Screening of cognitive impairment in childhood acute lymphoblastic leukemia survivors with P300 event-related potentials: comparison of different treatment protocols.


 BackgroundModern treatment protocols in childhood acute lymphoblastic leukemia (ALL) resulted in high cure rate and improved long-term survival. However, due to their high intensity, they are also associated with many side effects, including central nervous system toxicity. The aim of our study was to evaluate the use of P300 event-related potentials in screening of cognitive impairment in childhood ALL survivors.MethodsA group of 136 patients, 66 males (48.5%), aged 4.9 to 27.9 years who have completed ALL therapy, were screened for cognitive impairment with endogenous P300 potentials. ALL therapy was conducted according to modified New York (NY) (30 patients) and subsequent revisions of modified Berlin-Frankfurt-Münster (BFM): previous BFM protocols (pBFM) (32 patients) and BFM95 (74 patients). The control group consisted of 58 patients, 34 males (58.6%), aged 6 to 17 years after a syncope episode (n = 29) as well as healthy subjects (n = 29).ResultsThe total group of ALL survivors had significantly prolonged the mean latency of P300 (331.31 vs 298.14 ms, P < 0.001) and reaction time (439.51 vs 380.11 ms, P = 0.002) compared to the control group. Abnormalities in endogenous evoked potentials were observed in 10 (33.33%) NY, 5 (15.63%) pBFM and 21 (28.38%) BFM95 patients. The mean latency time was significantly longer compared to the control group (298.14 ms) in all analyzed protocols and the highest values were observed in pBFM patients (NY: 329.13 ms, P = 0.001; pBFM: 332.97 ms, P < 0.001; BFM95: 331.47 ms, P < 0.001). The reaction time was similarly prolonged compared to the control group. The largest and also significant prolongation was recorded in the NY group (461.8 vs 380.1 ms, P = 0.039).Analyzing the effect of radiotherapy on P300 potentials, a significantly higher frequency of prolonged reaction time in non-irradiated BFM95 patients was found (21.62 vs 15.85%, P=0.007). Radiotherapy methods used in NY and pBFM protocols have also significantly reduced the P300 wave amplitude (mean values: 10.395 vs 12.739 ms, P=0.027).ConclusionsEndogenous P300 event-related potentials may be useful in screening assessment of late cognitive impairment in ALL survivors. The type of treatment protocol significantly modulates the individual parameters of the registered P300 potentials.

The improvement in the overall survival of childhood acute lymphoblastic leukemia (ALL) currently exceeding 80% is considered as a real success in modern pediatric hematology [1]. There are many potential reasons for this contemporary breakthrough. It has been attributed to the introduction of new chemotherapeutic agents, enhanced supportive care and risk-adapted therapy [1]. As survival rates have signi cantly increased, more emphasis has been paid on the long-term side effects of the ALL treatment, also in scienti c research.
As shown repeatedly, ALL therapeutic protocols cause changes in the central nervous system.
Undoubtedly, white matter disturbances induced by demyelination and vascular abnormalities play a role in cognitive impairment caused by radiotherapy [2]. In turn, there is a signi cant gap in the knowledge of central nervous system-related chemotherapy toxicity. Direct destructive effects on cerebral endothelial cells, brain white matter, blood ow and glucose metabolism as well as modi cation of immunological mechanisms could be involved in the development of central nervous system damage [3][4][5].
The above described changes may contribute to the development of cognitive impairment in ALL survivors. Several meta-analyses concerning neuropsychological outcomes after treatment for childhood ALL have been developed so far. All of them unanimously emphasize the heterogeneity of the studied populations and the lack of a uniform cognitive impairment analysis scheme in ALL patients. In a literature review of neuropsychological consequences of ALL chemotherapy approximately two thirds of analyzed studies found declines in different aspects of cognitive functioning [6]. Cousens et al. [7] analyzed 31 studies reporting cognitive function in ALL children after cranial irradiation. Decrements were found in intellectual function, amounting to 10 intelligence quotient points. Campbell et al. [8] performed a complex meta-analysis of 28 studies, those with cranial irradiation, as well as studies in which treatment solely consisted of chemotherapy. As has been shown, ALL survivors represented signi cant de cits in intellectual and neurocognitive functioning which resulted in worse academic achievements. Moreover, ALL patients treated with cranial irradiation performed weaker on intellectual functioning than those who received only intrathecal chemotherapy. In turn, Peterson et al. [9] analyzed neuropsychological results of chemotherapy-only treatment for childhood ALL. Based on 13 articles published until 2004, some evidence for mild ne motor, executive function and verbal memory weaknesses exist in these patients. In addition, the direct relationship between higher levels of methotrexate and executive dysfunction has been recently reported [10]. This was also con rmed by the diverse activity of particular brain regions visualized with structural and functional magnetic resonance imaging [10].
Currently, the American Academy of Pediatrics indicates neuropsychological follow-up as an important element of long-term care for cancer survivors [11]. However, this recommended approach is not without drawbacks. There is still insu cient evidence to guide the speci c timing of comprehensive neuropsychological assessment for ALL children. Moreover, neuropsychometric evaluation can be expensive, time-consuming and provide limited insight into the neurobiological basis of cognitive dysfunction. As a consequence, there is a need for simple functional methods to screen which patients need extensive neuropsychological testing and possible rehabilitation to optimize their learning capabilities and academic achievements. Conventional EEG recordings have not turned out to be useful in predicting late effects of oncological treatment [12] but the endogenous event-related potentials, which detect the neuronal electrical activity associated with cognitive processing, seem to be more promising and encouraging as earlier studies suggested [13,14]. These diagnostic methods are known to give objective information about both attention-dependent and independent central auditory processing with quite simple and inexpensive test arrangements.
The aim of our study was to assess the value of screening of subtle neurocognitive dysfunction with P300 event-related potentials in childhood ALL survivors as well as to compare the observed changes in irradiated and non-irradiated groups of patients.

Methods
A group of 136 patients, 66 males (48.5%), aged 4.9 to 27.9 (average 13.5 ± 5.3) years who have completed ALL therapy, were included in the study. The study group was divided in 3 subgroups according to treatment protocols introduced gradually by Polish Leukemia/Lymphoma Study Group prophylactic and/or therapeutic central nervous system radiotherapy in addition to chemotherapy was used. In turn, 74 children (35 males, 47.3%) were treated with the BFM95 protocol without radiotherapy. Two of these children also received a second-line chemotherapy due to recurrence of the disease. Central The control group consisted of 58 patients, 34 males (58.6%), aged 6-17 years (mean 12.2 ± 3.3 years), who were hospitalized after a single syncope episode (n = 29) and healthy subjects (n = 29) who volunteered for consultation and consented to the examination (Fig. 1).

Methodology of P300 event-related potentials analysis
The auditory evoked P300 potential was performed in accordance with the recommendations of the International Federation of Clinical Neurophysiology (IFCN) [15]. In the study, a method of acoustic stimulation with two contrasting stimuli was used. Each time 60 responses to stimuli different from the background were averaged. Responses were recorded with surface cup electrodes located in the frontal (Fz), central (Cz) and parietal (Pz) zones. The reference electrodes were placed on the earlobes. Each patient underwent three procedures for averaging distinctive stimuli. The attention of the patients was controlled by pressing the counter at the moment of the appearance of the stimulus. To exclude the in uence of body temperature on the conduction speed, the temperature was measured with a validated surface thermometer in each patient. To avoid the impact of emotional factors on the course of the study and the obtained results, all measurements were made in a quiet shaded room after a thorough explanation of the purpose and course of the study.
According to the IFCN recommendations, the P300 potential was assumed as the positive wave with the highest amplitude recorded in the Pz lead, which appeared in the range of 280-500 ms. The latency, amplitude of the P300 wave and response time were evaluated in detail. Prolongation of the latency and the reaction time of the P300 wave above 2SD and a decrease in the amplitude below 1SD from the mean value were assumed as abnormal. To evaluate the effect of treatment, comparisons of ALL patients (NY, pBFM, BFM95) were made with the control group. In turn, to assess the impact of radiotherapy on the obtained P300 parameters, the NY + pBFM group was isolated and compared with the non-irradiated group (BFM95).
The study protocol was complied with the Declaration of Helsinki and was approved by the Jagiellonian University Medical College Ethics Committee (Consent No. KBET/131/B/207). All parents, adolescent patients and adult patients signed written informed consent before inclusion in the study.

Statistical analysis
Statistical analyses were performed with Statistica 12.0 (StatSoft, Statistica 12.0, Tulsa, Oklahoma, USA) software. Continuous variables are expressed as mean ± standard deviation and categorical variables as number (percentage). Continuous variables were rst checked for normal distribution by the Shapiro-Wilk statistic. Differences among two groups were compared by student's t-test when normally distributed or by the Mann-Whitney test with test for non-normally distributed variables. In turn, differences among the three groups were compared by ANOVA test when normally distributed or by the Kruskal-Wallis test with test for multiple comparisons for non-normally distributed variables. Categorical variables were analyzed by the chi-square test and Fisher's exact test depending on the size of the analyzed groups. P-value of less than 0.05 was considered statistically signi cant.

Results
Mean age of children at the time of starting treatment was 5.1 ± 3.2 years. In turn, mean age at the time of screening for cognitive disorders was 13.5 ± 5.3 years. The time that elapsed from the completion of treatment to performed screening ranged from 1.5 to 21.8 years.
Mean age of starting treatment in NY group was 6.5 ± 4.5 years and mean control age − 14.0 ± 5.6 ( Table 3). Children treated with pBFM developed ALL at the earliest (4.4 ± 3.1 years) and were controlled at the latest (18.3 ± 4.0 years). In this group, the average time from onset of the disease to control was therefore the longest. In turn, the difference between the average age of ALL onset (4.9 ± 2.5 years) and the average age of control (11.2 ± 4.0 years) was the shortest in the BFM95 group. However, the statistical analysis did not show any signi cant differences in the mean age of ALL onset. In turn, the mean age of cognitive control was signi cantly different (P < 0.001). Intergroup differences were shown between particular treatment regimens as well as in their direct comparisons with the control group (Table 3).
In groups with or without radiotherapy, the average age of starting treatment was similar, while patients with radiotherapy were signi cantly older at the time of control examination (mean age: 16.3 vs 11.2 years, P < 0.001) ( Table 4).

Analysis of P300 evoked potentials in individual protocols
The total group of ALL survivors had a signi cantly prolonged mean P300 latency (331.31 vs 298.14 ms, P < 0.001) and reaction time (439.51 vs 380.11 ms, P = 0.002) compared to the control group. No differences were observed in the average amplitude of the P300 potentials (11.67 vs 9.64 ms, P = 0.179).
Signi cant changes were found in the screening with endogenous evoked potentials in 10 (33.33%) NY patients. In turn, the results of this study were abnormal in 5 (15.63%) pBFM and 21 (28.38%) BFM95 patients. There was no signi cant difference in the total frequency of their occurrence in individual treatment groups (Table 3). In the NY group, prolonged P300 latency was found in 1 patient and a prolonged reaction time in 10 patients. The incidence of prolonged reaction time was signi cantly higher than in other groups (P = 0.007). In the pBFM group, P300 latency was normal in all patients, while in the BFM95 group latency was abnormal in 4 patients. In contrast, the prolonged reaction time was recorded in 3 pBFM and 16 BFM95 patients. There was no reduction in P300 amplitude in any patient.
Signi cant differences between the individual protocols were observed in all measured parameters characterizing the P300 evoked potentials ( Table 3). The mean latency time was signi cantly longer compared to the control group (298.14 ms) in all analyzed protocols (Fig. 2A). The highest values were observed in pBFM patients (NY: 329.13 ms, P = 0.001; pBFM: 332.97 ms, P < 0.001; BFM95: 331.47 ms, P < 0.001) ( Fig. 2A). At the same time, however, no intergroup differences were found between the protocols analyzed in this study.
The combined analysis of the P300 wave amplitude in individual groups using the Kruskall-Wallis test signaled a signi cant difference (P = 0.036) (Fig. 2B). However, further analysis of the average amplitude values in the pair-comparison test did not reveal differences between individual groups.
The reaction time was similarly prolonged compared to the control group. Its largest and signi cant prolongation was noted in the group treated with NY (461.8 vs 380.1 ms, P = 0.039) (Fig. 2C).

Impact of radiotherapy on P300 potential parameters
Abnormalities in the screening with endogenous evoked potentials were observed in 15 (24.19%) patients treated with NY + pBFM protocols. Analyzing the frequency of individual P300 potential abnormalities, a signi cantly higher frequency of reaction time prolongation was found in non-radiated patients treated with BFM95 (21.62 vs 15.85%, P = 0.007). Despite the lack of a decrease in P300 amplitude meeting adopted criteria in both analyzed groups, a statistical analysis showed a signi cant lowering impact of radiotherapy on the P300 wave amplitude (mean values: 10.395 vs 12.739 ms, P = 0.027) (Fig. 3). No signi cant differences were observed in the other analyzed parameters.

Discussion
According to our best knowledge, the presented data constitute the largest report about the implementation of event-related potentials in childhood ALL population. As we showed in our study, abnormalities in screening assessment of P300 potential were detected in more than a quarter of ALL survivors. Moreover, due to the inclusion of ALL patients treated with different protocols, a signi cant effect of the type of treatment on the nature of neuropsychological disorders in endogenous evoked potentials was observed. The analyzed protocols contribute to the prolongation of latency and reaction time of P300 potential. In turn, the use of regimens quite similar in terms of used chemotherapy regimens but containing radiotherapy reduces its amplitude. These abnormalities can be used to provide a more accurate characterization of subtle neurocognitive dysfunction in childhood ALL survivors.
The endogenous potentials analyzed in current study are the result of changes in the electrical voltage associated with information processing. They do not depend directly on the type of stimulus but on thinking processes and are classi ed as long-latency potentials constituting an electrophysiological indicator of cognitive processes [16]. The P300 potential is determined by the positive wave with the highest amplitude recorded in the central-parietal midline leads in response to the processing of the auditory or visual stimulus. Stimulation with two distinct acoustic stimuli (oddball paradigm) is commonly used. The P300 wave arises after 300-800 ms. This wave occurs when the stimulus recognition is associated with a high level of subjective uncertainty and is a measure of the degree of attention devoted to a particular cognitive task. It is assumed that it is generated in the hippocampus and in the temporal and parietal lobes of the cerebral cortex [16].
Among the greatest advantages of neurophysiological techniques are their high sensitivity, noninvasiveness and the ability to repeat them at relatively low costs. They are an objective although nonspeci c neurological diagnostic tool. Endogenous potentials are widely used in clinical practice, in particular in the diagnosis of oligosymptomatic disease processes, mainly dementia syndromes [17]. Their serial execution also allows to track the dynamics of the disease process and monitor the treatment, therefore they are helpful in determining the prognosis.
However, the usefulness of event-related potentials in the diagnosis and monitoring of adverse effects of childhood ALL treatment has not been su ciently understood yet. As already mentioned, event-related potentials re ect the synchronized post-synaptic potentials generated by the depolarization of neurons, primarily the large pyramidal cells of the cerebral cortex. Its latency is a measure of the time needed for the cognitive processes preceding the cognitive assessment of the task situation, while the wave amplitude de nes the involvement of cognitive structures. The changes observed by us indicate slower and more effortful target detection. Prolonged latency and a reduction in the amplitude of the P300 potential have already been shown in ALL survivors. However, all previous studies have been conducted on small groups of patients. P300 latency has been found to peak later and to have a smaller amplitude in childhood cancer survivors [13,18]. However, a study by Lahteenmaki et al. [13] was performed on a heterogeneous group of only 19 cancer survivors, in which there were 11 patients with ALL. In turn, Uberall et al. included only 13 long-time ALL survivors in their study. Our results are also consistent with the results by Sato et al. [19] who showed a signi cant increase in P300 latency in 33 patients treated with chemotherapy and radiotherapy compared to patients treated with chemotherapy alone and to the control group. Moore et al. also made similar observations on an equally large group of childhood cancer survivors [20]. Järvelä et al. demonstrated the usefulness of P300 potentials in monitoring of central nervous system toxicity of ALL therapy in 27 patients. They showed a relationship between progressive deterioration of mental performance and prolongation of the peak latency as well as poorer enhancement of P300 amplitude after treatment [21]. The previous observations presented above have also been con rmed by recently published preliminary results by Brace et al. [14]. Decreased amplitude of particular P300 components were observed in the analyzed small group of 8 ALL survivors treated exclusively with chemotherapy protocols.
Our study may also have potential therapeutic implications in the future. N-methyl-D-aspartate (NMDA) channels have a central role in the generation of event-related potentials [22]. Differences in particular parameters of P300 potentials between ALL survivors and controls are consistent with altered neurotransmission through NMDA receptors [23]. Recent preclinical study has revealed that memantine, non-competitive NMDA receptor antagonist, reduces the incidence of cognitive de cits in rats treated with intrathecal methotrexate [23]. Memantine has also shown promising effects in randomized trial among adults treated with cranial radiation for brain tumors [24]. Potentially, a group of patients with subtle neurocognitive dysfunction identi ed on the basis of screening with P300 event-related potentials can therefore experience the bene ts of prophylactic use of NMDA antagonists. Such behavior may protect this selected group of patients from the development of symptomatic cognitive impairment. However, large randomized trials using NMDA antagonists in patients with childhood ALL are necessary to con rm this hypothesis.
Our study has several limitations. First, our study compared different protocols previously used in clinical practice. However, it was our deliberate intention. Thanks to this it is possible to study the impact of radiotherapy withdrawn from many protocols currently used in ALL on P300 potentials. Second, neuroimaging and neuropsychological correlations with neurophysiological results were not performed. However, the purpose of our study was only to evaluate the value of cognitive disorders screening. The performed neurophysiological studies informed about maintaining the functional integrity of the nervous system. However, none of the patients exceeded the 5% margin of uncounted discriminating stimuli which indicates the correct concentration of attention. It should also be emphasized that all patients carried out their school duty in an undisturbed manner or they worked and were fully independent after completing their education. Third, genetic methods, which are increasingly used in the diagnosis of cognitive disorders in the pediatric population, have not been used [25,26].

Conclusions
Endogenous P300 event-related potentials may be useful in screening assessment of late cognitive impairment in ALL survivors. The type of treatment protocol signi cantly modulates the individual parameters of the registered P300 potentials. Understanding with the analysis of event-related potentials how ALL survivors brain responses are affected post-treatment will elucidate the type of the cognitive de cits and provide insights into new potential targets for intervention or prevention strategies. The study protocol was complied with the Declaration of Helsinki and was approved by the local Ethics Committee (Consent No. KBET/131/B/207). All parents, adolescent patients and adult patients signed written informed consent before inclusion in the study.

Availability of data and materials
The datasets used and analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.    Figure 1 The study ow chart. Abbreviations: ALL: acute lymphoblastic leukemia, BFM: Berlin-Frankfurt-Münster protocol, NY: New York protocol.

Figure 3
Negative impact of radiotherapy in NY and pBFM protocols on P300 wave amplitude. Abbreviations: BFM: Berlin-Frankfurt-Münster protocol, NY: New York protocol.