Skip to main content
NCBI home page
Journal of Pediatric Oncology Nursing logoLink to Journal of Pediatric Oncology Nursing
. 2020 Mar 6;37(4):244–254. doi: 10.1177/1043454220909785

The Longitudinal Parallel Process Analysis of Biomarkers of Oxidative Stress, Symptom Clusters, and Cognitive Function in Children With Leukemia

Mary C Hooke 1,2,✉, Daniel Hatch 3, Marilyn J Hockenberry 2,4, Susan Whitman 5, Ida Moore 5, David Montgomery 5, Kari Marano 5, Pauline Mitby 2, Michael E Scheurer 4, Olga Taylor 4, Wei Pan 3
PMCID: PMC7312345  PMID: 32141369

Abstract

Background: During treatment for acute lymphoblastic leukemia (ALL), children report co-occurring symptoms of fatigue, sleep disturbance, pain, nausea, and depression as a symptom cluster. Central nervous system–directed ALL therapies also put children at risk for cognitive impairments. Cancer therapies can cause an increase in oxidative stress, which may contribute to treatment-related symptoms. This study examined the longitudinal relationships between biomarkers of oxidative stress in the cerebrospinal fluid, the Childhood Cancer Symptom Cluster–Leukemia (CCSC-L), and cognition, in children over the first year of ALL treatment. Methods: Glutathione (GSH) biomarkers of oxidative stress were measured in cerebrospinal fluid collected during treatment lumbar punctures. GSH biomarkers, symptoms, and cognitive function of 132 children aged 3 to 18 years were evaluated at four time points during the first year of leukemia treatment. Participants, 7 years and older, completed self-report measures, and parents reported for younger children. Cognitive function measurements for all participants were completed by parents. A longitudinal parallel-process model was used to explore the influence of the initial measurement and the subsequent change over four time points of the GSH biomarkers on the CCSC-L and cognition. Results: GSH biomarkers increased over the four time points indicating decreasing oxidative stress. When GSH biomarkers were higher (less oxidative stress) at the initial measurement, the CCSC-L severity was lower, cognition was better, and cognition improved over the four measurements. Screening children for high levels of oxidative stress would be a foundation for future intervention studies to address symptom distress and cognitive impairments.

Keywords: cognitive functioning, biomarkers, leukemia, symptom cluster

Recent advances in treatment for pediatric cancers have resulted in an improved overall survival rate of 83.4% for all pediatric cancers in children aged 14 years and younger, and 91% specifically for acute lymphoblastic leukemia (ALL), the most common type of childhood cancer (Siegel et al., 2018). These gains in survival are the result of new therapeutic approaches that are, however, often more intensive and can result in distressing, multiple symptoms that frequently interfere with the child’s quality of life and ongoing development. Symptoms rarely occur individually, but usually they co-occur with other symptoms at the same time signified as a “symptom cluster.” Pediatric cancer researchers have identified different symptom clusters by using a single scale (i.e., the Memorial Symptom Assessment Scale; Collins et al., 2000) to measure multiple symptoms (Rodgers, Hooke, et al., 2016). More recently, researchers have selected, a priori, a group of symptoms known to occur in children and adolescents with leukemia and measured each symptom with a specific scale. Patients are then grouped by mild, moderate, and severe symptom levels using latent class growth analysis (Hockenberry et al., 2017). This symptom cluster of fatigue, sleep disturbance, pain, nausea, and depression is referred to as the Childhood Cancer Symptom Cluster–Leukemia (CCSC-L). In this cohort undergoing leukemia treatment, children who had more severe symptoms in the CCSC-L early in therapy had lower quality of life scores early in therapy and at the beginning of maintenance therapy (Rodgers et al., 2019).

In addition to symptom clusters, many children receiving cancer therapy experience changes in cognitive function. Cognitive function is a complex process that has multiple dimensions including visual-spatial skills, psychomotor ability, processing speed, learning, language, and memory (Moore et al., 2013). Children with leukemia receive intrathecal, intravenous, and oral methotrexate (MTX) as part of their treatment. Side effects of MTX can include neuroanatomical changes as well as difficulties in cognition and learning (Moore et al., 2013). Cognitive outcomes may also be affected by leukemia symptoms experienced by the child. A recent study showed that children with high composite scores on the CCSC-L, indicating severe symptoms early in leukemia treatment, had a significant decrease in their cognitive function (Hooke et al., 2018). Additionally, if their symptom cluster worsened during the first year of treatment, their cognitive function also declined over time (Hooke et al., 2018).

Our understanding of the biologic foundation for symptom clusters is in its early stages. Biomarkers of oxidative stress is an area that merits exploration. Oxidation is a normal biologic event (Da Poian et al., 2010). The process of oxidation involves the removal of electrons from a molecule that will then create a free radical. These free radicals are highly reactive molecules that can break and form new chemical bonds within their immediate environment. One example of the adverse effects of these newly formed bonds is the damage done to DNA and/or RNA during the normal cell cycle (Busu et al., 2013). To prevent constant damage, the reduction of these free radicals is critical to maintain homeostasis within the cell. The body’s natural defense includes compounds with both reduction and oxidation capacities. When this inherent balance of oxidation and reduction is altered, an increase in free radicals results in oxidative stress, which can lead to a wide variety of systemic problems, including neurological disorders, diabetes, cardiovascular diseases, cancer, and shortening one’s life span (Betteridge, 2000; Buettnera et al., 2013). Cancer therapies often disrupt the balance of oxidation and reduction causing an increase in oxidative stress. Oxidative stress is one underlying mechanism thought to be responsible for treatment-related symptoms and adverse side effects (Moore et al., 2018).

The glutathione (GSH) redox reaction is a critical element to decrease oxidative stress in its surrounding environment. In the reduced state, two GSH molecules donate one hydrogen molecule via glutathione peroxidase to neutralize a peroxide (hydrogen peroxide/reactive oxygen species). This reaction forms water or a neutral radical hydroxide molecule and oxidized glutathione (GSSG). The GSSG can be reduced back to 2 GSH molecules using glutathione reductase and reduced nicotinamide adenine dinucleotide phosphate (NADPH; Margaill et al., 2005; Meister & Anderson, 1983; Schafer & Buettner, 2001). Therefore, replenishing the redox cycle is one of the major pathways available to ameliorate oxidative stress within the cellular environment (Figure 1). This ongoing glutathione redox reaction can be used as an indicator of the severity of oxidative stress by measuring the levels of GSH (reduced) and GSSG (oxidized). More specifically, the ratio of reduced glutathione to oxidized glutathione (GSH/GSSG) indicates the level of oxidative stress. A decrease in oxidative stress is indicated by a higher level of reduced GSH and lower oxidized GSSG, while an increase in oxidized GSSG and low amounts of reduced GSH is associated with an increase in oxidative stress. MTX can inhibit NADPH-hydrogenases and NADPH-malic enzyme to decrease the availability of NADPH within the GSH redox cycle (Jahovic et al., 2003). GSH levels become lower and an oxidative stress environment is induced due to the lack of electrons available from GSH to neutralize free radicals. GSSG levels will remain high due to disruption of the redox reaction.

Figure 1.

Figure 1.

Open in a new tab

The glutathione redox reaction.

Note. The glutathione redox reaction is a critical element to decrease oxidative stress in its surrounding environment. In the reduced state, it takes two glutathione (GSH) molecules to donate a hydrogen molecule via glutathione peroxidase to neutralize a peroxide (hydrogen peroxide/reactive oxygen species), thus forming water or a neutral ROH molecule and oxidized glutathione (GSSG). The GSSG can be reduced back to 2 GSH molecules using glutathione reductase and reduced nicotinamide adenine dinucleotide phosphate (NADPH), therefore creating a replenishing redox cycle used to deplete oxidative stress within an environment.

During a routine intrathecal administration of MTX via a lumbar puncture, the cerebrospinal fluid (CSF) is first removed so that it can be analyzed for leukemia cells, and then the intrathecal chemotherapy is administered. Researchers have examined GSH levels in the CSF collected during these routine procedures and studied their relationship to side effects from leukemia treatment. Rodgers, Sanborn, et al. (2016) evaluated GSH/GSSG ratios and fatigue levels at three time points during the first year of intensive ALL treatment. GSH/GSSG ratios were lower than normal levels, and low ratios found earlier in treatment were correlated with higher levels of fatigue in the early phases of treatment. In a recent study, Moore et al. (2018) measured GSH and GSSG levels over multiple time points throughout ALL therapy. They found that both the GSH and the GSSG levels had a significant increase that showed the body’s oxidative defense; the children’s low GSH/GSSG ratio was a result of the oxidized extracellular environment.

To explore the longitudinal relationships of GSH and the GSH/GSSG ratio with the CCSC-L experienced by children undergoing ALL treatment, we used an advanced approach, the longitudinal parallel-process (LPP) modeling (Cheong et al., 2003), to data analysis. When applying the LPP model, the relationship between two variables (e.g., the symptom cluster and the biologic markers) is examined at the same time over the trajectory of treatment (Sousa et al., 2014). These longitudinal relationships can be represented by the estimated initial levels of the variables (i.e., the symptom cluster or biologic marker) and their growth trajectories over time (i.e., the slope of the change in each variable). Using this flexible LPP approach accommodates for modeling the trajectory of a symptom cluster and the biomarker that is likely to change concurrently over time during treatment. The purpose of this study was to examine the longitudinal relationship of the GSH and the GSH/GSSG ratio in the CSF, with the CCSC-L, and cognition, in children during the first year of treatment for ALL (see Figure 2).

Figure 2.

Figure 2.

Open in a new tab

Longitudinal model of GSH and GSH/GSSG ratios on CCSC-L and cognition.

Note. GSH = reduced glutathione; GSSG = oxidized glutathione; CCSC-L = Childhood Cancer Symptom Cluster–Leukemia.

Methods

Design

In this longitudinal study, symptoms were measured by child and parent report four times during the first year of ALL treatment. During each time point, when children were undergoing a lumbar puncture to receive intrathecal chemotherapy, they were placed in a lateral position, and an additional 2 milliliters of spinal fluid was drawn after the initial discard of spinal fluid according to the institution’s protocol. Additionally, children had a baseline lumbar puncture during their diagnostic work-up for the leukemia diagnosis. The four time points are summarized in Table 1. Measurements 3 and 4 occurred within a month of each time point.

Table 1.

Study Time Points.

Time 0 Time 1 Time 2 Time 3 Time 4
Diagnostic lumbar puncture pretreatmenta Consolidation 4 months
postinduction		 8 months
postinduction		 12 months
postinduction
Open in a new tab
a

Biomarker measurement from the cerebrospinal fluid collected at diagnosis was controlled for as a covariate.

Setting and Sample

The study was implemented at four pediatric cancer treatment centers in the United States. Inclusion criteria included the following: Children and adolescents newly diagnosed and receiving ALL chemotherapy, aged 3 to 18 years, without a developmental cognitive disability prediagnosis, and fluent in English or Spanish. The determination of a prediagnosis developmental cognitive disability was determined through review of the electronic medical record and included diagnoses such as autism and Down syndrome. The institutional review board at each site approved the study. Parents were first approached to consent to saving and storing their child’s diagnostic lumbar puncture specimen, which is performed as part of the diagnostic process of leukemia. Before the end of induction, parents and children were invited to participate in the study. As part of the consent process, participants were reminded that their banked specimen from the diagnostic lumbar puncture would be used in the study. Parents of children aged 3 to 17 years provided written consent, and children aged 7 to 17 years provided assent for study participation. Adolescents aged 18 years provided their own consent. The clinicaltrials.gov identifier was NCT01708421 .

Leukemia Treatment

Study participants were undergoing treatment on or off study according to Children’s Oncology Group’s protocols for pediatric ALL. The backbone of this therapy includes 4 weeks of induction chemotherapy during which participants received vincristine, corticosteroid, and pegaspargase, with high-risk and very-high-risk patients receiving an anthracycline (National Cancer Institute [NCI], 2019). After achieving remission, patients receive consolidation/intensification therapy with intrathecal MTX given every 3 months (NCI, 2019). Consolidation is administered after induction and includes cyclophosphamide, low-dose cytarabine, and mercaptopurine. The interim maintenance phase of treatment includes either multiple doses of high-dose MTX (i.e., 5 g/m2) with a leucovorin rescue or escalating doses of MTX (starting dose 100 mg/m2) without a leucovorin rescue. Delayed intensification includes chemotherapy similar to that used during the induction and initial consolidation phases (NCI, 2019). Children with higher-risk leukemia receive additional doses of vincristine and pegaspargase; interim maintenance and delayed intensification phases may also repeat (NCI, 2019). Maintenance therapy begins after these more intensive phases of treatment (NCI, 2019). Treatment schedules for study participants varied dependent on the child’s leukemia risk group, side effects, and blood count recovery

Measurements

Child and parent report measurements on symptoms and cognition were completed on a secure tablet computer with each individual questionnaire appearing separately on the screen. Respondents reported on the symptoms as experienced in the previous week. All instruments had published reliability and validity. Instruments were translated into Spanish and then back translated to assure quality and accuracy. During an outpatient oncology clinic visit, children aged 7 years and older reported on their own symptoms, while parents completed proxy reports for children aged 6 years and younger. Parents completed the cognition measurement for children of all ages.

Fatigue

Children aged 7 to 12 years completed the Childhood Fatigue Scale (CFS) in which they were asked 10 items related to how much they were bothered by fatigue (Hinds et al., 2010). Adolescents used the 13-item Adolescent Fatigue Scale (AFS; Mandrell et al., 2011), while parents reported for their children aged 3 to 6 years using the Parent Fatigue Scale (PFS) that included 17 items (Hockenberry et al., 2003). Cronbach’s α for the fatigue scales ranged from .67 to .95 (Hinds et al., 2010; Hockenberry et al., 2003; Mandrell et al., 2011). Scores of the three age-groups were then combined for analysis using a T score of the total fatigue score. The fatigue scores could range from 20 to 80 with a higher score indicating more severe fatigue.

Sleep Disturbance

Adolescents aged 13 to 18 years completed the self-report Adolescent Sleep Wake Scale (ASWS; Storfer-Isser et al., 2013), while the Child Sleep Wake Scale (CSWS) was used in children aged 3 to 12 years (LeBourgeois & Harsh, 2016). Both scales include five subscales related to going to bed, going to sleep, staying asleep, returning to sleep, and awakening. Scoring was calculated by averaging the subscale scores, which ranged from 1 to 6. Cronbach’s α for the two scales ranged between .84 and .91 (LeBourgeois & Harsh, 2016; Storfer-Isser et al., 2013). To be consistent with the other symptom scoring systems, reverse scoring was performed so that a higher score would indicate more severe sleep disturbance.

Pain

The Wong-Baker Faces Scale is a 6-point visual analogue scale (VAS; 0, 2, 4, 6, 8, and 10) that measures pain severity with a higher score indicating worse pain (Wong & Baker, 1988) with Cronbach’s α equaling .93 (Stinson et al., 2006).

Nausea

A VAS with a metric of 0 to 100 was used for rating nausea severity. The scale was pictured as a thermometer with a statement of the highest and lowest (i.e., no nausea) at each end (Scott & Huskisson, 1979). VASs have been used to measure functioning in children with cancer (Sherman et al., 2006; Cronbach’s α .72-.84) and nausea severity in adults, with a Spearman rank correlation of .90 (Meek et al., 2009.

Depression

The Child Depression Inventory (CDI-2) has established norms and includes 27 items with Cronbach’s α of .91; each item has three potential responses that indicate no symptom, mild symptom, or definite symptom (Kovacs, 2010). A T score of total score was calculated with a range of 20 to 80; a higher score indicated a more severe depressive symptom.

Childhood Cancer Symptom Cluster

At each time point, the measurements of the five symptoms (fatigue, sleep disturbance, pain, nausea, and depression) were shown to be highly correlated. Because of this, these symptoms were clustered and are defined as the Childhood Cancer Symptom Cluster–Leukemia (Hockenberry et al., 2017). This was done using principal component analysis, with a one-factor solution. Factor loadings ranging from .46 to .88 were obtained, and a minimum of 51% of the variance was explained by the factor at each time point. From this, standardized factor scores with a mean of 0 and a standard deviation of 1 were obtained.

Cognition

Parents completed the 32- item Parent-Perceived Child Cognitive Function (pedsPCF) scale, which measures the level of parental concern during the previous week about their child’s memory and thought process (Lai et al., 2011). The instrument has been tested in children undergoing cancer treatment as well (Lai et al., 2014). Parents rated the frequency and intensity of each item on a 0- to 5-point scale with results identifying the risk of problems with attention, social behaviors, and thought. Cronbach’s α for the scale was .99 (Lai et al., 2011). Scores on the PedsPCF in children with cancer have been shown to correlate significantly with scores from the CogState computerized cognitive test; additionally, the scale differentiates between children with cancer who have a brain tumor and those who do not (Lai et al., 2014). Scores were converted to a T score with a range of 20 to 80; a higher score indicates better cognitive function. Children and adolescents were undergoing a lumbar puncture for intrathecal therapy on the day of study measurements, which included medication(s) to manage the pain during the procedure. Because pain medications can affect response time on a computerized cognitive measurement, a valid parent report instrument was selected for the study.

Glutathione

The concentrations of GSH, GSSG, and the ratio GSH/GSSG were determined for the participants using a GSH assay from Promega (GSH Glo cat#V6911). We used a neutral thiol-free 0.05 mM solution of TCEP (tris(2-carboxyethly) phosphine hydrochloride) as the reducing agent (Fisher cat #77720) in this assay. The detailed methodology was described in a previous article (Moore et al., 2018). The results reported here are from patient samples that had not been previously thawed so as to avoid any freeze thaw effects that could have an impact on assay accuracy. We believe that these results best represent the most accurate biologic effects of GSH and presented symptoms in the study participants.

Statistical Analysis

Descriptive statistics were computed for background variables, including gender, age-group, race, and level of leukemia risk. Descriptive statistics were also computed for longitudinal variables, including GSH and the GSH/GSSG ratio, the five symptoms in CCSC-L, and cognition, at Times 1 through 4. Time point 1 was used as the initial time point for study measurements.

To test our hypothesis that participants with concentrations of increased GSH biomarkers had lower severity of leukemia symptoms, we used LPP (Cheong et al., 2003), a two-step modeling process. LPP allows one to examine longitudinal relationships between GSH biomarkers and the CCSC-L from Times 1 to 4 through estimated growth parameters. These include the initial level of the longitudinal variables and the slope, which characterizes rate of change. Specifically, in the first step, univariate growth models with mixed modeling were used to estimate the initial level and slope for each longitudinal variable separately. Assessing relationships by first conducting this step is particularly advantageous in that it also allows one to assess longitudinal relationships among variables with different time frames and different patterns of intermittent missing data. In this step, we were also able to control for covariates (gender, age-group, race, ethnicity, level of leukemia risk, and baseline GSH [time 0] and GSH/GSSG ratio [time 0]), so that they did not need to be controlled for again in the next step. These models were conducted using Proc Mixed in SAS version 9.4 (SAS Institute Inc., 2011).

In the second step, we used the structural equation modeling (SEM) framework to estimate relationships between initial levels and slopes of the GSH biomarkers, the CCSC-L, and cognition measurements. Because the sampling distribution of the estimates from the mixed models in the first step cannot be assumed to be normal, the bootstrap resampling method was used to obtain bias-corrected confidence intervals. SEM model fit was evaluated using chi-square of the estimated model (χ2), goodness of fit index (GFI), normed fit index (NFI), incremental fit index (IFI), relative fit index (RFI), comparative fit index (CFI), and root mean square error of approximation (RMSEA). A nonsignificant chi-square value will indicate a good fit, as will GFI, NFI, IFI, RFI, and CFI larger than .90 and RMSEA lower than .06 (Hu & Bentler, 1999). Based on the conceptual model (Figure 2), an initial SEM model was preliminarily conducted to fit the data. Then, modification indices produced by the SEM program were used, along with theoretical justification, to optimize model fit for our final SEM model. SEM models were conducted using IBM SPSS Amos (Arbuckle, 2010).

Results

Descriptive statistics for demographics are presented in Table 2. Participants in this study were fairly evenly distributed between females and males. Young children aged less than 7 years made up the largest age-group followed by school age children and then adolescents. The sample had a large percentage of children (48.48%) of Hispanic ethnicity with two of the four study sites in the Southwest. The largest proportion of participants was in the very-high-risk leukemia group (34.85%), followed by average/standard risk, high-risk, and low-risk patients. Among the very-high-risk patients (n = 46), 63% were of Hispanic ethnicity (n = 29).

Table 2.

Demographic Characteristics (N = 132).

Characteristic n (%)
Gender
 Male 70 (53.03)
 Female 62 (46.97)
Age-group
 Young child (3-6 years old) 56 (42.42)
 Child (7-12 years old) 49 (37.12)
 Adolescent (13-18 years old) 27 (20.45)
Race/ethnicity
 Hispanics 64 (48.48)
 Non-Hispanic White 44 (33.33)
 Non-Hispanic Black 11 (8.33)
 Non-Hispanic other 13 (9.85)
Leukemia risk group
 Low 12 (9.09)
 Average/standard 40 (30.30)
 High 34 (25.76)
 Very high 46 (34.85)
Open in a new tab

CCSC-L measurements over four time points are seen in Table 3. Fatigue decreased over the trajectory of study measurements during the first year of treatment. Pain was slightly higher at Time 1 and lower at Times 2, 3, and 4. Nausea generally decreased over time with a spike higher at Time 3. Scores on depression, sleep, and cognition were fairly stable over the 12 month study time period. As indicated by Table 4 and Figure 3, GSH levels started lower at Time 1, increased over the first three time points, and then decreased slightly at Time 4. The GSH/GSSG ratio also increased over the four measurements.

Table 3.

Individual Symptom Levels Over Time (N = 132).

Symptom Time 1: Consolidation
M (SD)
n = 130		 Time 2:
4 months
M (SD)
n = 125		 Time 3:
8 months
M (SD)
n = 107		 Time 4:Maintenance
M (SD)
n = 99
Fatigue 53.53 (10.37) 50.55 (10.21) 48.23 (8.69) 46.97 (7.84)
Sleep 2.65 (0.72) 2.81 (0.78) 2.78 (0.81) 2.66 (0.79)
Pain 2.69 (2.39) 1.59 (2.39) 1.26 (1.79) 1.39 (1.95)
Nausea 18.39 (23.43) 15.74 (23.30) 18.58 (24.23) 12.33 (21.11)
Depression 51.18 (9.86) 51.57 (11.53) 49.22 (10.15) 47.83 (8.37)
Cognition 51.63 (8.76) 49.55 (10.46) 47.89 (10.10) 48.61 (11.00)
Open in a new tab

Table 4.

Oxidative Stress Biomarker Levels Over Time (N = 132).

Biomarker Time 1
Consolidation
M (SD)
n = 130		 Time 2
4 months
M (SD)
n = 125		 Time 3
8 months
M (SD)
n = 107		 Time 4 Maintenance
M (SD)
n = 99
GSH 229.40 (156.0) 288.07 (164.61) 303.68 (157.69) 288.68 (135.97)
GSH/GSSG 0.79 (0.48) 1.02 (0.48) 1.10 (0.46) 1.14 (0.58)
Open in a new tab

Note. GSH = reduced glutathione; GSSG = glutathione disulfide.

Figure 3.

Figure 3.

Open in a new tab

Mean GSH levels at the four study time points.

Note. GSH = reduced glutathione; CI = confidence interval.

A preliminary analysis of the initial SEM model indicated that fit of the model could be improved by adding correlations among variables in the model, or among their error terms as appropriate, to achieve our final SEM model. Results of this final SEM model are presented in Figure 4. Fit for this final SEM model was found to be good: χ2(4) = 2.78, p = .60; GFI = 1.0; NFI = 1.0; IFI = 1.0; RFI = .99; CFI = 1.0; and RMSEA < .01. This final model indicated that when initial levels of GSH were higher at Time 1, initial levels of the CCSC-L were significantly lower (β = −0.25, p = .02). Higher levels of GSH at the initial time point (Time 1) were also associated with better initial cognition (β = 0.29, p = .007) and an increase in cognition over time (β = 0.29, p = .007). All other associations between GSH biomarkers (change in GSH over time, GSH/GSSG ratio at initial measurement and over time) and the CCSC-L and cognitive measurements were nonsignificant at the .05 level.

Figure 4.

Figure 4.

Open in a new tab

The final growth model of the longitudinal relationships between GSH biomarkers and childhood leukemia symptoms. Standardized estimates are reported.

†p < .10. *p < .05. **p < .01.

Discussion

Study results demonstrated that the GSH biomarkers had a significant relationship with the CCSC-L and cognition at the first time point and with cognition over time. This is the first study, to our knowledge, to demonstrate the influence of GSH biomarkers on a symptom cluster and on cognition over the trajectory of leukemia treatment.

Relationships Between Glutathione Biomarkers, CCSC-L, and Cognition

When GSH level at the initial measurement was higher, meaning oxidative stress was lower, there was significantly less symptom distress at the first measurement as measured by the CCSC-L. This significant relationship enhances our understanding of how oxidative stress may contribute to the child or adolescent’s symptom experience. When GSH was higher at the initial measurement (less oxidative stress), there was better cognition at the initial time point and a greater increase in cognition over the four study measurements. This indicates that the ability to defend against free radicals may be protective of cognitive function; patients with less oxidative stress may have a greater concentration of GSH to pull from during treatment, which would account for less symptoms and cognitive decline over time.

Individual Symptoms

The measurement of the symptoms of fatigue, depression, and cognition were reported as a T score, a standardized score with a mean of 50 and a standard deviation of 10. Participants scored within 1 standard deviation on these scales with group means decreasing over the four time points for fatigue and depression. Group means for cognition decreased indicating worsening cognition although overall mean scores remained within the first standard deviation. Scores on nausea decreased at Time 2 but then increased at Time 3; the change in mean nausea scores may be attributed to some participants receiving delayed intensification at the time, which is more emetogenic. Sleep scores remained stable and were comparable with published norms from children in a community control group (LeBourgeois & Harsh, 2016).

Glutathione Biomarkers

Levels of reduced GSH increased over the trajectory of the study, which showed that there was a decrease in oxidative stress measured in the CSF. The ratio of GSH/GSSG also increased over the study measurements, as reflected in a higher level of reduced GSH and lower oxidized GSSG; these changes indicate a lowering of oxidative stress. This demonstrates the body’s effective defense mechanism in managing free radicals. The levels and increase in biomarkers were greater than those reported by Rodgers, Sanborn, et al. (2016) and Moore et al. (2018), during the first year of leukemia treatment. The differences in findings may be due to the processing of the CSF specimens; in our study, only specimens that had not been previously thawed were included in the analysis so to avoid any potential freeze thaw effects on the testing accuracy. The 2 milliliter volume of CSF that was collected during the child’s usual clinical care was small and consistent with previous pediatric ALL CSF biomarker research studies (Cheung et al., 2019; Ӧsterlundh et al., 2008). The volume was constant across participants and within each participant across sample collection points. Therefore, we would expect the variability on biomarker concentration related to collection procedure or volume to be small.

In the study sample, almost half of the study participants were of Hispanic ethnicity, which could be a reflection of two of Southwestern United States study sites. Young children were the largest of the three age-groups, which is consistent with the peak age of ALL diagnosis being 2 to 5 years (Rabin et al., 2016). The very-high-risk treatment group (n = 46) was 34.85% of the total study sample; this is a larger portion of the four leukemia risk groups than national estimates of 24% patients who are very-high risk (NCI, 2019). This may be related to the large portion of patients with Hispanic ethnicity (48.8%). Patients with Spanish surnames and those of Hispanic ethnicity are more likely to have adverse prognostic features at diagnosis (Pollock et al., 2000) including genetic variables (Lim et al., 2014).

This study focused on one diagnostic group of children with cancer, those undergoing the first year of ALL treatment. The study population included a wide range of ages from 3 to 18 years, and symptoms and cognition may be experienced and exhibited differently at different developmental stages. Additionally, study participants included four risk levels of children with leukemia; although they received the same backbone of therapy, treatment intensity varied by risk level and treatment arms. In our previously published analysis, however, we found that symptom trajectories were not influenced by age-group, gender, or leukemia risk group; Hispanic participants did experience more symptoms over time (Hockenberry et al., 2017). Future research may focus on one developmental group and/or one risk level within a diagnostic group. However, as is necessary with many pediatric oncology studies, a multisite approach is needed to recruit enough participants with similar characteristics.

Child and parent report measurements at the time of diagnosis would strengthen the analysis of symptoms over the trajectory of treatment. The first days and weeks of treatment are stressful for families, however, and the consenting process focuses on ensuring the parents’ and patient’s understanding of treatment itself as well as enrollment in any treatment clinical trials that are available. Therefore, the researchers decided to approach patients and parents for assent/consent after a month period of adaptation and adjustment to the new diagnosis.

When measuring cognitive function in children with cancer, computerized cognitive assessment tools such as Cogstate or the NIH Toolbox® cognitive measures provide a precise quantitative measure with normative standards. For this study, however, measurements occurred during clinic visits that included lumbar punctures with pain medication(s) that can temporarily affect cognitive performance, and so a validated parent report measurement was utilized to avoid the influence of procedure side effects and decrease measurement burden to participants.

Nurses are often the first to assess children and adolescents for distressing symptoms related to their cancer and treatment. With the recognition that symptoms rarely occur as single events, patients and families should be asked about what other symptoms are co-occurring (clustering) and what are the most distressing and disruptive to them so that interventions can be initiated. As precision medicine moves forward, nurses will need to advance their knowledge of biomarkers that influence responsiveness to cancer treatment but also biomarkers that influence symptoms that may occur with that treatment.

Our understanding of the biologic foundation for symptom clusters is in its early stages. Biomarkers related to inflammation have been identified in adults experiencing symptom clusters during cancer treatment (Kelly et al., 2016), but little is known in the pediatric oncology population. This is the first longitudinal study that identifies GSH biomarkers as a biomechanism associated with a specific symptom cluster in a pediatric cancer population. This first step is identifying children with high levels of oxidative stress who are at risk for increased symptom distress and poorer cognitive outcomes, and it provides a foundation for future testing of tailored, personalized intervention to ameliorate distressing side effects.

Author Biographies

Mary C. Hooke, PhD, APRN, PCNS, CPON, FAAN, is an associate professor at the University of Minnesota’s School of Nursing. She has a clinical appointment as an advanced practice nurse in the Cancer & Blood Disorders program at Children’s Minnesota. Her research interests include supportive care and symptom management for children and adolescents with cancer.

Daniel Hatch, PhD, is a biostatistician III at Duke University School of Nursing. His research interests include latent variable modeling and longitudinal data analysis.

Marilyn J. Hockenberry, PhD, RN, PPCNP, FAAN, is a professor of Pediatrics at Baylor College of Medicine and professor Emerita at Duke University School of Nursing. She is currently the director of Global HOPE Nursing, training nurses in sub-Saharan Africa in the care of children with cancer. Her long term research interests involve symptom assessment and management during cancer treatment.

Susan Whitman, MS, is a researcher, scientist III for the College of Nursing at the University of Arizona. She performs scientific experimentations to help support research interests for the College of Nursing.

Ida Moore, PhD, RN, FAAN, is the Anne Furrow professor and dean, College of Nursing at the University of Arizona. Her research focus is cognitive outcomes of CNS-directed treatment and biomarkers of neurological injury.

David Montgomery, PhD, is a research professor at The University of Arizona College of Nursing and a Research Scientist at the Southern Arizona VA Health Care Center. His recent focus is on ALL and the mechanisms of CNS injury caused by intrathecal methotrexate treatment in children and adolescents with ALL.

Kari Marano, MPH, has been a research specialist at the University of Arizona for 10 years.

Pauline Mitby, MPH, is the clinical research manager for the Cancer & Blood Disorders Program at Children’s Minnesota. Her research interests include supportive care and long-term survivorship.

Michael E. Scheurer, PhD, MPH, is a professor of Pediatrics at Baylor College of Medicine. He also directs the Center for Epidemiology and Population Health. His research interests include the epidemiology of cancer in children and adolescents and understanding the factors associated with adverse treatment outcomes.

Olga Taylor, MPH, is a clinical research manager in the Center for Epidemiology and Population Health in the Department of Pediatrics at Baylor College of Medicine/Texas Children’s Hospital. Her research interest is in the area of treatment-related toxicities and patient-reported outcomes among pediatric cancer patients.

Wei Pan, PhD, is an associate professor at the Duke University School of Nursing. His research interests include causal inference, advanced statistical modeling, meta-analysis, psychometrics, and their applications in the social, behavioral, and health sciences.

Footnotes

Declaration of Conflicting Interests: The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Institutes of Health R01CA1693398 and the Alex’s Lemonade Stand Foundation.

References

  1. Arbuckle J. L. (2010). IBM SPSS Amos 19 user’s guide. Amos Development Corporation. [Google Scholar]
  2. Betteridge D. J. (2000). What is oxidative stress? Metabolism, 49(2), 3-8. 10.1016/S0026-0495(00)80077-3 [DOI] [PubMed] [Google Scholar]
  3. Buettnera G. R., Wagnera B. A., Rodgers V. G. (2013). Quantitative redox biology: An approach to understanding the role of reactive species in defining the cellular redox environment. Cell Biochemistry & Biophysics, 67(2), 4774-83. 10.1007/s12013-011-9320-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Busu C., Li W., Caldito G., Aw T. Y. (2013). Inhibition of glutathione synthesis in brain endothelial cells lengthens S-phase transit time in the cell cycle: Implications for proliferation in recovery from oxidative stress and endothelial cell damage. Redox Biology, 1(1), 131-139. 10.1016/j.redox.2013.01.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Cheong J., Mackinnon D. P., Khoo S. T. (2003). Investigation of mediational processes using parallel process latent growth curve modeling. Structural Equation Modeling, 10(2), 238 10.1207/S15328007SEM1002_5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Cheung Y. T., Khan R. B., Liu W., Brinkman T. M., Edelmann M. N., Reddick W. E., Pei D., Panoskaltsis-Mortari A., Srivastava D., Cheng C., Robison L. L., Hudson M. M., Pui C.-H., Krull K. R. (2018). Association of cerebrospinal fluid biomarkers of central nervous system injury with neurocognitive and brain imaging outcomes in children receiving chemotherapy for acute lymphoblastic leukemia. JAMA Oncology, 4(7), e180089 10.1001/jamaoncol.2018.0089 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Collins J. J., Byrnes M. E., Dunkel I. J., Lapin J., Nadel T., Thaler H. T., Polyak T., Rapkin B., Portenoy R. K. (2000). The measurement of symptoms in children with cancer. Journal of Pain & Symptom Management, 19(5), 363-377. 10.1016/S0885-3924(00)00127-5 [DOI] [PubMed] [Google Scholar]
  8. Da Poian A. T., El-Bacha T., Luz M. R. M. P. (2010). Nutrient utilization in humans: Metabolism pathways. Nature Education, 3(9), 11. [Google Scholar]
  9. Hinds P. S., Yang J., Gattuso J. S., Hockenberry M., Jones H., Zupanec S., Li C., Crabtree V. M., Mandrell B. N., Schoumacher R. A., Vallance K., Sanford S., Srivastava D. K. (2010). Psychometric and clinical assessment of the 10-item reduced version of the Fatigue Scale—Child instrument. Journal of Pain & Symptom Management, 39(3), 572-578. 10.1016/j.jpainsymman.2009.07.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Hockenberry M., Hinds P. S., Barrera P., Bryant R., Adams-McNeill J., Hooke C., Rasco-Baggott C., Patterson-Kelly K., Gattuso J. S., Manteuffel B. (2003). Three instruments to assess fatigue in children with cancer: The child, parent and staff perspectives. Journal of Pain & Symptom Management, 25(4), 319-328. https://www.jpsmjournal.com/article/S08853924(02)00680-2/fulltext [DOI] [PubMed] [Google Scholar]
  11. Hockenberry M. J., Hooke M. C., Rodgers C., Taylor O., Koerner K., Moore I., Sheurer M., Pan W. (2017). Symptom trajectories in children receiving treatment for leukemia: A latent class growth analysis with multitrajectory modeling. Journal of Pain and Symptom Management, 54(1), 1-8. 10.1016/j.jpainsymman.2017.03.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Hooke M. C., Rodgers C., Taylor O. A., Koerner K., Mitby P., Moore I., Scheurer M., Hockenberry M. J., Pan W. (2018). Physical activity, the Childhood Cancer Symptom Cluster–Leukemia, and cognitive function; A longitudinal mediation analysis. Cancer Nursing, 41(6), 434-440. 10.1097/NCC.0000000000000634 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Hu L. T., Bentler P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6(1), 1-55. 10.1080/10705519909540118 [DOI] [Google Scholar]
  14. Jahovic N., Cevik H., Sehirli A. O., Yeğen B. C., Sener G. (2003). Melatonin prevents methotrexate-induced hepatorenal oxidative injury in rats. Journal of Pineal Research, 34(4), 282-287. 10.1034/j.1600-079X.2003.00043.x [DOI] [PubMed] [Google Scholar]
  15. Kelly D. L., Dickinson K., Hsiao C. P., Lukkahatai N., Gonzalez-Marrero V., McCabe M., Saligan L. N. (2016). Biological basis for the clustering of symptoms. Seminars in Oncology Nursing, 32(4), 351-360. 10.1016/j.soncn.2016.08.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Kovacs M. (2010). Children’s Depression Inventory 2 (CDI 2). Pearson. [Google Scholar]
  17. Lai J. S., Zelko F., Butt Z., Cella D., Kieran M. W, Krull K. R., Magasi S., Goldman S. (2011). Parent-perceived child cognitive function: Results from a sample drawn from the US general population. Child’s Nervous System, 7(2), 285-293. 10.1007/s00381-010-1230-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Lai J. S., Zelko F., Krull K. R., Cella D., Nowinski C., Manley P. E., Goldman S. (2014). Parent-reported cognition of children with cancer and its potential clinical usefulness. Quality of Life Research, 23(4), 1049-1058. 10.1007/s11136-013-0548-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. LeBourgeois M., Harsh J. (2016). Development and psychometric evaluation of the Children’s Sleep-Wake Scale (CSWS). Sleep Health, 2(3), 198-204. 10.1016/j.sleh.2016.04.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Lim J. Y., Bhatia S., Robison L. R., Yang J. J. (2014). Genomics of racial and ethnic disparities in childhood acute lymphoblastic leukemia. Cancer, 20(7), 955-962. 10.1002/cncr.28531 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Mandrell B. N., Yang J., Hooke M. C., Wang C., Gattuso J. S., Hockenberry M., Jones H., Zupanec S., Hinds P. S. (2011). Psychometric and clinical assessment of the 13-item reduced version of the Fatigue Scale–Adolescent instrument. Journal of Pediatric Oncology Nursing, 28(5), 287-294. 10.1177/1043454211418667 [DOI] [PubMed] [Google Scholar]
  22. Margaill I., Plotkine M., Lerouet D. (2005). Antioxidant strategies in the treatment of stroke. Free Radical Biology & Medicine, 39(4), 429-443. 10.1016/j.freeradbiomed.2005.05.003 [DOI] [PubMed] [Google Scholar]
  23. Meek R., Kelly A. M., Hu X. F. (2009). Use of the visual analog scale to rate and monitor severity of nausea in the emergency department. Academic Emergency Medicine, 16(12), 1304-1310. 10.1111/j.1553-2712.2009.00581.x [DOI] [PubMed] [Google Scholar]
  24. Meister A., Anderson M. E. (1983). Glutathione. Annual Review of Biochemistry, 52, 711-760. 10.1146/annurev.bi.52.070183.003431 [DOI] [PubMed] [Google Scholar]
  25. Moore I. M., Hockenberry M. J., Krull K. R. (2013). Cancer-related cognitive changes in children, adolescents and adult survivors of childhood cancers. Seminars in Oncology Nursing, 29(4), 248-259. 10.1016/j.soncn.2013.08.005 [DOI] [PubMed] [Google Scholar]
  26. Moore I. M., Koerner K. M., Gundy P. M., Montgomery D. W., Insel K. C., Harris L. L., Insel K. C., Harris L. L., Taylor O. A., Hockenberry M. J. (2018). Changes in oxidant defense, apoptosis, & cognitive abilities during treatment for childhood leukemia. Biological Research for Nursing, 20(4), 393-402. 10.1177/1099800418763124 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. National Cancer Institute. (2019). Childhood acute lymphoblastic leukemia treatment (PDQ®)–health professional version. https://www.cancer.gov/types/leukemia/hp/child-all-treatment-pdq#_565 [PubMed]
  28. Ӧsterlundh G., Kjellmer I., Lannering B., Rosengren L., Nilsson U. A., Márky I. (2008). Neurochemical markers of brain damage in cerebrospinal fluid during induction treatment of acute lymphoblastic leukemia in children. Pediatric Blood & Cancer, 50(4), 793-798. 10.1002/pbc.21378 [DOI] [PubMed] [Google Scholar]
  29. Pollock B. H., DeBaun M. R., Camitta B. M., Shuster J. J., Ravindranath Y., Pullen D. J., Land V. J., Mahoney D. H., Jr., Lauer S. J., Murphy S. B. (2000). Racial differences in the survival of childhood B-precursor acute lymphoblastic leukemia: A Pediatric Oncology Group study. Journal of Clinical Oncology, 18(4), 813-823. 10.1200/JCO.2000.18.4.813 [DOI] [PubMed] [Google Scholar]
  30. Rabin K. R., Gramatges M. M., Margolin J. F., Poplack D. G. (2016). Acute lymphoblastic leukemia. In Pizzo P. A., Poplack D. G. (Eds.), Principles and practice of pediatric oncology (7th ed., pp. 463-544). Wolters Kluwer. [Google Scholar]
  31. Rodgers C., Hooke M. C., Ward J., Linder L. A. (2016). Symptom clusters in children and adolescents with cancer. Seminars in Oncology Nursing, 32(4), 394-404. 10.1016/j.soncn.2016.08.005 [DOI] [PubMed] [Google Scholar]
  32. Rodgers C., Sanborn C., Taylor O., Gundy P., Pasvogel A., Moore I. M., Hockenberry M. J. (2016). Fatigue and oxidative stress in children undergoing leukemia treatment. Biological Research for Nursing, 18(5), 515-520. 10.1177/1099800416647794 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Rodgers C., Hooke M. C., Taylor O., Koerner K., Mitby P., Moore I. M., Scheurer M. E., Hockenberry M. J., Pan W. (2019) Childhood Cancer Symptom Cluster: Leukemia & health-related quality of life. Oncology Nursing Forum, 46(2), 228-237. 10.1188/19.ONF.228-237. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. SAS Institute Inc. (2011). SAS/STAT® 9.3 user’s guide. [Google Scholar]
  35. Schafer F. Q., Buettner G. R. (2001). Redox environment of the cell as viewed through the redox state of the glutathione disulfide/glutathione Couple. Free Radical Biology & Medicine, 30(11), 1191-1212. 10.1016/S0891-5849(01)00480-4 [DOI] [PubMed] [Google Scholar]
  36. Scott J., Huskisson E. C. (1979). Vertical or horizontal visual analogue scales. Annals of Rheumatic Diseases, 38(6), 560 10.1136/ard.38.6.560 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Sherman S. A., Eisen S., Burwinkle T. M., Varni J. W. (2006). The PedsQL™ Present Functioning Visual Analogue Scales: Preliminary reliability and validity. Health and Quality of Life Outcomes, 4(1), 75 10.1186/1477-7525-4-75. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Siegel R. L., Miller K. D., Jemal A. (2018). Cancer statistics, 2018. CA: A Cancer Journal for Clincians, 68(1), 7-30. 10.3322/caac.21442 [DOI] [PubMed] [Google Scholar]
  39. Sousa K. H., Kwok O. M., Schmiege S. J., West S. G. (2014). A longitudinal approach to understanding the relationship between symptom status and QOL. Western Journal of Nursing Research, 36(6), 732-747. 10.1177/0193945913510980 [DOI] [PubMed] [Google Scholar]
  40. Stinson J. N., Kavanagh T., Yamada J., Gill N., Stevens B. (2006). Systematic review of the psychometric properties, interpretability and feasibility of self-report pain intensity measures for use in clinical trials in children and adolescents. Pain, 126(1-2), 143-157. 10.1016/j.pain.2006.05.006 [DOI] [PubMed] [Google Scholar]
  41. Storfer-Isser A., Lebourgeois M. K., Harsh J., Tompsett C. J., Redline S. (2013). Psychometric properties of the adolescent sleep hygiene scale. Journal of Sleep Research, 22, 707-716. 10.1111/jsr.12059 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Wong D., Baker C. (1988). Pain in children: Comparison of assessment scales. Pediatric Nursing, 14(1), 9-17. [PubMed] [Google Scholar]

Articles from Journal of Pediatric Oncology Nursing are provided here courtesy of SAGE Publications

RESOURCES