The effect of chemotherapy on cognitive functions in children with leukemia

Overview

Leukemia is one of the most common childhood diseases with a malignant course. Fortunately, today there is a high probability of its successful cure. The nature of the disease entails a high risk of initial spread, or later relapse of the disease into the central nervous system. Standard treatment including intrathecal application of cytostatics reduces this risk fundamentally; nevertheless it can damage some cognitive functions. There is extensive research dealing with this issue. Through a systematic literature research, we selected 16 articles that examined cognitive functions through neuropsychological tests. Result analysis indicates a statistically significant decrease in performance in tests of attention, processing speed, working memory and executive functions. Conversely, the overall intelligence quotient, learning ability and long-term memory seem to be only slightly affected by the treatment.

Keywords:

Chemotherapy – Intelligence – cognitive function – leukemia – pediatric hematology/oncology

This is an unauthorised machine translation into English made using the DeepL Translate Pro translator. The editors do not guarantee that the content of the article corresponds fully to the original language version.

 

 

Introduction

After accidents, cancer is the second most common cause of death in children and adolescents in developed countries [1]. Approximately 400,000 children are diagnosed with this disease worldwide each year [2]. Childhood and adult cancers are similar in their nature, i.e., uncontrolled growth and proliferation of abnormal cells. Without treatment, they are incompatible with life. Due to their different dynamics, childhood cancers are more sensitive to chemotherapy than adult cancers [2] and chemotherapy is therefore the main treatment modality. The success rate of treatment, measured as survival of 5 years or more from diagnosis, has been increasing over the last few decades and exceeds 80% [3]. With the increasing success rate, there is also a growing need to monitor the side effects of treatment, which may manifest at the somatic and neuropsychiatric levels.

Leukaemia is the most common cancer in children under 14 years of age, accounting for 30% of all cancers, and up to 36% in the age group under 4 years [4]. The incidence of leukemias decreases with age, with a prevalence of around 15% in adolescents between 15 and 19 years of age, with lymphomas and epithelial tumors being more common [5].

Leukemia is the collective name of a group of diseases whose common characteristic is malignant transformation of hematopoietic cells. According to the type of malignant cells affected, we distinguish acute lymphoblastic leukemia (ALL), acute myeloblastic leukemia (AML) and chronic myeloid leukemia (CML). ALL is the most typical for children, accounting for about 80% of cases [4,6].

Leukemia is a chemosensitive disease. Treatment takes approximately 2 years and the combination of cytostatics and the intensity of treatment depend on the response to treatment in the first, so-called induction phase [7]. From a neurological perspective, the highest risk of adverse effects is associated with vincristine, which can cause polyneuropathy [8,9]. The concomitant effect of glucocorticoids and leukemia itself can then lead to osteoporosis (and compressive vertebral textures affecting spinal nerves) [10,11].

Leukemia cells have the ability to cross the blood-brain barrier, infiltrate the CNS and become trapped in the brain membranes. Here, due to their adaptability, they can survive for a long time or start proliferating [12]. Cytostatics penetrate the blood-brain barrier only to a low extent [13] and therefore may not affect leukemic cells. Surviving leukemic cells can later cause disease relapse [14]. Without systemic prophylaxis, which is already a common part of treatment, CNS relapse occurred in 30-50% of children [4,14].

The original method of CNS prophylaxis was cranial radiotherapy, which increased the cure rate to the current 80-90% [15], but radiation caused some cognitive decline, uncontrollable epileptic seizures, cerebral haemorrhages, endocrinopathies and secondary CNS tumours [16,17]. Therefore, radiotherapy has been replaced by intrathecal administration of methotrexate in most children, which yields comparable results regarding CNS relapses with concomitant lower neurotoxicity [17]. Radiotherapy now remains part of the treatment protocols for children at highest risk, i.e., about 15% of all children with leukemia [7]. Children with an inadequate response to treatment in the induction phase and a cure probability of less than 50% are indicated for bone marrow transplantation [18].

Unfortunately, even more sparing chemotherapy prophylaxis carries the risk of CNS damage. Studies confirm the association between intrathecal administration of cytostatics and leukoencephalopathy [19,20]. Intrathecally administered methotrexate can cause immediate toxicity, leading to disruption of cerebral gray matter and/or demyelination of white matter fibers [21,22]. The leukoencephalopathy thus produced may be acute ("stroke-like" symptoms, epiparoxysmal symptoms, aphasia) or asymptomatic. An additional risk is the secondary immunological response to methotrexate, which induces an inflammatory reaction that may result in microvascular injury. Both leukoencephalopathy and vascular injury can lead to a decline in mental performance. In particular, skills critical for success in school are affected [23], such as attention and acquisition of new knowledge (learning capacity), visuospatial skills, and others [24]. Difficulties in mastering school material can subsequently cause a widening of the already existing gap between children with leukaemia and their healthy peers [25].

Cognitive function in children with leukemia has been studied by many different authors. In our article we would like to summarize the results of the studies of the last 10 years, to focus on the intelligence level and the most frequently observed cognitive functions and to compare the results of different authors or scientific groups. We conducted the literature search in response to children's recurrent difficulties in coping with the demands of learning when returning to normal school routines after leukemia treatment. The results of the search can serve as a basis for long-term monitoring of the cognitive performance of children with leukaemia and subsequent design of appropriate educational and psychological support measures.

 

Methods

For this review, we searched all English-language publications related to the topic of cognitive function and intelligence level in children with leukemia published between January 2011 and December 2021. The publications were searched in the PubMed electronic database using a combination of keywords and the terms: (leukemia or ALL) AND (child*) AND (neurocognit* or neuropsych* or cognit*). We further supplemented the search with articles from the Google Scholar database using the same keywords and time constraints. The third source was references, which referred to relevant articles found and studied in the search.

The first PubMed database search yielded 439 articles, of which 114 were selected based on title. In the next step, we selected papers that met the criteria listed below based on the abstract or reading of the article:

a)
diagnosis of childhood leukaemia;

b)
purely chemotherapeutic treatment, with intrathecal administration of cytostatics;

c)
the examination of cognitive function was carried out by performance tests;

d)
the neuropsychological test scores and the differences or statistical significance compared to the control group;

e)
a minimum of 20 children in the study population.

 

Articles obtained by other methods were duplicated or did not meet some of the selection criteria.

 

Sixteen studies met the established criteria (Table 1) [26-41]. The size of the groups studied ranged from 22-243, with a median of 71 patients. Outcomes were compared with population norms in nine studies and with healthy, demographically matched controls in four studies. One study [38] compared intelligence test scores with healthy controls, and subtests of cognitive function with population norms. Finally, two groups were constructed prospectively, i.e., they monitored the change in performance of children with leukemia over time.

The studies that we obtained by literature search were based on different assumptions and clinical studies. Thus, their selection of the cognitive functions studied varied. In this article, we focus on intelligence level and cognitive functions that represent the intersection of studies, i.e., we operationalized the most frequently observed cognitive functions as follows: total intelligence quotient (CIQ), learning and memory, attention, processing speed, working memory, and executive functions. We are aware that the definitions of the above cognitive functions may differ, especially for executive functions and working memory. In this paper, we use the term executive function in accordance with the interpretation of Lezak et al. [42] as a set of interacting abilities necessary for planning intentional behavior and executing sub-steps to achieve a desired goal. Working memory, in turn, as the ability to temporarily hold information in memory in order to use it for more complex mental operations.

 

Results

Statistical significance

Statistical significance, i.e., the assumption of cognitive impairment by CNS prophylaxis, is reported at the significance level of p < 0.05 and adequate values, i.e., with 95% confidence. A summary of the results, i.e., their statistical significance with respect to the criteria above, is presented in Table 2 [26-41]. Statistically significant impairment in any of the cognitive functions studied was reported by 14 of the 16 studies. We give equal weight to all studies in the results, and any adverse effect of outliers was controlled by summing all children tested in the target group and controlling for proportions in this variable as well. The number tested ranged from 732 for executive functions to 1,326 for attention. The percentage of results indicative of impairment is comparable when we look at the ratio of studies and the ratio of the number of children included in those studies (Figure 1). The results differ only for executive functions, where 67% of studies are suggestive of impairment. However, if we add up all the children included in the studies, only 41% of them show impairment. The difference may be due to methodological reasons (see below).

 

Observed group

All subjects had childhood leukemia and were treated with chemotherapy alone. The treatment included intrathecally administered methotrexate in all cases. The doses of methotrexate were determined according to the currently valid international treatment protocols, or adjusted according to the clinical status and tolerability of treatment in individual children. Only a few studies have reported the average amount or number of doses [28,31,33,34], and the values are very similar and range around 170 g/mm².

 

Time of testing since diagnosis, age of probands

The time of testing children from the diagnosis of leukemia was heterogeneous and ranged from 2 or 3 years [27,32,35,40] to more than 20 years [31,33]. Studies conducted within 5 years of diagnosis were more likely to report significant reductions in performance in subdomains (10 of the 14 domains studied; note that later in this article we use the word domain as a parent term for the CIQ and the subdomains of cognitive function described), despite the fact that results were compared with normative data in most cases (see below). Studies conducted more than 5 years after diagnosis reported impairment in approximately half of the cases (22 of 41 domains studied), with more than half comparing results with healthy controls. The median age of the children at the time of testing was 13.1 years, and the median age reported in each study ranged from 9 [27,32,35] to 28 years [31,33]. Because the studies reported only the median ages of the children, a more detailed analysis of the effect of age at disease diagnosis on outcomes was not possible. The mean age of the children at diagnosis was tentatively between 2.4 and 9.9 years with a median age of approximately 6 years.

 

Control group

Performance was most often compared with normative data (9 of 16 studies). Four studies compared performance with healthy, demographically matched peers and only two studies compared outcomes prospectively [32,40]. One study [38] compared intelligence test scores with healthy controls, and subtests of cognitive function with normative data. The studies comparing results with normative data and healthy controls looked at a similar number of domains, averaging 3.9 and 3.6 domains, respectively. The overall results were more positive when comparing with normative data, reporting impairment "only" in 49% of cases. When comparing the results with healthy controls, damage was reported in 72% of cases. Prospective studies are minimally represented -⁠ two studies, each looking at one domain. Regardless of statistical significance and control group, all authors reported lower performance in the cognitive functions studied in children after undergoing leukemia treatment. Only Watanabe et al [40] observed a slight increase in performance on the CIQ.

 

Domains

Cecal IQ was looked at in ten studies, seven of which found no statistically significant change in performance in children with leukaemia. Only the study by Watanabe et al [40] reported even further decomposition of CIQ scores. This showed weaker performance on verbal than on performance subtests, which corresponded with earlier studies [43,44].

 

Learning ability and memory

Eight studies investigated the ability to remember, retain and recall information with distance. Two reported a statistically significant decline in performance in children after chemotherapy treatment for leukemia [29,30]-Genschaft et al. [29] in verbal and nonverbal memory, Jacola et al. [30] in particular in learning capacity. The results of the remaining studies showed only mild impairment in both short-term and long-term memory. The results are comparable across different control groups and at different times since diagnosis.

 

Attention

This category included studies that measured attention with one of the neuropsychological tests of attention or looked at the number of errors from oversight in comprehensive test batteries. Eight of the ten studies that included attention tests reported statistically significant decreases. The remaining two studies [28,33] were administered at the longest time points since diagnosis, nearly 21 and 15 years, respectively. The median time between diagnosis and test administration for the remaining studies ranged between 2 [27] and 9 years [38].

 

Processing speed

Processing speed was measured by various tests in the studies. We excluded purely motor tests from the analysis, the outcome of which may be influenced by other side effects of the treatment, especially neuropathy. For completeness, we can note that in the case of motor tests, three of the four studies reported statistically significant reductions in performance. The group of tests reported in the article includes verbal, visual, and visuomotor tests, and most studies measured processing speed by combining multiple psychodiagnostic methods. The results of the different subtests in terms of statistically significant or non-significant performance decrements were mostly consistent within the same study, or only one of the several psychodiagnostic tests came out as non-significant, such as in Zając-Spychała et al. [41]. Six of the eight studies reported a significant performance decrease in processing speed.

 

Working memory

Working memory, the ability to temporarily retain and work with information in memory, is classified as an executive function along with inhibition control and cognitive flexibility [45]. Because of their direct relationship to academic achievement [46], they are often tested separately. Out of 16 studies, nine focused on working memory, and four of them assessed only working memory performance, no longer executive functions. Significantly lower performance was reported by six of the nine studies. All three studies not suggestive of a significant reduction in performance compared results with normative data, the others equally with healthy controls and normative data.

 

Executive functions

Six studies assessed executive function, five of which also assessed working memory. The results were mostly concordant, i.e., both working memory and executive function performance were impaired or not impaired across studies. Statistically significant reductions in performance were reported by four of the six studies. All studies were conducted more than five years after diagnosis.

 

Discussion

The subject of our research was to investigate the possible effect of pure chemotherapeutic treatment of childhood leukemia on cognitive function. In the first part of our review, a systematic literature search was conducted to select 16 articles that examined intelligence and cognitive function through neuropsychological performance tests. We then included in the review those domains that have been investigated by at least six different studies. For the purposes of our review, we divided the published results into two bands, i.e., indicative vs. not indicative of performance decline, according to quantitative results at the 95% confidence level. Conclusions from the analysis indicate that intrathecal administration of cytostatics in the treatment of leukemia has a significant negative effect on attention, processing speed, working memory, and executive function. Lower performance was also observed for CIQ and memory and learning ability, but the decrease was not statistically significant.

Damage to only some functions may be due to their localisation. The functions for which performance impairment has been observed are associated with frontal lobe activity, whereas structures essential for memory and learning are mainly located in the parietal and temporal lobes and hippocampus. This reasoning also corresponds with the findings of Krull et al [34], who describe structural changes in the frontal lobe and intact parietal and occipital lobes during functional MRI of children in our study group. In line with the above, it is hypothesized that the fluid components of intelligence (the so-called fluid " g"), which include some aspects of working memory and executive functions, may indeed be affected by leukemia treatment, but because the CIQ is also strongly saturated by crystalline abilities (the so-called crystalline " g"), which are fortunately less impaired, CIQ differences between healthy and treated children do not reach statistical significance. Tests of executive function are constructed in a similar way to the CIQ, i.e. the score is based on performance in multiple cognitive domains. However, these are primarily dependent on the integrity of the prefrontal cortex.

Another variable in the research is the large variance in the time elapsed between examinations and diagnosis or treatment. Six of the 16 studies we analysed were conducted within 5 years of diagnosis. The authors did not give reasons in their articles. The distribution of disease in the different age groups could have influenced the choice of study time. The highest incidence of leukemia is in children between 2 and 6 years of age, when tests of partial cognitive functions are difficult to perform because of CNS immaturity and the high variance of results even in the normal population. Thus, excluding younger children could significantly reduce the research sample. The profile of the functions studied would also support the consideration, as most early research focuses on more complex abilities (CIQ, memory and learning) that we assess relatively reliably in younger children.

In spite of the above, we consider the comparison of research conducted up to and beyond 5 years after diagnosis to be very interesting. For the analysed studies, it was evident that the results of studies conducted with a longer delay were more positive for all domains studied. A similar slightly positive trend was reported for the CIQ in the study by Watanabe et al [40], who tested children at three time points in their prospective study. The improvement may be due to brain maturation, the positive effect of brain plasticity, but also due to cognitive training.

Another variable that may influence the results presented here is the control group. The results comparing children's performance with the normative sample may be slightly overestimated. This may be due to the ability to better control for the health status and cognitive performance of children, both healthy and cured, entering the study, the greater willingness of more educated parents to participate in the study, and finally, a better understanding of the reasons for testing, and thus greater motivation of children after leukemia.

The results obtained may be affected by several variables that we could not control for for the purposes of our review. In addition to the type of control group mentioned above, we consider the following points to be key.

1)
Studies looking at individual cognitive functions have worked with different test batteries. The different methods used in the studies differ in terms of their difficulty or which aspect of cognitive function they focus on.

2)
As indicated in the introduction, the treatment of leukaemia in children varies in frequency and dose of cytostatics. Consistent with available sources, it can be assumed that the risk and degree of cognitive impairment increases with higher doses of intrathecally administered cytostatics [27,47]. Unfortunately, interindividual differences in performance status as a function of treatment intensity cannot be assessed from the studies included here.

3)
The age at which the treatment takes place may also have an influence. Several studies have looked at the age dependence of the degree of cognitive impairment, but typically they divide children under and over five years of age at the time of diagnosis. Very young children have been shown to be more susceptible to cognitive impairment [30,48]. The effect of leukaemia treatment on older children remains a question, as adolescence is a time of intense development of higher cortical functions and thus greater vulnerability might be expected [49].

 

Conclusion

Chemotherapeutic treatment of leukaemia, including intrathecal administration of cytostatics, in addition to its clear health benefits, also carries the risk of side effects in the form of reduced performance of some cognitive functions. In particular, cognitive functions with centres in the frontal lobe of the brain -⁠ attention, information processing speed, working memory and executive functions -⁠ appear to be highly vulnerable. Less so are CIQ and memory and learning. Decreases in performance in the above abilities can have a significant negative impact on school performance and daily life in children after leukaemia treatment and should be monitored regularly and, if necessary, rehabilitated and compensated for early.

 

Acknowledgements

Thank you, doc. Mgr. Ondřej Bezdíček, Ph.D. for his help and valuable advice in the preparation of the article.

 

Grant support

Supported by the Charles University Grant Agency (GAUK 260121).

 

Conflict of interest

The authors declare that they have no conflict of interest in relation to the subject of the paper.

 

Tables

Table 1: Articles on the effect of chemotherapy on cognitive function in children with leukaemia.

Table 2. Overview of selected cognitive functions studied by individual authors.