Longitudinal change in fine motor skills after brain radiotherapy and in vivo imaging biomarkers associated with decline

Abstract

Background

We explored fine motor skills (FMS) before and after brain radiotherapy (RT), analyzing associations between longitudinal FMS and imaging biomarkers of cortical and white matter (WM) integrity in motor regions of interest (ROIs).

Methods

On a prospective trial, 52 primary brain tumor patients receiving fractionated brain RT underwent volumetric brain MRI, diffusion tensor imaging, and FMS assessments (Delis-Kaplan Executive Function System Trail Making Test Motor Speed [DKEFS-MS], Grooved Pegboard Dominant Hands [PDH], and Grooved Pegboard Nondominant Hands [PNDH]) at baseline and 3-, 6-, and 12-month post-RT. Motor ROIs autosegmented included: sensorimotor cortices and superficial WM, corticospinal tracts, cerebellar cortices and WM, and basal ganglia. Volume (cc) was measured in all ROIs at each timepoint. Diffusion biomarkers (FA [fractional anisotropy] and MD [mean diffusivity]) were additionally measured in WM ROIs. Linear mixed-effects models assessed biomarkers as predictors of FMS scores. P values were corrected for multiple comparisons.

Results

Higher RT dose was associated with right paracentral cortical thinning (β = −2.42 Gy/(month × mm), P = .03) and higher right precentral WM MD (β = 0.69 Gy/(month × µm²/ms), P = .04). Higher left (β = 38.7 points/(month × µm²/ms), P = .004) and right (β = 42.4 points/(month × µm²/ms), P = .01) cerebellar WM MD, left precentral cortical atrophy (β = −8.67 points/(month × mm), P = .02), and reduced right cerebral peduncle FA (β = −0.50 points/month, P = .01) were associated with worse DKEFS-MS performance. Left precentral cortex thinning was associated with worse PDH scores (β = −17.3 points/(month × mm), P = .02). Left (β = −0.87 points/(month × cm³), P = .001) and right (β = −0.64 points/(month × cm³), P = .02) cerebellar cortex, left pons (β = −19.8 points/(month × cm³), P = .02), and right pallidum (β = −10.8 points/(month × cm³), P = .02) atrophy and reduced right internal capsule FA (β = −1.02 points/month, P = .03) were associated with worse PNDH performance.

Conclusions

Biomarkers of microstructural injury in motor-associated brain regions were associated with worse FMS. Dose avoidance in these areas may preserve FMS.

Key Points.

• Fine motor skills (FMS) decline after brain radiotherapy (RT) in older patients.

• Higher RT dose is associated with greater injury to brain motor structures.

• Imaging biomarkers of injury in brain motor structures correlate with post-RT FMS.

Importance of the Study.

Fine motor skills (FMS), including motor coordination and speed, may decline after brain radiotherapy (RT); yet, this domain is often overshadowed by evaluation of higher-level cognitive functions such as memory. FMS are critical for many activities of daily living, and efforts to preserve these skills are crucial for maintaining functional independence and quality of life. No studies have comprehensively investigated microstructural damage to motor-associated brain regions and its relationship to post-RT FMS decline in primary brain tumor patients. We found that FMS declined among older, non-white, and left-handed patients after brain RT, as well as among patients taking anti-seizure medications. We also identified in vivo imaging biomarkers of microstructural damage in the primary motor cortex, corticospinal tracts, cerebellum, and basal ganglia associated with FMS performance. These regions may represent areas for RT avoidance, and these at-risk patient populations may require more intensive pre- and post-RT physical therapy to preserve FMS.

Radiotherapy (RT) is a cornerstone of primary brain tumor treatment; yet, it can be associated with progressive and irreversible neurocognitive decline.¹ Multiple large, randomized trials have documented decline in fine motor skills (FMS), including motor speed and coordination, after brain RT^2,3; however, the focus on impairment in this domain is often overshadowed by cognitive functions such as memory and executive function.⁴

FMS involve the ability to coordinate the eyes and hands to perform precise, manipulative movements. Many activities of daily living, including handwriting, typing, and eating, require a combination of cognitive and motor functions that include speed, dexterity, and coordination. Moreover, FMS can contribute to performance on many of the tests commonly used to assess neurocognition in brain tumor clinical trials.⁵ Strategies to maximize the quality of life and functional independence should therefore focus on improving FMS in addition to higher-level cognitive domains.

In the clinical setting, patients with brain tumors are counseled on the possibility of worsening motor deficits after brain RT if their tumor is located close to or within the primary motor cortex. Yet, motor functioning is subserved by a network of neuroanatomic regions distributed throughout the brain, including the somatosensory cortex,⁶ the corticospinal tracts,⁷ the cerebellum,⁸ the basal ganglia,⁹ and the superficial white matter (WM) adjacent to the primary sensorimotor cortex, which enables communication across neighboring gyri.¹⁰ It is unclear whether microstructural damage to these structures also contributes to post-RT FMS decline.

Volumetric MRI and diffusion tensor imaging (DTI) represent noninvasive, in vivo methods of quantifying WM and cortical injury after brain RT.^11,12 Such injury has been associated with post-RT memory,^13,14 executive functioning,¹⁵ attention and processing speed,¹⁶ and language¹⁷ deterioration; however, no studies to date have explored the microstructural correlates of post-RT FMS changes. We aimed to explore the relationship between FMS and volumetric and DTI biomarkers of injury to regions previously shown to be critical to FMS, namely, the primary sensorimotor cortex and superficial WM, corticospinal tracts, cerebellum, and basal ganglia. We hypothesized that changes in volumetric (ie, atrophy and cortical thinning) and diffusion (ie, reduced WM FA [fractional anisotropy] and higher WM MD [mean diffusivity]) metrics within these structures would correlate with worse FMS following RT independent of tumor location.

Materials and Methods

Study Design

This was a prospective longitudinal observational clinical trial examining neurocognitive functioning in the domains of memory, attention and processing speed, executive functioning, language, and FMS among primary brain tumor patients before and after fractionated, partial brain RT on an IRB-approved (IRB# 131457) clinical trial. High-resolution volumetric and diffusion imaging was also performed prospectively. Patients completed comprehensive neurocognitive evaluation, high-resolution 3D volumetric brain MRI, and DTI at baseline (pre-RT) and 3, 6, and 12 months after RT completion.

Study Participants

We enrolled 63 patients in this prospective clinical trial. All subjects provided written informed consent. Our inclusion criteria were broad and included high-functioning patients (Karnofsky performance status [KPS] ≥70) with any type of primary brain tumor and a life expectancy of ≥1 year to increase the generalizability of our results to any patient presenting for brain RT and living long enough to experience the sequelae of therapy. Additional inclusion criteria included ability to complete neurocognitive assessments in English and age ≥18 years. This specific analysis included 52 primary brain tumor patients with baseline FMS outcomes and imaging performed that followed the trial-specified protocol (Supplementary Figure 1).

FMS Assessment

FMS was evaluated using two robust, validated measurements. Fine motor speed was evaluated with the Delis-Kaplan Executive Function System Trail Making Test Motor Speed¹⁸ (DKEFS-MS) and fine motor coordination/dexterity was evaluated with the Grooved Pegboard¹⁹ test. In the DKEFS-MS test, participants connect a series of circles joined by a dashed line as quickly as possible. Patients’ scores are represented by the time taken to complete the test in seconds. The Grooved Pegboard requires patients to insert metal pegs into slots in sequence with their dominant and nondominant hands as quickly as possible. The score is the time in seconds required to complete the array with the Grooved Pegboard Dominant Hands (PDH) and Grooved Pegboard Nondominant Hands (PNDH). The number of peg drops is also recorded. Higher scores on both tests reflect worse FMS performance. See Supplementary Table 1 for a list of test name abbreviations.

Imaging Acquisition

The imaging acquisition of high-resolution volumetric and diffusion-weighted MRIs for this study have been described in detail elsewhere.^11,20 Briefly, imaging for all patients at each time point were acquired on a 3.0T 750 GE system (GE Healthcare, Milwaukee, WI) equipped with an 8-channel head coil. Sequences selected for the protocol included a 3D volumetric T1-weighted inversion recovery spoiled gradient echo sequence (echo time [TE]/repetition time [TR] = 2.8/6.5 ms; inversion time [TI] = 450 ms; flip angle = 8°; field of view [FOV] = 24 cm), a 3D FLAIR sequence (TE/TR = 125/6000 ms, TI = 1868 ms, FOV = 24 cm, matrix = 256 × 256, slice thickness = 1 mm), and a diffusion-weighted imaging (DWI) sequence using a single-shot pulsed-field gradient spin EPI (echo planar imaging) sequence (TE/TR = 96 ms/17 s; FOV = 24 cm, matrix = 128 × 128 × 48; in-plane resolution 1.87 × 1.875; slice thickness = 2.5 mm; 48 slices) with b = 0, 500, 1500, and 4000 s/mm², with 1, 6, 6, and 15 unique gradient directions for each b value, respectively, and one average for each nonzero b value. Two additional b = 0 volumes were acquired with either forward or reverse phase-encode polarity for use in nonlinear B₀ distortion correction.²¹

Image Processing

Preprocessing of imaging data was completed using in-house algorithms derived in MATLAB. We corrected for anatomical imaging distortions due to gradient nonlinearities,²² as well as diffusion scan spatial distortions caused by susceptibility and eddy currents.²¹ FA and MD maps were derived by fitting the DWI data from b values of 0, 500, and 1500 s/mm² to a tensor. Within each voxel, an ellipsoid defined by three perpendicular axes (eigenvectors) approximated the diffusion process. FA is a unitless expression of the degree of directionality of diffusion that ranges from 0 to 1 and decreases with WM injury.²³ MD is an average of the three eigenvalues expressed in µm²/ms that represents the average mobility of water molecules. MD increases with WM injury.²³ Volume is reported here as cm³.

Region of Interest (ROI) Segmentation

We investigated brain regions strongly linked to FMS. Selected FMS ROIs included the sensorimotor cortex and superficial WM (precentral, postcentral, and paracentral gyri), the corticospinal tracts (including the corona radiata, posterior limb of the internal capsule, crus cerebri, and pontine segments), the cerebellar cortex and WM, and the basal ganglia (caudate, putamen, and pallidum) (Figure 1). The medulla was outside the DWI and volumetric MRI FOV; thus, accurate imaging parameters could not be captured in this region. Cortical, superficial WM, cerebellar, and basal ganglia ROIs were segmented using the automated FreeSurfer processing pipeline (version 5.3; http://surfer.nmr.harvard.edu), available on the Neuroscience Gateway Portal.²⁴ Superficial WM is defined here as the volume of WM up to 5 mm below the surface of cortical ROIs defined by FreeSurfer’s Desikan-Killiany atlas.²⁵ The DWI-derived maps and high-resolution volumetric MRI were co-registered and atlas-based tractography was used to segment the DWI into WM tracts in order to evaluate the corticospinal tracts.²⁶ Volume, FA, and MD were assessed in the corticospinal tracts, superficial WM, and cerebellar WM, while volume alone was assessed in the cerebellar cortex and basal ganglia. Cortical thickness was measured in the sensorimotor cortex. FA and MD values were averaged across each WM ROI. Cortical thickness for each sensorimotor ROI was estimated as the mean thickness across all cortical vertices of that ROI. A censoring mask was drawn manually slice by slice on each image to exclude tissue affected by tumor, surgical cavity or scarring, or edema. The voxels within the censoring mask were excluded from the final ROI to avoid confounding by tumor and edema-related effects.²⁷ Planning CT and RT dose maps were co-registered to the baseline T1 and DWI volumes and used to estimate the average dose to each ROI.

Fig. 1.

FMS ROIs. Fine motor skill ROIs were assessed in this analysis. (a) Precentral cortex, blue; postcentral cortex, orange. (b) Paracentral cortex, magenta. (c) Precentral SWM, green; precentral cortex, blue; postcentral SWM, brown; postcentral cortex, orange; paracentral SWM, yellow; paracentral cortex, magenta. (d) Corona radiata, light blue; posterior limb of the internal capsule, gold; crus cerebri, green; pons, purple. (e) Cerebellar cortex, dark purple; cerebellar white matter, lavender. (f) Caudate, light green; putamen, dark green; pallidum, pink. Abbreviations: FMS, fine motor skills; ROIs, regions of interest; SWM, superficial white matter.

Reliable Change Indices

Reliable change indices adjusted for practice effects (RCI-PEs) were calculated using DKEFS-MS t-scores, corrected for age, sex, and education when appropriate, and raw PDH and PNDH scores. We calculated RCI-PEs for each patient on each neurocognitive test measure between 0-6 and 0-12 months to detect subacute and potential long term changes in FMS over time. RCI-PE calculation is based on the test-retest data of a reference group that has completed the same assessment multiple times.²⁸ A positive DKEFS-MS RCI-PE indicates improved performance over time, while a positive PDH or PNDH RCI-PE indicates diminished performance over time.

Statistical Analysis

One sample t-tests were used to detect significant change in RCI-PEs for each test between 0-6 and 0-12 months (H₀ = 0). Associations between RCI-PEs and patient characteristics were evaluated using nonparametric tests. Covariates investigated included sex, age, race, highest education achieved, baseline KPS, marital status, handedness, tumor histology, tumor laterality, RT modality, total prescription RT, receipt of concurrent and/or adjuvant chemotherapy, history of brain surgery, history of seizures, and anti-seizure medications during the study period (Table 1). These variables were chosen for their potential to independently influence neurocognitive and/or FMS outcomes.²⁹

Table 1.

Baseline Patient, Tumor, and Treatment Characteristics

Characteristic	Patients, n (%)
Median age, y (range)	47 (21-75)
Sex
Male	32 (61.5)
Female	20 (38.5)
Ethnicity
Hispanic	5 (9.62)
Non-Hispanic
White	43 (82.7)
Asian/Pacific Islander	3 (5.77)
Black	1 (1.92)
Highest education achieved (median, range)	16 y (10-20)
High school	12 (21.1)
College	28 (49.1)
Graduate school	16 (28.1)
Handedness
Right	43 (84.3)
Left	8 (15.7)
Tumor diagnosis
Glioma
WHO Grade I-IIa	9 (17.3)
WHO Grade III-IVb	23 (44.2)
Meningioma	12 (23.1)
Pituitary adenoma	3 (5.77)
Otherc	5 (9.62)
Tumor side
Left	28 (53.8)
Right	22 (42.3)
Central	2 (3.85)
Tumor location
Frontal	18 (34.6)
Temporal	12 (23.1)
Suprasellar	7 (13.5)
Parietal	6 (11.5)
Base of skull	4 (7.69)
Cerebellar	2 (3.85)
Cavernous sinus	3 (5.77)
Perimotor tumor locationd
Yes	10 (19.2)
No	42 (80.8)
Radiotherapy type
IMRT/VMAT	39 (75.0)
Proton	13 (25.0)
Median RT dose, Gy (range)	59.4 (50.4-70)
Chemotherapy	32 (61.5)
Concurrent/adjuvant temozolomide	23 (44.2)
Concurrent/adjuvant temozolomide, and othere	5 (9.62)
Adjuvant procarbazine, lomustine, and vincristine	4 (7.69)
Surgery
Gross total resection	9 (17.3)
Subtotal resection	33 (63.5)
Biopsy	3 (5.77)
None	7 (13.5)
History of seizures during the study period	23 (44.2)
Antiepileptic drug use during the study period	29 (55.8)

Abbreviations: IMRT/VMAT, intensity-modulated radiotherapy/volumetric modulated arc therapy; RT, radiotherapy; IDH1, Isocitrate dehydrogenase 1; MGMT, O[6]-methylguanine-DNA methyltransferase; WHO, World Health Organization.

^aWHO grade I-II gliomas included: grade II diffuse astrocytoma (IDH1 wild type [n = 2], IDH1 mutated [n = 1]; MGMT methylated [n = 2], MGMT unmethylated [n = 1]), grade I pilocytic astrocytoma (n = 1), grade II IDH1 mutated, 1p19q codeleted oligodendoglioma (n = 3), grade II anaplastic astrocytoma (IDH1 wild type [n = 1], IDH1 mutated [n = 1]; 1p19q intact [n = 2]).

^bWHO grade III-IV gliomas included: grade IV glioblastoma multiforme (IDH1 wild type [n = 10], IDH1 mutated [n = 1]; MGMT methylated [n = 6], MGMT unmethylated [n = 5]), grade III anaplastic astrocytoma (IDH1 wild type [n = 2], IDH1 mutated [n = 7]; MGMT methylated [n = 6]; 1p19q intact [n = 5]), grade III IDH1 mutated, 1p19q codeleted oligodendroglioma (n = 2).

^cIncludes craniopharyngioma (n = 2), schwannoma (n = 2), and low-grade chondrosarcoma (n = 1).

^dPerimotor defined as tumor within 1 cm of the primary motor cortex.

^eOther chemotherapy included: vaccine clinical trial (n = 1), poly (ADP-ribose) polymerase-inhibitor clinical trial (n = 2), adjuvant lomustine (n = 1), proteasome inhibitor clinical trial (n = 1).

We subsequently investigated the longitudinal relationship between RT dose received and biomarkers of injury within each ROI using linear mixed-effects (LME) analysis. LME models are well suited for longitudinal analysis because they account for within-subject correlation between repeated measures and allow for incomplete outcome data.³⁰ All LME models were fitted using the lme4 package for R.³¹ LME models contained volume, FA, MD, or cortical thickness as the outcome, mean dose and time as main effects, and subject-specific random intercepts. Models estimating volume additionally controlled for the percentage of the ROI that was censored at each time point to control for longitudinal censoring trends in volume measurements.¹⁷ This was not necessary for FA and MD models because FA and MD are average values.

We then evaluated imaging biomarkers as predictors of raw FMS scores over the study period. LME models contained raw FMS scores as the outcome with imaging parameters, time, and perimotor tumor location (defined as within 1 cm of the primary motor cortex) as main effects. Models contained subject-specific random intercepts. Outliers were identified and removed via Mahalanobis distance based on a chi-square distribution (assessed using P < .001).³² Statistical significance was set at α = 0.05 for two-tailed tests. P values were corrected for multiple comparisons using the false discovery rate.³³

Results

Patient Characteristics

Demographic and clinical characteristics are shown in Table 1. All 52 patients included in the final analysis completed baseline neurocognitive assessments and imaging following the trial-specified MRI protocol (Supplementary Figure 1). This cohort was mostly male (61.5%), white (82.7%), highly educated (over 75% of participants achieving a college degree or higher), and right-handed (84.3%).

FMS over Time

At baseline and 6 months, 41 and 39 patients completed DKEFS-MS and PDH/PNDH tests, respectively. At baseline and 12 months, 32 and 33 patients completed DKEFS-MS and PDH/PNDH tests, respectively. There was no group change in DKEFS-MS, PDH, or PNDH RCI-PEs between 0-6 months and 0-12 months.

Older age was associated with declining DKEFS-MS performance between baseline and 6 months (ρ = −0.50, P = .001) (Figure 2). Compared to non-white patients (RCI-PE = −1.13), white patients (mean RCI-PE = 0.38, P = .04) had improved PNDH performance between baseline and 6 months. Patients taking anti-seizure medications experienced decline in PDH (RCI-PE = 0.34) and PNDH (RCI-PE = 0.74) performance between baseline and 6 months compared to those not taking them (PDH and PNDH RCI-PE = −0.72, P < .05). Patients with seizures demonstrated deterioration in PDH performance between baseline and 6 months (RCI-PE = 0.34) compared to patients without seizures (RCI-PE = −0.63, P = .04). Compared to left-handed patients (RCI-PE = 1.51), right-handed patients experienced improved PNDH performance (RCI-PE = −0.05, P = .002) between baseline and 12 months.

Fig. 2.

Association between age and FMS change between baseline and 6-month post-RT. Association between age and baseline to 6-month RCI-PE. DKEFS-MS RCI-PE is negatively correlated with age (ρ = −0.50, P < .001). Negative correlation remains significant when the outlier (red circle) is removed (ρ = −0.43, P = .01). Abbreviations: DKEFS-MS, Delis-Kaplan Executive Function System Trail Making Test Motor Speed; FMS, fine motor skills; RCI-PE, reliable change indices adjusted for practice effects; RT, radiotherapy.

Mean Dose as a Predictor of Injury Biomarkers

Mean dose to each segmented ROI after censoring is shown in Supplementary Table 2. There was no difference in mean dose to left- vs right-sided ROIs. Results of LME modeling assessing mean ROI dose as a predictor of longitudinal volume, FA, MD, and thickness are shown in Supplementary Table 3. Higher mean dose was associated with greater MD and reduced thickness of the right precentral superficial WM (β = 0.69 Gy/(month × µm²/ms), P = .04) and paracentral cortex (β = −2.42 Gy/(month × mm), P = .03), respectively.

Biomarkers of Injury as Predictors of Longitudinal FMS

White matter ROIs

Results from LME analysis of WM ROI injury biomarkers as predictors of FMS are shown in Table 2. Atrophy of the left pontine corticospinal tract segment (β = −19.8 points/(month × cm³), P = .02) and reduced FA in the right posterior limb of the internal capsule (β = −1.02 points/month, P = .03) were associated with worse PNDH scores. Lower FA in the right crus cerebri was associated with worse DKEFS-MS performance (β = −0.50 points/month, P = .01).

Table 2.

Association Between Biomarkers of Injury (Volume, FA, MD) and Fine Motor Skill in White Matter ROIs

	FMS ROI	FMS Test	Volumea		FA		MD
			β Points/ (Month × cm3)	P Value	β Points/Month	P Value	β Points/(Month × µm2/ms)	P Value
Superficial white matter	L postcentral WM	DKEFS-MS	−0.31	0.62	−0.23	0.28	−6.72	0.67
		PDH	0.52	0.68	−0.52	0.23	−13.75	0.66
		PNDH	−1.92	0.12	−0.57	0.21	40.99	0.20
	L precentral WM	DKEFS-MS	0.78	0.08	−0.14	0.48	−7.03	0.62
		PDH	−0.25	0.75	−0.52	0.17	−12.80	0.65
		PNDH	−0.93	0.26	−0.06	0.89	30.66	0.29
	R postcentral WM	DKEFS-MS	−0.66	0.40	−0.17	0.43	−8.08	0.56
		PDH	0.78	0.61	−0.44	0.33	−24.55	0.42
		PNDH	−0.48	0.77	−0.33	0.49	−4.24	0.89
	R precentral WM	DKEFS-MS	−0.13	0.76	−0.28	0.10	−5.15	0.69
		PDH	−0.36	0.64	−0.59	0.10	−16.98	0.52
		PNDH	−0.18	0.84	−0.43	0.25	3.73	0.89
Corticospinal tract	L corona radiata	DKEFS-MS	−2.30	0.18	0.28	0.26	6.46	0.88
		PDH	−2.39	0.45	0.23	0.58	−12.69	0.87
		PNDH	−3.09	0.37	0.33	0.50	137.02	0.11
	L posterior limb of internal capsule	DKEFS-MS	−6.57	0.11	0.01	0.95	22.84	0.48
		PDH	9.71	0.23	−0.20	0.62	88.22	0.17
		PNDH	3.51	0.71	−0.31	0.48	42.70	0.54
	L crus cerebri	DKEFS-MS	−3.43	0.49	0.05	0.84	2.82	0.90
		PDH	6.20	0.52	−0.32	0.49	28.09	0.53
		PNDH	−5.829	0.57	−0.62	0.19	27.14	0.55
	L pons	DKEFS-MS	−2.82	0.52	0.001	0.99	7.38	0.68
		PDH	−4.81	0.57	0.30	0.33	7.89	0.83
		PNDH	−19.82	0.02 b	−0.10	0.74	−0.01	0.99
	R corona radiata	DKEFS-MS	−1.87	0.23	0.46	0.10	50.56	0.23
		PDH	−1.15	0.71	0.05	0.93	8.84	0.91
		PNDH	−3.89	0.22	0.28	0.63	131.22	0.11
	R posterior limb of internal capsule	DKEFS-MS	−3.85	0.38	−0.46	0.04	39.20	0.21
		PDH	11.44	0.18	−0.30	0.48	−8.28	0.89
		PNDH	3.20	0.73	−1.02	0.03 b	66.89	0.29
	R crus cerebri	DKEFS-MS	−6.77	0.13	−0.50	0.01 b	41.80	0.046
		PDH	−3.76	0.66	−0.11	0.79	56.29	0.16
		PNDH	−14.04	0.13	−0.28	0.52	91.61	0.04
	R pons	DKEFS-MS	−3.72	0.37	−0.28	0.12	8.64	0.64
		PDH	−2.17	0.79	0.29	0.42	−10.78	0.77
		PNDH	−9.13	0.28	0.26	0.48	2.82	0.94

Abbreviations: DKEFS-MS, Delis-Kaplan Executive Function System Trail Making Test Motor Speed; FA, fractional anisotropy; FMS, fine motor skills; MD, mean diffusivity, ROI, region of interest, PDH, Grooved Pegboard Dominant Hands; PNDH, Grooved Pegboard Nondominant Hands; WM, white matter.

Bold indicates significant at P < .05 level.

^aModels control for percentage volume censored.

^b P values remained significant after correction for multiple comparisons.

Cerebellum

Results from LME analysis of cerebellar ROI injury biomarkers as predictors of FMS are shown in Table 3. Atrophy of the left (β = −0.87 points/(month × cm³), P = .001) and right (β = −0.64 points/(month × cm³), P = .02) cerebellar cortex was associated with worse PNDH performance, while higher MD in the left (β = 38.7 points/(month × µm²/ms), P = .004) and right (β = 42.4 points/(month × µm²/ms), P = .005) cerebellar WM correlated with poorer DKEFS-MS scores.

Table 3.

Association Between Biomarkers of Injury (Volume, FA, MD) and Fine Motor Skill in Cerebellar ROIs

FMS ROI	FMS Test	Volumea		FA		MD
		β Points/(Month × cm3)	P Value	β Points/Month	P Value	β Points/(Month × µm2/ms)	P Value
L cerebellum WM	DKEFS-MS	−0.93	0.03	0.17	0.38	38.71	0.004 b
	PDH	−0.41	0.59	−0.80	0.03	−11.90	0.65
	PNDH	−1.76	0.05	−0.09	0.82	26.72	0.33
L cerebellum cortex	DKEFS-MS	−0.21	0.12	-	-	-	-
	PDH	−0.51	0.045	-	-	-	-
	PNDH	−0.87	0.001 b	-	-	-	-
R cerebellum WM	DKEFS-MS	−0.53	0.16	0.17	0.45	42.41	0.005 b
	PDH	1.26	0.12	−0.76	0.08	−45.11	0.14
	PNDH	−0.17	0.84	−0.27	0.54	2.66	0.93
R cerebellum cortex	DKEFS-MS	−0.15	0.26	-	-	-	-
	PDH	−0.25	0.34	-	-	-	-
	PNDH	−0.64	0.02 b	-	-	-	-

Abbreviations: DKEFS-MS, Delis-Kaplan Executive Function System Trail Making Test Motor Speed; FA, fractional anisotropy; FMS, fine motor skills; MD, mean diffusivity, ROI, region of interest, PDH, Grooved Pegboard Dominant Hands; PNDH, Grooved Pegboard Nondominant Hands; WM, white matter.

Bold indicates significant at P < .05 level.

^aModels control for percentage volume censored.

^b P values remained significant after correction for multiple comparisons.

Sensorimotor cortices

Results from LME analysis of cortical thickness as a predictor of FMS are shown in Table 4. Left precentral cortex thinning was associated with worse DKEFS-MS (β = −8.67 points/(month × mm), P = .02) and PDH (β = −17.3 points/(month × mm), P = .02) performance.

Table 4.

Association Between Cortical Thickness and Fine Motor Skill in Cortical ROIs

FMS ROI	FMS Test	Cortical Thickness
		β Points/(Month × mm)	P Value
L postcentral cortex	DKEFS-MS	−2.80	0.51
	PDH	−0.57	0.95
	PNDH	−6.34	0.45
L precentral cortex	DKEFS-MS	−8.67	0.02 a
	PDH	−17.35	0.02 a
	PNDH	−10.06	0.18
R postcentral cortex	DKEFS-MS	−2.80	0.51
	PDH	3.42	0.72
	PNDH	−8.21	0.397
R precentral cortex	DKEFS-MS	−8.26	0.04
	PDH	−15.50	0.06
	PNDH	−12.14	0.15

Abbreviations: DKEFS-MS, Delis-Kaplan Executive Function System Trail Making Test Motor Speed; FMS, fine motor skills; ROI, region of interest, PDH, Grooved Pegboard Dominant Hands; PNDH, Grooved Pegboard Nondominant Hands.

Bold indicates significant at P < .05 level.

^a P values remained significant after correction for multiple comparisons.

Basal ganglia

Atrophy of the right pallidum was associated with worse PNDH performance (β = −10.8 points/(month × cm³), P = .02). There were no significant associations between imaging biomarkers of injury and FMS in remaining basal ganglia ROIs.

Discussion

To our knowledge, we present the first comprehensive analysis of longitudinal changes in FMS in primary brain tumor patients receiving RT, including in vivo biomarkers of cortical and WM damage associated with motor speed and coordination decline. Despite FMS’ significance for daily functioning and as a mediator of cognitive performance in other domains, the underlying microstructural brain changes that contribute to post-RT FMS dysfunction have remained unexplored. We found that imaging biomarkers of injury to the primary motor cortex, corticospinal tracts, cerebellum, and basal ganglia were associated with worse FMS among patients in our cohort, as summarized in Supplementary Table 4.

Loss of WM integrity in subregions of the corticospinal tract was associated with poorer FMS. Specifically, atrophy of the left pontine corticospinal tract segment and reduced FA of the right internal capsule and crus cerebri correlated with worse function. Injury to the caudal portions of the corticospinal tract likely produces greater functional deficits than damage to the corona radiata due to the dense concentration of motor fibers within and beyond the internal capsule. Indeed, a study by Maruyama et al. identified a greater risk of motor complications among gamma-knife radiosurgery patients receiving >20 Gy to the internal capsule vs the corona radiata.³⁴ While DTI-based tractography is used intraoperatively to spare the corticospinal tracts,³⁵ this structure is largely ignored in RT planning. Nevertheless, corticospinal tractography has been shown to reduce the risk of motor complications after radiosurgery.³⁶ Our findings highlight the role of damage to the distal regions of the corticospinal tract in post-RT FMS decline, and suggest that informed consent for brain RT should cover the possibility of motor dysfunction with higher doses to this area. Furthermore, dose constraints for the corticospinal tract and subregions may facilitate FMS preservation.

This is the first study to demonstrate the functional consequences of RT-associated cortical thinning measured with volumetric MRI. We found that left precentral, rather than postcentral, cortical thinning correlated with worse dominant hand function in a cohort of mostly right-handed patients. Importantly, this association remained significant after controlling for tumor location within the primary motor cortex. This indicates that, in alignment with neurosurgical planning, radiation avoidance of the dominant primary motor cortex may be particularly important for sparing dominant hand function. RT dose-dependent cortical atrophy has been described previously¹¹; however, the sensorimotor cortex may be less vulnerable to RT dose-mediated thinning compared to other cortical areas.¹² Nevertheless, our findings imply that when sensorimotor cortical atrophy does manifest, it appears to affect FMS.

Cerebellar injury is classically linked to dysmetria and ataxia, but may also affect fine motor coordination and complex attentional skills.³⁷ We found that microstructural injury to the bilateral cerebellar cortex and WM correlated with reduced FMS. Interestingly, left cerebellar cortical atrophy was associated with both dominant and nondominant hand function. Studies exploring cerebellar involvement in motor function have detected contralateral cerebellar activation when performing tasks with the nondominant hand.³⁸ We also found that atrophy of the right pallidum correlated with reduced nondominant hand motor coordination and dexterity. The basal ganglia are implicated in motor control, motor learning, and executive function.⁹ It follows that RT-mediated basal ganglia damage could have important consequences for patient cognitive and motor functioning. Neuro-oncologic RT dosing guidelines³⁹ specify dose constraints for the brain, brainstem, and visual structures; none are currently defined within the cerebellum or basal ganglia. RT avoidance strategies within these structures may aid in maintaining FMS.

Superficial WM injury on quantitative MRI did not correlate with FMS in our cohort. The superficial WM comprises short-range, U-shaped association fibers that form the local WM connections in the brain.¹⁰ Superficial WM damage has been associated with executive function,¹⁵ memory,¹³ and language¹⁷ decline after RT. Functional connectivity within the brain is particularly important for these higher-order cognitive functions, which require synchronous coordination between numerous cortical regions.⁴⁰ RT-mediated superficial WM damage may thus portend greater cognitive, rather than motor, dysfunction among primary brain tumor patients.

In accordance with similar studies of the microstructural sequelae of brain RT,^11,20 we found significant correlations between RT dose and microstructural damage within the right precentral WM and paracentral cortex. RT-mediated WM injury results in demyelination, neuroinflammation, and axonal and vascular damage.⁴¹ These processes alter diffusion properties within WM structures and manifest as elevated MD and reduced FA on DTI. Similarly, imaging evidence of cortical thinning has been noted in brain regions exposed to higher RT doses.¹¹ Of note, higher RT dose was not associated with corticospinal tract WM damage in this study. A previous study found reduced sensitivity to RT dose in the corticospinal tract compared to the corpus callosum, fornix, and cingulum bundle.²⁷ Investigations of WM changes in Alzheimer’s disease have suggested that early myelinating structures, such as the corticospinal tract, are less susceptible to injury compared to those that myelinate later, including the superficial WM.⁴² Moreover, oligodendrocytes within the superficial WM myelinate many axon segments with fewer wraps around each segment,⁴³ thus providing less protection against damage. Despite this, we found that loss of integrity in the corticospinal tract, rather than the superficial WM, portended worse FMS performance. Although the corticospinal tract is less sensitive to RT-mediated damage, our results imply that corticospinal tract injury has important functional consequences when it does occur. Efforts to preserve the structural integrity of the corticospinal tract are thus important for FMS preservation in this patient population.

Older patients in our cohort experienced greater post-RT FMS decline. Prior studies have demonstrated that older age may portend a greater risk of RT-mediated cognitive deficits,⁴⁴ suggesting that older patients may be more vulnerable to RT-mediated injury. This may be due to greater neuroinflammation. For example, greater microglial activation has been observed in older vs younger rats undergoing whole-brain RT.⁴⁵ Additionally, we observed worsening FMS over time among patients taking anti-seizure medications. Cognitive and motor slowing secondary to anti-seizure medication use has been well established in the epilepsy literature.⁴⁶ Right-handed patients, the majority (62.2%) of whom had left-sided tumors, showed significant nondominant hand FMS improvement compared to left-handed patients. This may reflect nondominant hand compensation among right-handed patients in this cohort secondary to tumor- and/or treatment-mediated damage to dominant hand motor structures.⁴⁷ These patient groups may benefit from early interventions with physical or occupational therapy focused on FMS.

This study has several limitations. Motor ROIs were ascertained by FreeSurfer⁴⁸ and AtlasTract,²⁶ auto-segmentation programs designed using reference non-tumor anatomy. To minimize error in ROI definition, segmentations for each patient at each timepoint were meticulously inspected slice by slice to manually censor areas of tumor, surgical cavities, and edema.^12,27 This may have limited our power to detect the effects of higher RT doses in brain motor areas. Though our sample size was small, our cohort was similar in size, if not larger, than other studies assessing imaging and cognitive outcomes among brain tumor patients.^14,49 Moreover, we collected longitudinal data rich in specialized imaging and detailed neurocognitive parameters for each patient. Our cohort was heterogeneous, including patients with both benign tumors and gliomas. Nevertheless, tumor type (ie, glioma vs non-glioma) was not associated with the degree of FMS decline. Furthermore, cognitive functioning is a key outcome for all brain tumor patients, and our findings are generalizable to all primary brain tumor patients undergoing intracranial RT. While chemotherapy may drive neurocognitive decline and contribute to changes in WM integrity,⁵⁰ receipt of chemotherapy did not portend greater FMS decline in our cohort. Most patients received temozolomide chemotherapy, and this study was not powered to analyze the impact of different chemotherapy agents on FMS. Future studies are needed to understand the contribution of chemotherapy in combination with RT on imaging and functional outcomes. Finally, biomarkers of FMS likely differ in left- vs right-handed patients. Our study was not powered to detect specific biomarkers based on handedness; yet, we did identify biomarkers associated with both dominant and nondominant hand function. Future studies with larger sample sizes are needed to delineate motor biomarkers in left- and right-handed individuals. These may enable physicians to execute more personalized treatment planning.

We present the first evidence for imaging biomarkers of microstructural injury which predict for FMS after brain RT. Volume and diffusion metrics within motor-associated structures correlated with both dominant and nondominant hand FMS performance. We also demonstrated associations between RT dose and microstructural damage within the motor cortex and superficial WM. Normal tissue complication probability analyses that explore the functional implications of RT dose to brain motor regions are needed to develop clear dose constraints in these areas. Finally, we identified subsets of patients at increased risk of FMS decline after RT who may benefit most from strategies to preserve FMS. Interventions such as regular physical and occupational therapy as well as dose avoidance in motor structures, when possible, may help spare FMS after brain RT in these patients.

Funding

This work was supported by the National Institutes of Health (1TL1TR001444 to K.T., 1TL1TR001443 to M.S., M.D.T., and A.Y., F31 NS111883-01 to A.R., UL1TR001442 of CTSA funding in support of Clinical and Translational Research Institute (CTRI) at UC San Diego and 1KL2TR001444, UL1TR000100, and R01 CA238783-01 to J.A.H.-G.); National Cancer Institute and UC San Diego Moores Cancer Center (P30 CA02310029 to J.A.H.-G.); and American Cancer Society (RSG-15-229-01-CCE to C.R.M.). The content is solely the responsibility of the authors and does not necessarily represent the official views of any of the funding agencies, had no direct role in designing, conducting, or reporting the study.

Conflict of interest statement. J.A.H.-G. reports grant funding from Varian Medical Systems, unrelated to the present study. C.R.M. has research funding from GE Healthcare, unrelated to the current study. There are no other financial or other relationships that might lead to a perceived conflict of interest.

Authorship statement. Concept and design: M.S., R.K., J.A.H.-G., and C.R.M. Acquisition, analysis, or interpretation of data: all authors. Drafting of the manuscript: all authors. Statistical analysis: M.S. and M.D.T. Obtained funding: J.A.H.-G. and C.R.M. Administrative, technical, or material support: J.A.H.-G. and C.R.M.