Association of white matter hyperintensities with long‐term EGFR‐TKI treatment and prediction of progression risk

Abstract Purpose The purpose of this study was to test the hypothesis that brain white matter hyperintensities (WMH) are more common in patients receiving epidermal growth factor receptor tyrosine kinase inhibitor (EGFR‐TKI) and identify clinical risk factors associated with WMH. Experimental design This multiple‐center, prospective cohort study was conducted from March 2017 to July 2020. Two groups of patients with non‐small cell lung cancer (NSCLC) who received or did not receive EGFR‐TKI were included and followed up for more than 24 months. The progression of WMH was defined as an increase of ≥1 point on the Fazekas visual rating scale between the baseline and at the 2‐year follow‐up. A modified Poisson regression model was performed to evaluate risk factors on increased WMH load. Results Among 286 patients with NSCLC, 194 (68%) patients with NSCLC who received EGFR‐TKI and 92 (32%) patients with NSCLC without EGFR‐TKI treatment were analyzed. Modified Poisson regression analysis showed that EGFR‐TKI treatment was independently associated with the WMH progression (EGFR‐TKI: aRR 2.72, 95% confidence interval [CI] 1.46–5.06, p = .002). Interleukin (IL)‐2, IL‐4, and IL‐10 were associated with increased WMH in the adjusted model (IL‐2: aRR 1.55 [95% CI 1.06–2.25], p = .023; IL‐4: aRR 1.66 [95% CI 1.13–2.43], p = .010; IL‐10: aRR 1.48 [95% CI 1.06–2.06], p = .020). Conclusion Patients with NSCLC who received EGFR‐TKI may be at higher risk of developing WMH or worsening of WMH burden. The impact of increased WMH lesions in these patients is to be further assessed. IL‐2, IL‐4, and IL‐10 may be used as potential biomarkers to monitor the risk of increased WMH burden


INTRODUCTION
The treatment of non-small cell lung cancer (NSCLC) is "individualized" with targeted drugs.This approach is more effective, has fewer side effects, and is gradually replacing traditional chemotherapy (Ettinger et al., 2016;Wu et al., 2019).Epidermal growth factor receptor (EGFR) mutation is the most common agent used to target the mutation in NSCLC (Holleman et al., 2019;Loong et al., 2018).Therefore, the use of epidermal growth factor receptor tyrosine kinase inhibitor (EGFR-TKI) is one of the research hotspots in lung cancer treatment.
Many studies have found that epidermal growth factor (EGF) promotes cell proliferation, regeneration, neuronal development, and injury repair (Furnari et al., 2015;Goldshmit et al., 2004;I.-S. Kim et al., 2017).Therefore, the inhibition of the EGF pathway or anti-EGFR therapy by EGFR-TKI may have a negative impact on neuronal differentiation, maturation, and recovery.The common adverse effect of anti-EGFR therapy includes skin toxicities (31.4%), diarrhea (14.2%), pruritus (6.7%), and hepatic toxicity (3.8%) (Shah & Shah, 2019).However, studies have also reported that patients with NSCLC treated with long-term EGFR-TKI have developed mental sluggishness, memory deterioration, and cognitive disorder.(Kang et al., 2019;Zhu et al., 2020).These symptoms are related to the development of brain white matter lesions that are insidious.White matter lesions appear as white matter hyperintensities (WMH) on magnetic resonance imaging (MRI) T2 images (Martorell et al., 2012).They are often associated with older age, cerebrovascular diseases, and dementia (Erten-Lyons et al., 2013;Young et al., 2008).Some studies have reported that targeted drugs might induce cognitive and psychiatric problems in patients with NSCLC.However, increased WMH was found in patients treated with EGFR-TKI during our routine follow-ups.In this study, we investigated EGFR-TKI treatment with WMH load and explored the possible mechanisms of increased WMH load in patients with NSCLC receiving EGFR-TKI.

Participants
This study was approved by the Medical Ethics Committee of Tongji

Image acquisition and processing
All participants underwent baseline and follow-up MRI scans.
MRI images from the baseline and second years were analyzed.

Outcome measures
WMH progression was defined as a one-point or greater increase on the Fazekas visual rating scale between baseline and follow-up.

Statistical analysis
We first compared the data distribution of each covariate among the groups, using the t-test (normal distribution) or Kruskal-Wallis rank sum test (non-normal distribution) for continuous variables and chi-square tests for categorical data.The effect values and statistical significance were then assessed using modified Poisson regression without and with adjustments for sex, age, and body mass index (BMI).
Inter-and intrarater reliability for the Fazekas scale was assessed using Dice kappa.All analyses were performed with R (http://www.R-project.org)and the EmpowerStats software (X&Y Solutions, Inc.; www.empowerstats.com).A significance level of 0.05 and a power of 0.9 were used for all statistical tests.

Identification of common pathways and biological processes
The following transcriptome data from the GEO database were used to study (https://www.ncbi.nlm.nih.gov/geo/):PC9 cells were treated for 2 weeks with 300 nM Gefitinib (GSE114647), and periventricular white matter lesions were extracted from frozen human post-mortem CNS tissue (GSE157363) (Fadul et al., 2020;Raoof et al., 2019).Quantile normalization and log2 transformation were employed for data pre-processing with R software.Differentially expressed genes (DEGs) were performed using functions in the limma package (version 3.20.1).
Genes with an adjusted p-value <.05 and |log fold change| (|log FC|) > 1 were assigned as DEGs in both datasets.Gene ontology (GO) biological process and Kyoto encyclopedia of genes and genomes (KEGG) pathway enrichment analysis of these significant DEGs were performed by Metascape (https://metascape.org)with a p-value cutoff of .01, a minimum enrichment of 1.5 and a minimum overlap of 3 (Zhou et al., 2019).The heatmap of overlapping DEGs at the common KEGG pathway was generated using the TBtools software (https://github.com/CJ-Chen/TBtools)(Chen et al., 2020).

Patient characteristics
Between March 1, 2017, and July 30, 2020, we screened a total of 537 patients with NSCLC.Among them, 322 were treated with EGFR-TKI, and 215 received no EGFR-TKI.A total of 199 patients had missing follow-up images or inadequate/missing MRI sequences; 39 had missing baseline imaging and 13 had incomplete clinical information.Hence, statistical analyses were conducted on 286 patients (194 in the EGFR-TKI group and 92 in the non-EGFR-TKI group).Among the 286 patients with NSCLC, the mean age was 59.1 (SD, 9.5) years, and 62% were male.Clinical and demographic comparisons between the groups are presented in Table 1.The inter-and interrater reliability was high (intrarater reliability: 0.92 and 0.87; interrater reliability: 0.83).

WMH differences between groups
Clinical variables were obtained for the risk factor of WMH progres-  2).We reported effect size between laboratory biomarkers and WMH in Table S2).

WMH differences between the EGFR-TKI group and non-EGFR-TKI group
The proportions of raw scores (total Fazekas score, periventricular WMH score, and deep WMH score) for each group are presented in Figure S1, and it was evident that the EGFR-TKI group had higher scores in the total and periventricular than the non-EGFR-TKI group.
Data on the distribution of Fazekas scores change revealed that the EGFR-TKI group had more participants with a progression of two points or more (Table 2).Figure 3 illustrates the spatial distribution of WMH in the non-EGFR-TKI and received EGFR-TKI treatment groups.
WMH in the patient who had received EGFR-TKI involved more of the periventricular and deep white matter.

Possible mechanisms by which EGFR-TKI enhances the risk of WMH progression
We performed GO and KEGG pathway enrichment analysis to determine the relationship between age-related periventricular lesions (GSE157363) and PC9 treated with EGFR-TKI (GSE114647).The top 20 results of enrichment are shown in Figure S2A,B.The KEGG pathway contained nine common pathways including MAPK signaling pathway (hsa04010), oxytocin signaling pathway (hsa04921), cAMP signaling pathway (hsa04024), and so forth (Figure S2A).In addition, the DEGs were involved in nine common biological processes (Figure S2B), such as metal ion homeostasis (GO: 0055065), synaptic signaling (GO: 0099536), and extracellular matrix (GO: 0031012).As shown in the heatmap, genes that were enriched in the MAPK signaling pathway, oxytocin signaling pathway, and cAMP signaling pathway were upreg-ulated in the PC9 treated with EGFR-TKI (Figure S2C).Similarly, genes on the common signaling pathway showed elevated expression in the periventricular lesions dataset.Figure 4 shows the possible mechanism of white matter effects after long-term administration of EGFR-TKIs

DISCUSSION
In our study, there was a clear association between more than 2 years of EGFR-TKI therapy and WMH progression in patients with NSCLC.
After adjusting for sex, age, BMI, hyperlipidemia, hypertension, diabetes, brain metastases, and radiotherapy, the EGFR-TKI group had an almost 2.78-fold higher relative risk of increased WMH load compared to non-EGFR-TKI.The pathophysiological mechanisms of the formation of WMH have not been fully elucidated, and ischemic demyelination, axonal loss, and gliosis can be observed histologically (K.W. Kim et al., 2008).According to the literature, EGF is expressed in the cortical plate during neural development and promotes axonal outgrowth of cortical neurons, and the EGFR pathway is associated with a variety of neuronal cellular events such as proliferation, differentiation, and apoptosis (Goldsh- Several studies have demonstrated the role of Th2 inflammatory markers including IL-4 and IL-10 in intracerebral inflammation.Treatment with anti-IL-4 increased the number of mature oligodendrocytes and myelin proteins (Zanno et al., 2019).IL-10 is considered an antiinflammatory factor, but it also indirectly responds to the severity of neuroinflammation (Sanchez-Molina et al., 2021).IL-2, reported to be toxic to oligodendrocytes and myelin, could play a role in the molecular cascade leading to white matter damage in periventricular leukomalacia (Kadhim et al., 2002).
In patients with brain metastases, radiation therapy is associated with a higher incidence of WMH.Radiation therapy could result in vascular injury.Increased permeability and disruption of the blood-brain TA B L E 3 Risk factor and laboratory biomarker analysis for increased white matter hyperintensities load in subgroup of epidermal growth factor receptor tyrosine kinase inhibitor (EGFR-TKI).Note: aRR adjust for: Age.Abbreviations: CI, confidence interval; DBP, diastolic blood pressure; NA, not applicable; SBP, systolic blood pressure.
barrier (BBB) characterize vascular injury.High-dose focal radiation treatment induces endothelial cell deficiency, leading to vasogenic edema and ischemia via acid sphingomyelinase-dependent apoptosis (Soussain et al., 2009).Hypoxia and the release of reactive oxygen radicals caused by tissue ischemia and vasogenic edema impaired cellular function.Radiation directly damages glial cells, resulting in impaired myelin formation, myelin sparing, reactive gliosis, and even coagulative necrosis and the formation of white matter cavities (Piao et al., 2015).
It is also commonly believed that WMH is caused by inadequate perfusion of small vessels.Many studies suggest that hypertensioninduced changes in large arteries can affect small arteries in a variety of ways, including small artery remodeling and arterial stiffness (Laurent & Boutouyrie, 2015;Stefanadis et al., 1995).In general, rising blood pressure causes arterial injury due to biomechanical fatigue of the elastic arterial wall (Lacolley et al., 2009).Individuals with higher SBP have less cerebrovascular tortuosity and branch number, resulting in white matter ischemia and WMH (Zhang et al., 2021).
Treatment with EGFR-TKI may result in the growth of white matter hyperintensity through the pathways as shown in Figure 4. From the results of enrichment analysis, white matter lesions and EGFR-TKI treatment overlap in multiple signaling pathways.The use of EGFR-TKIs directly affects the MAPK signaling pathway.Studies have shown that MAPK is closely related to brain axonal and white matter remodeling (Fattah et al., 1991;Guo et al., 2020).The use of EGFR-TKI results in decreased expression of tight junction proteins, which can lead to dysfunction of the BBB (Holcmann & Sibilia, 2015).Long-term use of EGFR-TKIs increases VEGF, which affects the stability of the BBB (Viloria-Petit et al., 2001).Increased BBB permeability is associated with the progression of white matter hyperintensity.The release of VEGF also affects the immune microenvironment, causing an increase in inflammatory factors, neuroinflammation, and white matter damage (Proescholdt et al., 2002).As a result, EGFR-TKI administration may have an adverse effect on the white matter brain structure through various pathways.
There were several limitations to this study.First, relatively small sample sizes may limit the reliability of conclusions.Second, there is no analysis of neurological performance.Third, no healthy subjects were included in this retrospective study for comparison.Furthermore, for a one-point increase on the Fazekas scale, the change must be of a certain magnitude, limiting the detectability of change in WMH.Due to the different purposes of the study, the selected transcriptomic data do not represent the actual pathology.Therefore, prospective studies as well as in vitro and in vivo research are required to confirm the risks of anti-EGFR therapy.

CONCLUSION
In patients with NSCLC treated with EFGR-TKI, the risk of developing brain WMH is increased by nearly threefold.Insidious changes in WMH may lead to poor outcomes such as cognitive decline.Furthermore, IL-2, IL-4, and IL-10 may act as independent predictors to increase the WMH load in patients with NSCLC.Lastly, genes enriched in signaling pathways such as MAPK may be related to WMH progression.Future studies are needed to confirm these results.

TA B L E 2
Distribution of Fazekas scores change in white matter hyperintensities progressed patient.

F
The dynamic changes of white matter hyperintensities white matter hyperintensities (WMH).(a) Spatial distribution of baseline WMH in patients who had not received epidermal growth factor receptor tyrosine kinase inhibitor (EGFR-TKI) treatment.(b) WMH in the patient had not received EGFR-TKI at the 2-year follow-up.(c) Baseline WMH in patients had received EGFR-TKI.(d) WMH in the patient had received EGFR-TKI at the 2-year follow-up.T2 Flair images from magnetic resonance imaging (MRI) scans of (e) a 64-year-old man with non-small cell lung cancer (NSCLC) taking EGFR-TKI 1 year before the scan, a total Fazekas score of 3 (2 for periventricular and 1 for deep) and (f) a 61-year-old man has not received EGFR-TKI with NSCLC, no known vascular risk factors, with a total Fazekas score of 1 (1 for periventricular and 0 for deep).
mit et al., 2004;I.-S. Kim et al., 2017).Therefore, long-term EGFR-TKI treatment use may result in brain abnormalities.Previous research has found an average annual increase of 0.6 mL of WMH in people between the ages of 60-70 years.We have found an annual increase of 2.7 mL/year in patients taking EGFR-TKI (deLeeuw et al., 2001).A 7-year follow-up of patients with carotid artery stenosis revealed an increase in Fazekas scale scores in 21% of participants(Ihle-Hansen et al., 2021).In our cohort, WMH progressed in 39 of 106 (36.7%) patients receiving EGFR-TKI.