chemotherapy
in 28/31 (90%), and during consolidation in 3/31 (10%) children. Of the 31, 25 (80%)
were gram-negative bacilli (GNB) while 6 (20%) were Gram-positive cocci (GPC). Among
the GNB, the most common organism was Klebsiella pneumoniae in 18/25 (72%), Escherichia coli in 5/25 (20%), Pseudomonas aeruginosa in 1/25, and Vibrio cholerae in 1/25. We documented GPC in six children-3/6 Streptococcus viridans, 2/6 coagulase-negative
Staphylococcus aureus, and 1/6 methicillin-sensitive S. aureus. Among the GNB, 4/5 E. coli (80%) and 10/18
(55%) of the K. pneumoniae were carbapenem-resistant. All children with carbapenem-resistant bacteremia received
colistin. The focus for infection was predominantly gut in 26/31 (83%). We had documented
that perianal ulcers preceded GNB bacteremia in 20/31 cases (64%), and this was statistically
significant (P value 0.001).
Pre-and post-2012
Nineteen children underwent treatment before 2012, and 29 children underwent treatment
post-2012. Ten (52.6%) developed culture-proven bacteremia prior to 2012, while 21
(72.4%) developed post 2012. K. pneumoniae has remained the predominant organism throughout. We documented carbapenem-resistance
in only one child before 2012. However, after 2012, we had documented 13 children
with carbapenem-resistance. There was a documented increase in carbapenem-resistance
post 2012, which was statistically significant (14% vs. 67%, P = 0.032).
Fungal infections
We documented fungal infections in 7/48 (14.5%). Proven fungal infection with the
isolation of candida species in the bloodstream was recorded in one child (2%), while
6 (12.5%) had probable fungal infections, localized in the lung.
Interventions to combat severe sepsis
Febrile neutropenia necessitated admission to the pediatric critical care unit (PICU) in 32 (66%) children. Of the 32 admissions, 28 (87%) children were discharged
from PICU when they recovered from septic events. Early granulocyte transfusion administered
in 13/32 (40%) children prevented organ damage. Granulocyte transfusions were indicated
in those with neutropenia and culture positive sepsis with or without features of
early septic shock and where neutropenia was not expected to recover within the next
48 h. Central lines were removed and changed soon in all children demonstrating features
of sepsis before the onset of septic shock. Monitoring and supportive care for mucositis,
Sitz bath for perianal care twice a day reduced bacterial translocation. Meticulous
perianal hygiene involved the use of a barrier cream of lanolin ointment alternating
with “triple cream,” a combination of clotrimazole, mupirocin, and zinc oxide. We
had a unit policy for neutropenic diet, enteral nutrition, including the use of protein
supplements and antibiotic stewardship as per the HIC team.
Outcome and survival
Overall survival in our cohort was 53% with a median follow up of 85 months. The mean
survival time was 95 months (confidence interval [CI] 95% 70.4 months–119 months),
with a plateau in the survival curve after 24 months [Figure 1]. Causes of death are shown in [Table 1]. Overall, the induction mortality due to sepsis was 6.2%. Of the three children
who died of sepsis during induction, two were treated prior to 2012 and one child
post 2012. Among the 17 children who did not develop culture proven bacteremia, 10
(58%) are alive and in remission, while 7 died due to disease relapse.
Figure 1: Kaplan–Meier survival curve analysis demonstrating overall survival of 53%
with mean survival time was 95 months (confidence interval 95% 70.4–119 months), with
a striking plateau in the survival curve after 24 months
Table 1
Causes of death in the cohort
|
Cause of death (n=23/48; 47%)
|
n (percentage of total 48)
|
|
HSCT: Hematopoietic stem cell transplantation
|
|
Disease relapse
|
15 (31)
|
|
Disease progression
|
3 (6.2)
|
|
Induction sepsis
|
3 (6.2)
|
|
Sepsis post-HSCT
|
1 (2)
|
|
Uncontrolled seizures 1-year posttreatment
|
1 (2)
|
The relapse rate in our cohort was 20/48 (41%). HSCT was performed in five children
who relapsed, one matched sibling donor HSCT and two each undergoing matched unrelated
donor and unmanipulated T replete haploidentical transplantation with posttransplant
cyclophosphamide, respectively. Four children are alive and in remission (80%), while
one child died of sepsis before engraftment. HSCT had a statistically significant
survival advantage with mean survival time post-HSCT of 66 months (CI 95% 42–90) as
against 14 months (CI 95% 12.3–15.8) among those not transplanted (P value 0.002) [Figure 2]. All children who relapsed and did not undergo transplant succumbed to progressive
disease.{Figure 2}
Figure 2: Kaplan–Meier survival curve analysis with mean survival time post- hematopoietic
stem cell transplantation of 66 months (confidence interval 95% 42–90) as against
14 months (confidence interval 95% 12.3–15.8) among those not transplanted (P = 0.002)
Discussion
Our study reports a significant increase in carbapenem resistance among GNB post-2012.
However, the induction mortality due to sepsis has remained low and constant over
two decades, with overall survival of 53%. The myelodysplasia-related changes AML-10
trials published their results with overall survival of 57%. Infections, mainly fungal,
were a predominant cause of death.[4]
Several groups over the years have reported GPCs to be the predominant organisms isolated,
up until 2016. In a publication from 2007, where 492 children enrolled in the CCG2961
study were analyzed for infection-related mortality (IRM), fungi and GPCs were most
commonly associated.[3] Similar results were published from Italy in 2010, where 240 children undergoing
AML chemotherapy were analyzed with bacteremia noted in 32%, GPCs being the predominant
organisms isolated.[5] Bochennek et al. published data on infections during chemotherapy with AML-BFM 2004 protocol, where
isolates were predominantly GPCs as against GNBs (240 vs. 90), with an IRM of 5.4%.[6]
There has been, however, a global change in the patterns of MDROs being isolated.
In 2019, a group from Germany reported inferior outcomes in patients colonized with
carbapenem-resistant organisms among those undergoing AML chemotherapy.[7] Similar results have also been reported from India in 2018, where Jaiswal et al. demonstrated poor outcomes in adults undergoing chemotherapy who were colonized
with carbapenem-resistant enterobacteriaceae.[8] In 2016, a multicenter study from Italy found that of the K. pneumoniae isolates among patients with hematological malignancies, 57.9% were carbapenem-resistant
and contributed significantly to mortality.[9] In another study from Italy, among the patients with carbapenem-resistant Klebsiella,
mortality was highest in those with AML.[10] In 2019, a multicenter study from India analyzed over 5000 isolates and found that
E. coli and K. pneumoniae were isolated in 37% and 26%, respectively. Mortality was
higher among those with MDROs.[11]
Fungal infections were not the predominant cause of morbidity or mortality in our
cohort, as noted in 14% of children with only one child diagnosed to have a proven
fungal infection. This is in contrast to several groups worldwide reporting invasive
fungal infections as an essential contributor to infection rates.[3],[4] In 2019, Patel et al. reported invasive mold infections in 28/162 adults undergoing AML chemotherapy.[12] The use of micafungin in prophylaxis and early imaging and intervention has helped
keep the incidence of fungal sepsis low in our cohort.
Hand hygiene, frequent surface cleaning, isolation, and antibiotic stewardship are
essential to contain MDROs.[13],[14] In our cohort, early intervention with the pediatric intensive care team in sepsis-related
events, including granulocyte transfusions, has a positive impact on survival.[15]
Limitations of the study include the retrospective nature, the lack of a control group
and the nonavailability of MRD assessment prior to 2009.
Conclusions
Despite changes in the bacterial isolates over two decades and increase in carbapenem-resistant
organisms, adequate infection control and treatment by a team including pediatric
oncologists, pediatric intensivists, infectious disease specialists, and oncology
nurses helps keep sepsis related mortality down to <10% in children undergoing chemotherapy
for AML, with an overall survival of 53%. The bigger challenge is the relapse rate
which has remained at around 30%, particularly in the first 2 years' post treatment.
HSCT provides a significant survival advantage in relapsed AML. Precise risk stratification
and safer HSCT techniques are the need of the hour for children with AML.
Acknowledgments
We would like to acknowledge the immense support provided by the departments of hematology,
stem cell apheresis, and infectious disease specialists in the care of these children.
