Toxicity of CNS Prophylaxis for Childhood Leukemia
Toxicity of CNS Prophylaxis for Childhood Leukemia
Central nervous system (CNS) prophylaxis has been an essential component
of the treatment of childhood acute lymphoblastic leukemia (ALL) for several
decades. Early prophylactic treatment of CNS minimal residual disease is
intended to guard against the possibility that CNS blasts not eradicated
by systemic therapy will reseed the bone marrow, leading to relapse of
the disease. For many years, the preferred approach to CNS prophylaxis
has been cranial radiation therapy (CRT), currently given at an 18-Gy dose,
combined with intra-thecal methotrexate. This strategy is highly effective
in preventing CNS relapse.[2,3]
The potential for long-term neurotoxicity associated with such intensive
treatment in developing children, however, is an ongoing concern; cognitive
impairment and short stature are commonly observed. Children diagnosed
with ALL and treated with modern protocols experience excellent survival
rates. As the number of survivors grows, the prevalence of these developmental
toxicities becomes more clinically relevant. Although efforts have been
undertaken to develop alternative methods of CNS prophylaxis that do not
involve irradiation (such as intrathecal methotrexate, often given in combination
with other drugs), treatment that includes CRT remains the preferred approach
for children with high-risk disease.
Although it is generally assumed that late neurotoxic effects result
from CNS prophylactic therapy, the experimental designs of leukemia protocols
may or may not allow this hypothesis to be tested. Central nervous system
prophylaxis occurs in the context of complex treatment protocols that include
a variety of neurotoxic agents. Therefore, it is important to document
the risk for various toxicities and to determine to what extent they can
be reliably associated with specific treatment components.
In the context of a randomized trial, outcomes can be clearly evaluated,
and alternative approaches to CNS prophylaxis can be examined. Such opportunities
are rare, however. More often, different approaches to CNS therapy are
evaluated by comparing children assigned to different risk groups or protocols[5-7]
or even children with different diseases.[8,9] In these situations, variations
in CNS treatment may be confounded by other aspects of the protocol or
the disease, rendering evaluation less reliable.
Furthermore, components of CNS prophylaxis may interact synergistically
with systemic therapy[5-10]; CNS prophylactic treatment is associated with
different outcomes depending on the systemic therapy. Therefore, in the
discussion that follows, the late effects of CNS prophylactic therapy will
be discussed and, when possible, related to specific components of such
therapy. It should be understood, however, that in most instances, the
more conservative approach is to relate late effects to the therapeutic
picture with the assumption that CNS prophylaxis plays a major role.
Although the major focus of this discussion is long-term toxic effects,
significant acute toxic effects can be associated with CNS prophylaxis
(Table 1). Cranial irradiation has
been associated with vomiting, anorexia, headache, and somnolence. Intrathecal
methotrexate may be associated with acute arachnoiditis, pain at the site
of the lumbar puncture, nausea and vomiting, fever, and an increase in
intracranial pressure. Most of these acute toxic effects are short-lived,
and few result in long-term consequences. Myelopathy and encephalopathy
may also occur but are rare.
The late toxic effects with the highest prevalence affect physical and
mental development (Table 1). These effects
are permanent, rather than acute, with the potential for causing lifelong
problems. The risk for a second malignancy is also increased in children
treated with CRT, but the frequency is low.
There are no good epidemiologic estimates on the prevalence of problems
with cognitive development and physical growth, for several reasons: First,
as will be discussed in detail below, the incidence and severity of these
problems are highly dependent on a variety of factors, such as the specific
characteristics of the leukemia treatment protocol and the age and gender
of the child at the time of diagnosis. Prevalence, therefore, varies from
institution to institution, depending on the treatment protocol used and
the demographic characterisitics of the children treated.
Second, as will also be discussed below, with respect to cognitive issues,
there is no standard way of defining outcome. Indeed, the definition of
a learning disability, as a diagnosis, remains the subject of some controversy.
Consequently, it is difficult to specify the prevalence of a "disorder."
Risk for a particular child, therefore, is best estimated by consideration
of the various factors discussed below.
Ocular morbidity, attributed largely to corticosteroids, consists of
posterior subcapsular cataracts. They tend to be asymptomatic and do not
usually progress or cause loss of vision. Dental complications are
minimal with the low CRT dose (18 Gy), but preventive dental programs are
The cognitive sequelae of therapy have been of greatest concern, and
this issue remains a controversial topic. On the one hand, it is undesirable
to employ a therapy that is known to cause lifelong impairment when such
therapy is not necessary to achieve the desired medical outcome. On the
other hand, effective therapy should not be withheld if it does not, in
fact, cause such sequelae or if the medical outcome will be compromised.
Moreover, the same therapy may be toxic in one group of patients but not
in another. In such a situation, it would be unwise to deny the benefit
of such therapy to the group who would not have experienced adverse effects
to spare the group who would have been affected. Because of the confusion
surrounding cognitive outcomes, the bulk of the discussion here focuses
on these effects.
Children treated for ALL are at risk for a decrement in final adult
height, often on the order of one standard deviation from expectation.
Acute changes in growth rate are common during treatment, but recovery
often occurs after cessation of treatment. Of greater concern is the
decreased amplitude of peak height velocity, a long-term effect that may
occur during the adolescent growth spurt and is associated with a decrement
in final height.
Some studies have indicated that girls are more vulnerable to growth
delay than boys. Menarche, as well as peak height velocity, may occur earlier
in girls who have undergone treatment than in girls who have not. Young
age at treatment (before the age of 7) is a particular risk factor for
girls,[16,17] with boys appearing to be far less vulnerable
In general, treatment protocols that include CRT are associated with
a permanent decrease in height, whereas those that do not include CRT are
not [14,16,18]. The specificity of the impact of cranial irradiation on
growth is substantiated by preclinical studies. Rats exposed to CRT,
either alone or in combination with antileukemic drugs (eg, prednisolone,
methotrexate), show disturbances in bone growth, whereas those exposed
to drugs without CRT show no such disturbances. Clinically, children treated
with a high dose of cranial irradiation (24 Gy) show more significant decrements
in height than do those treated with a low dose (18 Gy), but growth
can be adversely affected by both doses. Children receiving lengthy, intensive
maintenance drug therapy, however, may be at greater risk for diminished
Whereas the relationship between treatment and outcome is reasonably
straightforward in the case of physical growth, it is more controversial
with respect to mental development. A fundamental question centers on the
end point to be measured. For height, the end point is clear, easily agreed
upon, and reliably defined; however, this is not the case for mental development.
There is no widely accepted technique for measuring cognitive outcomes,
with the possible exception of the IQ test. The IQ test is a good indicator
of a child's general level of cognitive functioning, and most investigators
use it, which permits good comparability among studies. However, the IQ
test is not sensitive to the subtler aspects of information processing
that can manifest themselves in problems related to learning and social
skills. Indeed, the hallmark of the definition of a child with a learning
disability is that the child functions in the normal range of intelligence
but experiences unexplained difficulties at school.
The IQ test is sensitive to neurotoxicity in severely affected children
but can be insensitive to more subtle deficits, even though they may have
a substantive functional impact. Consequently, investigators, particularly
neuropsychologists, typically supplement the IQ test with other measures
of information processing, although there is no standard way of doing so.
Thus, outcome measures vary widely from study to study, depending on the
theoretical bent or clinical practices of the investigator. This makes
comparison of study results quite difficult.
Role of CRT
Probably the central questions throughout the literature on late cognitive
sequelae are whether CRT affects outcome and, if so, to what extent. Of
all the agents used in the treatment of leukemia, CRT has stimulated the
most concern; the assumption is that observed cognitive deficits are primarily
referable to CRT. A number of studies have indicated that children treated
with cranial irradiation fare worse than a comparison group not given such
treatment.[21-25] The majority of these studies evaluated cognitive outcomes
in children treated with 24 Gy of CRT, a dose that is no longer commonly
Because contemporary treatment protocols employ an 18-Gy dose of CRT,
data that evaluate the impact of that dose are more relevant. In general,
the trend is toward less severe toxic effects for protocols that involve
the lower dose. Several studies comparing groups treated with and without
CRT at the 18-Gy dose document no differences in IQ or other measures.[5,27]
Another multicenter study, however, did observe lower IQ scores in children
treated with the 18-Gy dose than in those not treated with CRT. The
adverse impact of CRT, however, emerged only in the youngest children (les
than 3 years of age at treatment), suggesting that for older children,
CRT at the lower dose poses less risk.
Concurrent methotrexate therapy may affect the degree of toxicity associated
with CRT. Whereas an 18-Gy dose of cranial irradiation resulted in no
discernible decrease in IQ scores, the same dose preceded by a single high
dose (4 g/m²) of methotrexate resulted in a reliable decrement. Significantly,
the same high dose of methotrexate did not result in adverse cognitive
sequelae when treatment did not include CRT. Again, comparable findings
have emerged from an animal model. Whereas CRT or methotrexate alone did
not induce behavioral changes in animals, the combination of the two agents
was associated with significant behavioral changes.
Furthermore, adverse cognitive sequelae have been observed in children
who did not receive CRT. In general, these sequelae do not entail IQ deficits,
but rather, problems involving more specific skills. For example, comparable
deficits in rote memory have been found in children treated with or without
CRT. Giralt and colleagues compared children treated with CRT (24-Gy
dose) and intrathecal methotrexate or intrathecal cytarabine and intrathecal
methotrexate in the context of a randomized trial design. Both groups showed
deficits relative to those of controls, but the magnitude of the deficits
was similar in the two groups. Kaufmann and associates describe declines
in attentional, visual-motor, and academic skills in children for whom
triple intrathecal therapy (methotrexate, cytarabine, and hydrocortisone)
was used for CNS prophylaxis. In none of these studies, however, was it
possible to determine to what extent any adverse outcomes were attributable
to CNS prophylaxis per se and to what extent they reflected the impact
of other components of treatment.
Glucocorticoid therapy may play a role in inducing cognitive problems.
Prednisone is associated with acute memory problems in children and
adults. These agents cross the blood-brain barrier and are active in
areas of the brain essential for learning. Moreover, steroids may be
associated with disordered behavior in animals when administered alone
and especially when administered in combination with CRT and methotrexate.
Although this question has yet to be examined systematically in the clinical
setting, it should be a consideration in evaluating the potential toxicity
associated with protocols.
Effects Related to Gender and Age
As previously indicated, girls are more vulnerable than boys to growth
changes associated with CRT. Heightened cognitive vulnerability in girls
has also been observed; this vulnerability has emerged clearly in treatment
protocols that included a high dose of cranial irradiation (24 Gy).[34-36]
The basis for this phenomenon is unclear. Heightened vulnerability to toxicity
in girls is not confined to the CNS; girls are also more vulnerable to
anthracycline-induced cardiac toxicity. Moreover, gender differences
in treatment efficacy have been observed in some circumstances, with the
rate of CNS relapse higher for boys than for girls.
Gender-related differences are less evident, however, with a low dose
of CRT (18 Gy). At this lower dose, boys and girls appear to have essentially
comparable cognitive outcomes. As previously indicated, the impact of cranial
irradiation can be exacerbated (for girls) by a high dose of methotrexate,
a synergistic interaction that has not been observed in boys.
Age-related effects have also been observed, with younger children more
severely affected than older chil-dren.[7,34,36,39] Again, this effect
is more prominent in girls.[34,36,39]
Clinical Presentation of Neurobehavioral Problems
As previously indicated, contemporary leukemia protocols are likely
to result in subtle problems, rather than in global depression of IQ, as
was seen following earlier, more intensive pro-tocols. This is not to say
that such problems will not have important consequences for a child's daily
functioning, but rather, that they are probably less severe than those
seen a decade ago, when a 24-Gy dose of CRT was the normative dose. It
is important to appreciate the qualitative nature of these problems, the
way in which they manifest themselves in children's lives, and their developmental
In the general population, the most commonly diagnosed learning disability
syndromes are reading disability (dyslexia) and disordered attention (attention
deficit hyperactivity disorder [ADHD]). These problems are relatively specific,
tend to be idiopathic, and often seem to run in families. The problems
seen in children who have undergone treatment of ALL, however, tend to
be of a different nature. Learning problems seen in these children often
involve so-called nonverbal learning disabilities. Most of these children
acquire elementary reading skills without too much difficulty, although
some may require support, and few exhibit the impulsivity, overactivity,
and distractibility characteristic of ADHD. Although many children exhibit
apparent attentional problems, they are best understood as being secondary
to cognitive issues, as described in greater detail below, rather than
as being a primary attentional syndrome or ADHD.
Metacognitive problems have been described in survivors of leukemia
in a variety of contexts than can involve working with complex material
that is either verbal or nonverbal.[41-43] The term "metacognition"
means "knowing about knowing," that is, developing strategies
for solving problems and understanding the nature and organization of knowledge.
Metacognition implies knowledge and insight into one's own thought and
problem-solving processes. Children with metacognitive problems can approach
new material in a relatively concrete fashion, becoming overly involved
with superficial aspects without moving on to the more conceptual level.
Exhibiting inferential reasoning and making connections from prior knowledge
to new material can be difficult.
Another common problem involves rote memory and retrieval of factual
information. Children with this problem may experience greater-than-
expected difficulty in remembering names and facts in social studies and
science classes. In mathematics, retrieval of number facts may lack fluency,
thus compromising speed, efficiency, and accuracy of numeric computation.
Spelling is a related area of vulnerability.
The metacognitive nature of these problems has clinical implications.
First, these children's difficulties will not necessarily manifest as a
deficit in a discrete skill area, and therefore, their formal scores on
psychometric testing may not deviate substantially from those expected
for children their age. Nevertheless, parents may report that their children
are struggling academically because they work inefficiently, reverting
to a deliberate, step-by-step approach to compensate. School personnel
may not be sure that there is a problem because test scores can be relatively
normal. This can be a source of friction as families struggle to support
children who are experiencing unexplained school-related stress.
Second, problems may appear inconsistently. Because the primary issue
involves how children approach a task, or their level of efficiency, there
is no specific reference symptom; rather, there is a general sense of struggle.
Children with these problems may be able to learn a mathematics topic well
enough to pass a test or complete a homework assignment, and yet when the
same material is presented in another context or several months later,
they may appear bewildered. This can be puzzling to parents and teachers,
often raising questions of motivation. The inconsistency may be interpreted
as attentional, and ADHD may be suspected. The apparent inattention or
inconsistency, however, may be secondary to cognitive overload or confusion.
Third, children with these problems often exhibit a characteristic developmental
course. In essence, they may be successful in the early grades but may
begin to falter in the upper elementary grades and middle-school grades.
This does not usually represent a progressive deterioration of function
as a consequence of therapy, although neuropathologic changes may occur
for some years after the completion of therapy. Instead, the demands
and expectations of the upper grades demand greater efficiency, an ability
to digest larger volumes of material and infer concepts, and an ability
to work independently. It is with regard to such expectations that children
treated for ALL can be most vulnerable. Thus, they may appear to develop
problems when, in fact, the problems were always present but the challenges
were not yet sufficient to elicit them.
Finally, the same cognitive issues that affect academic performance
may also manifest in the development of social skills, particularly as
children approach adolescence.[45,46] With development, social interaction
involves processing information from multiple sources, making inferences
about intention that may not be supported by the literal meaning of language,
and grasping subtle cues embedded in the social context. All of these tasks
can be difficult for some children, and problems may be manifested by occasional
withdrawal or isolation, initially more as a function of bewilderment than
of depression. This may occur at the same age at which differences in linear
growth become most apparent, and this constellation can be particularly