Methylene tetrahydrofolate reductase A1298C polymorphisms influence the adult sequelae of chemotherapy in childhood-leukemia survivors

This retrospective correlation study investigated the putative link between methylene tetrahydrofolate reductase (MTHFR) A1298C mutations and chemotherapy-related brain function changes in adult childhood-leukemia survivors. To this end, we determined the relationship between the particular MTHFR1298 genotype (AA, AC or CC) of 31 adult childhood-leukemia survivors, and (1) their CSF Tau and phosphorylated Tau (pTau) levels at the time of treatment, (2) their adult performance intelligence quotient (PIQ), and (3) their regional brain connectivity using diffusion magnetic resonance imaging (dMRI) and resting-state functional MRI (rsfMRI). We confirmed that neuropathology markers Tau and pTau significantly increased in CSF of children after intrathecal methotrexate administration. Highest concentrations of these toxicity markers were found during the induction phase of the therapy. Moreover, CSF concentrations of Tau and pTau during treatment were influenced by the children’s particular MTHFR1298 genotype. CSF Tau (but not pTau) levels significantly dropped after folinic acid supplementation. At adult age (on average 13.1 years since the end of their treatment), their particular MTHFR1298 genotype (AA, AC or CC) influenced the changes in PIQ and cortical connectivity that we found to be related to their childhood exposure to chemotherapeutics. In summary, we suggest that homozygous MTHFR1298CC individuals are more vulnerable to the adult sequelae of antifolate chemotherapy.


Introduction
Acute lymphoblastic leukemia (ALL) and non-Hodgkin lymphoma (NHL) account for one third of pediatric malignancies. Long-term survival has increased markedly after the introduction of central nervous system (CNS) prophylaxis [1]. Developmentally adverse cranial irradiation was successfully replaced by intrathecal and high-dose intravenous methotrexate, an antifolate agent, but the functional sequelae of this treatment continue to be of concern [2]. Executive functioning, and more specifically, processing speed and cognitive flexibility appear to be typically affected in non-irradiated childhood-leukemia survivors [3][4][5][6]. Defects in executive functioning have been linked to defects in white matter integrity, especially in tracts with more prolonged development [7]. Using diffusion tensor imaging (DTI), we previously demonstrated higher fractional anisotropy (FA) and lower orientation dispersion index (ODI) in the left centrum semiovale of childhood-leukemia survivors. Moreover, these structural changes coincided with hypoconnectivity between the inferior temporal gyrus (ITG), and the default mode network (DMN) that is involved in executive functioning [8,9]. Altered functional connectivity of the DMN in non-irradiated childhood-leukemia survivors has been described by others as an adult marker of chemotherapy-induced brain damage [10][11][12].
However, adult sequelae of chemotherapy remain difficult to predict. Younger age [3,6,13,14]and higher doses of intrathecal methotrexate [15,16] have been linked to more adverse outcomes. Also, it has been demonstrated that CSF levels of the microtubule-associated protein Tau-a marker for acute axonal damage-increase after intrathecal methotrexate in the treated children. This increase is especially prominent during the induction phase, during which patients are not supplied with folinic acid (as opposed to later treatment phases) [17]. In addition, CSF levels of pTau were also elevated, possibly signaling even more pervasive brain impact [18]. We were able to link CSF pTau during treatment to adult PIQ, but failed to observe a correlation between intrathecal methotrexate and CSF pTau 14, even though genetic and/or dietary-induced folate disturbances have been shown to induce Tau hyperphosphorylation [19]. This suggests the involvement of intermediary risk factors that modulate the late effects of methotrexate exposure.
In this report, we therefore investigate A1298C mutations in the gene for methylenetetrahydrofolate reductase (MTHFR), one of the main regulatory enzymes of homocysteine metabolism that catalyzes the conversion of 5,10-methylenetetrahydrofolate to 5-methyltetrahydrofolate [19][20][21]. A1298C substitution in the MTHFR gene has been linked to increased risk for executive dysfunction in childhood-leukemia survivors, as opposed to the common C677T mutation [22,23]. We therefore hypothesize that the MTHFR1298 genotype of former patients (AA, AC or CC) might influence some of the previously identified outcomes, shown to be vulnerable to methotrexate exposure, such as (1) Tau and pTau CSF levels during treatment [14,16], (2) adult performance intelligence quotient (PIQ), and (3) functional and structural MRI measures, such as FA and ODI in the left centrum semiovale and resting state functional connectivity (RSFC) between the default mode network (DMN) and the inferior temporal gyrus (ITG) [24].
See Tables 1 and 3 for treatment and demographic data. We could retrace CSF levels in 30 (for Tau) and 26 (for pTau) survivors (see [18], for collection and analysis), but some dossiers were incomplete (see Table 2). The available values were pooled as long as fewer than 2 data points per treatment phase were missing.
The study was approved by the Institutional Ethical Commission of the University Hospital Leuven (UZ Leuven, dossier no. B322201419664). Written informed consent was obtained from all participants (including parents in case of minors). An informed consent form (ICF) for every participant was provided according to the instructions and guidelines of the Commission. These instructions included a template of the form as well as general instructions (e.g., that the form must be drawn up in a language that is clear and understandable to the participant; https://www.uzleuven.be/nl/ethische-commissie-onderzoek/templates-en-internerichtlijnen-bij-starten-van-dossier-bij-ec-onderzoek/informed-consent-formulier-icfopstellen). The ICF is understandable to people who are not health professionals and who are unfamiliar with professional or technical language (i.e., understandable for a 12-year-old). Briefly, the ICF informed participants (in their mother tongue) about the aims of the study, content of the examination, and possible risks of participation and discovery of yet unknown neurostructural findings. Participants received further oral clarification about the contents of the informed consent document, and were particularly informed about the possibility to withdraw from the study, at any time, without any need for justification. The instructions of the UZ Leuven Commission are consistent with the recommendations of the European expert group on clinical trials (implementation of EU Regulation No. 536/2014 about clinical trials on medicinal products for human use; https://ec.europa.eu/health/sites/health/files/files/ eudralex/vol-10/2017_01_26_summaries_of_ct_results_for_laypersons.pdf

2.2.MTHFR genotyping
Obtained from a blood sample in survivors at the time of cognitive testing and image aquisition, DNA segments were amplified by polymerase chain reaction (Promega1 PCR kit) using GoTaq1 reaction buffer, 1.5 mM MgCl 2 , 100 ng of genomic DNA, 0.2 μM of forward and reverse primer, respectively and 1.25 U of Taq polymerase, according to the following protocol: (1) 5-min initial denaturation cycle at 95˚C and (2) 35 cycles each of 30 s at 95˚C, 30 s at 58˚C and 30 s at 72˚C. Primers were based on Krull et al. [23]. Resulting DNA fragments of 138bp were sequenced and genotyped at the respective positions using CLC workbench1 software.

2.3.Cognitive testing and imaging
Adult intelligence was measured using the Wechsler Adult Intelligence Scale (WAIS IV) [25]. MRI scans were acquired at the Leuven University Hospital using a 3T Philips (Achieva) MRI scanner and a 32-channel phased-array head coil. After preprocessing and quality control, the final sample size was 30 for diffusion-weighted magnetic resonance imaging (dMRI) and 28 for resting-state functional MRI (rsfMRI). Image acquisition, selection of imaging outcome measures and image methodology were as previously described [8]. Based on this previous study, higher fractional anisotropy (FA) and lower orientation dispersion index (ODI) in the left centrum semiovale, and lower connectivity between the default mode network and the inferior temporal gyrus (DMN-ITG) were defined as imaging outcome measures.

2.4.Statistical analyses and data availability
Graphs were designed in Graphpad Prism 5. Concentrations of biomarkers were logarithmically transformed. Statistical analyses were performed with SPSS software (v20). Paired sample Table 2. Some of the patient dossiers in this study were incomplete.

MTHFR1298CC MTHFR1298AC MTHFR1298AA
Tau, diagnosis -5 1 Tau, LP2  -2  1 Tau, LP5 -1 - t-tests were used to compare CSF biomarker levels between treatment phases and lumbar punctures. One-way ANOVA was used to assess differences in single or pooled CSF biomarker levels with "genotype" as factor (MTHFR1298AA, -AC or -CC), followed by post-hoc least significant difference (LSD) testing. Repeated-measures two-way ANOVA was used to assess the differences in CSF biomarker levels between participant groups with "genotype" as between factor and "time of sample" as repeated measure, followed by post-hoc least significant difference (LSD) testing.

PLOS ONE
Correlations were calculated as Pearson's r (pooled CSF Tau or pTau, PIQ, FA and ODI at the left centrum semiovale and the DMN-ITG RSFC), whereas correlations including discrete genotypes (AA, AC, CC) were assessed using Spearman's rank correlation coefficients. We used one-sided correlational testing in our sample of former patients that was further subdivided in MTHFR1298 genotype groups. All our hypotheses were indeed distinctly one-sided as it is well established that (1) reduced folate heightens CSF Tau and pTau [17,18,24,25], (2) cognitive performance is reduced in some survivors of childhood leukemia [23,26], (3) MTHFR genotype adversely affects cognitive outcome in child leukemia survivors [22,23], and (4) reduced brain connectivity affects cognitive performance [10].
Raw data that refer to the figures are included as Supplementary Materials. Any further information or data can be obtained from the corresponding author on simple request.

Discussion
We examined the effect of MTHFR1298 mutations on methotrexate-induced increases in CSF neurotoxicity markers in children treated for leukemia, but even more importantly, also on their adult PIQ, and MRI measures related to frontal connectivity and executive functioning. We observed a marked increase in CSF Tau and pTau levels in children after the first intrathecal methotrexate administration. Highest concentrations of these markers were measured during the induction phase, and MTHFR genotype influenced their levels after methotrexate administration and folinic acid supplementation. Intrathecal methotrexate showed the highest impact on CSF toxicity markers in children with MTHFR1298CC genotype compared to MTHFR1298AA, whereas heterozygotes displayed an intermediate response. Moreover, we found indications that the MTHFR genotype still influenced adult PIQ as well as cortical connectivity, many years after treatment.
Methotrexate-induced increases in CSF Tau levels is modulated by the CNS folate status of the children. By inhibiting dihydrofolate reductase, methotrexate depletes the pool of active folates. Besides, methotrexate increases renal folate excretion and inhibits folate uptake into the CSF [27]. The acute rise in CSF Tau after intrathecal methotrexate could be attributed to reduced CSF folate levels during the induction phase [17,18,25,[27][28][29][30]. Despite continued intrathecal and intravenous methotrexate, CSF Tau significantly declined after systemic folinic acid supplementation, which suggests that the supplementation indeed attenuates the toxicity of the anti-folate regimen. This is in line with previous observations by Winick et al. [31], who reported unexpected signs of acute neurotoxicity, including seizures and other neurological symptoms, in 18% of their children receiving intrathecal MTX. They found that folinic acid supplementation indeed prevented these symptoms in 24 subsequently entered patients.
Notably, folate receptor alfa-one of the two pathways by which folates (including folinic acid) and methotrexate enter the cell-has a higher affinity for folinic acid than for methotrexate [32]. Systemic folinic acid might saturate these receptors, diminish uptake of methotrexate and prevent acute toxicity. We observed that, unlike CSF Tau, CSF pTau failed to respond to folinic acid supplementation. It has been argued that the difference in half-life of CSF Tau and pTau cannot account for this discrepancy [18]. The fact that CSF pTau remained elevated after the induction phase (as opposed to CSF Tau) might be related to the increase in CSF homocysteine during later stages of treatment [28]. This finding is consistent with an earlier report Methylenetetrahydrofolate reductase (MTHFR) genotype correlates with differences in adult brain connectivity and performal IQ in childhood-cancer survivors. Panels A-C depict different functional brain connectivity parameters derived from MRI scans, based on our previous study [8], in adult survivors with diverse MTHFR genotype. Fractional anisotropy (FA, panel A), but not orientation dispersion index (ODI, panel B) in the left centrum semiovale, was moderately correlated with MTHFR genotype. Also, functional connectivity between the default mode network (DMN) and the inferior temporal gyrus (ITG) is moderately correlated with MTHFR1298 genotype (C). Finally, performal intelligence quotient (PIQ) is moderately correlated with MTHFR genotype (D). At the top of each figure, r values are 1-sided Spearman's rank correlations (with resp. p values).
Our data are also consistent with the general notion that genetic constitution plays a crucial part in chemotherapy-induced brain toxicity [22,23,26,33,34]. MTHFR1298 genotype influenced the levels of CSF Tau and pTau, which could be at least one of the reasons why we previously failed to find a correlation between intrathecal methotrexate dose and CSF pTau levels [14]. In the present study, MTHFR1298CC homozygotes presented with the highest levels of CSF Tau, followed by heterozygotes. Notably, the latter group demonstrated a comparable response to the first intrathecal methotrexate injection as MTHFR1298CC individuals. This suggests that MTHFR1298 polymorphisms may only become relevant when CSF folate levels are challenged. Indeed, the differences between genotypes disappeared after folinic acid supplementation, also indicating that MTHFR1298CC individuals in particular might benefit from this intervention.
Reportedly, polymorphisms in the MTHFR gene neither affect white matter structure or function, nor intelligence scores, in healthy individuals. However, we propose that certain MTHFR genotypes could be more vulnerable than others to the acute and late effects of MTX exposure. There are several indications that the proposed interaction would be physiologically plausible, consistent with some other reports [22,23]. In the present study, CSF pTau (but not Tau) levels continued to differ between MTHFR1298AA and MTHFR1298CC genotypes during the maintenance phase. Additionally, other potentially neurotoxic compounds, such as homocysteine, have been shown to increase after intrathecal MTX [27,28]. Although the proposed interaction might still be debated [34,35], we suggest that the genetically influenced activity of MTHFR could interact with chemotherapy-induced brain folate depletion, which may enhance Tau phosphorylation and concomitant neural toxicity [36,37].
Furthermore, polymorphisms of the MTHFR gene correlated with defects in functional connectivity in the brain of adult childhood-cancer survivors. We report that individuals with the MTHFR1209CC genotype presented the highest FA at the left centrum semiovale. A drug that affects white matter integrity would increase anisotropy especially in brain regions with many crossing fibers [8,38], such as the centrum semiovale, where corpus callosum, corticospinal tract and superior longitudinal fasciculus (SLF) intersect. Such regions have been shown to be especially vulnerable to the acute effects of chemotherapy [7]. Morioka et al. [7] performed DTI in 17 children with ALL and NHL, before and after chemotherapy, and reported defects in corpus callosum and frontal white matter. Such white matter sensitivity has been demonstrated both in vitro [39] and in vivo [38,40]. A possible explanation for this could be that oligodendrocytes are particularly sensitive because of their high metabolic activity. Additionally, suppressed CNS methylation, resulting from folate alterations, may hinder synthesis of myelin basic protein (MBP), a key component of the myelin sheet [21,41,42]. A healthy myelin sheath not only enhances nerve impulse propagation, it also protects axonal fibers against noxious influences [43]. Therefore, defects in the myelination process renders later developing tracts (such as SLF) especially at risk for chemotherapy-induced injury. Since long associative fibers (e.g., SLF) are crucial role for connecting DMN to other brain areas [44], the correlation between DMN-ITG connectivity and MTHFR1298 polymorphisms is of particular interest. Indeed, we observed that genetically determined, decreased MTHFR activity coincided with lower connectivity between DMN and ITG.