Neurodegenerative disease after hematopoietic stem cell transplantation in metachromatic leukodystrophy

Abstract

Objective

Metachromatic leukodystrophy is a lysosomal storage disease caused by deficient arylsulfatase A. It is characterized by progressive demyelination and thus mainly affects the white matter. Hematopoietic stem cell transplantation may stabilize and improve white matter damage, yet some patients deteriorate despite successfully treated leukodystrophy. We hypothesized that post‐treatment decline in metachromatic leukodystrophy might be caused by gray matter pathology.

Methods

Three metachromatic leukodystrophy patients treated with hematopoietic stem cell transplantation with a progressive clinical course despite stable white matter pathology were clinically and radiologically analyzed. Longitudinal volumetric MRI was used to quantify atrophy. We also examined histopathology in three other patients deceased after treatment and compared them with six untreated patients.

Results

The three clinically progressive patients developed cognitive and motor deterioration after transplantation, despite stable mild white matter abnormalities on MRI. Volumetric MRI identified cerebral and thalamus atrophy in these patients, and cerebellar atrophy in two. Histopathology showed that in brain tissue of transplanted patients, arylsulfatase A expressing macrophages were clearly present in the white matter, but absent in the cortex. Arylsulfatase A expression within patient thalamic neurons was lower than in controls, the same was found in transplanted patients.

Interpretation

Neurological deterioration may occur after hematopoietic stem cell transplantation in metachromatic leukodystrophy despite successfully treated leukodystrophy. MRI shows gray matter atrophy, and histological data demonstrate absence of donor cells in gray matter structures. These findings point to a clinically relevant gray matter component of metachromatic leukodystrophy, which does not seem sufficiently affected by transplantation.

Introduction

Metachromatic leukodystrophy (MLD) is an inherited lysosomal storage disorder caused by biallelic pathogenic variants in ARSA, resulting in deficient arylsulfatase A (ASA) activity. This leads to sulfatide accumulation within cells. Oligodendrocytes and Schwann cells are among the most affected cell types, explaining the prominent central and peripheral demyelination of the white matter (WM). ¹ Hematopoietic stem cell transplantation (HSCT) and hematopoietic stem cell gene therapy (HSC‐GT), both aimed at enzyme replacement within the central nervous system (CNS), are effective treatments when performed before symptom onset or early in the disease course. ² , ³ , ⁴ , ⁵ Besides clinical stabilization, HSCT also improves pre‐existent WM changes in some patients. ⁶ , ⁷

Although primarily a WM disorder, MLD also affects the gray matter (GM). Early histopathological descriptions of MLD brain tissue already noted sulfatide accumulation in cortical and deep gray matter (DGM) neurons, ⁸ and recent neuroimaging studies have shown atrophy of cortical and DGM structures already at diagnosis. ⁹ , ¹⁰ , ¹¹ , ¹² These findings suggest neuronal pathology independent of WM involvement. The significance of this neuronal pathology for disease progression, especially after disease modifying treatments as HSCT or HSC‐GT, has not yet been studied.

In our center, we follow a cohort of patients with MLD, including patients treated with HSCT. Three of our long‐term follow‐up patients developed a neurodegenerative disease course despite no evidence of leukodystrophy on their follow‐up MRIs. We hypothesized that possible GM involvement explained this evolution and therefore analyzed available MRI and histopathological data to shed light on this question.

Methods

This cross‐sectional study was carried out at the Amsterdam Leukodystrophy Center and approved by the Institutional Review Board Commission of the VU University Medical Center. Parents gave written informed consent to participate.

Subjects

Three patients (MLD‐4, MLD‐37, and MLD‐45), with early juvenile and late infantile MLD, from an existing MLD cohort at our center were selected because they (1) received HSCT early in the MLD disease course, (2) showed a consistently stable or even improving WM involvement, assessed on MRI, after treatment, and (3) yet developed progressive neurological symptoms and signs in the years following HSCT. Motor function was scored using the gross motor function classification system for MLD (GMFC‐MLD). Cognitive function was scored with different test methods, depending on patient age.

For the histopathological analyses, brain tissue from three transplanted (MLD‐50, MLD‐53, MLD‐64) and six untreated MLD patients was used, as well as tissue from three age matched controls. MLD‐50 and MLD‐53 had died 10 and 12 months after HSCT, respectively; MLD‐64 had died 4 years after transplantation. MLD‐50 and MLD‐53 have previously been published. ¹³

Brain imaging and analysis

To acquire normative volumetric data, sixteen controls from a larger control group used by Steenweg et al. ¹⁴ were included. This group consisted of ten males and six females with an age range of 2.1–14.8 years (mean age ± SD: 7.5 ± 4 years).

MRI data acquisition

Until April 2013, MR images were acquired at 1.5T (until 2007: Siemens Vision, 2007–2013: Siemens Sonata, Erlangen, Germany). Since May 2013, MRI was acquired at 3T (2013–2021: General Electric Signa MR750, Milwaukee, WI, USA, and since 2022: Philips Ingenia Elition X, Philips Medical Systems, Best, Netherlands).

On each scanner, we acquired a 3D T1‐weighted sequence (MPRAGE or FSPGR)—with 1 mm isotropic voxels and 2T‐weighted sequence (1 × 1 mm in‐plane, 4–5 mm slice thickness), as previously described. ¹⁴ , ¹⁵

MRI data analysis

Presence and progression of brain lesions and atrophy were assessed by experienced child neurologists using the MLD MRI severity scoring system. ¹⁶

Brain segmentation was performed on the 3D T1 images using Synthseg. ¹⁷ This segmentation method is robust to the presence of WM inhomogeneities. We extracted the following volumes: thalamus, as previous literature suggests early involvement, ¹¹ , ¹² cerebrum, cerebellum, and CSF. To account for differences in head size, all volumes were shown as percentage of intracranial volume.

Autopsy and tissue staining

Autopsy was performed in three patients: MLD‐50, MLD‐53, and MLD‐64. Tissue samples were acquired from the frontal white matter, frontal cortex, and thalamus. Brain autopsy was performed within 6 hours post‐mortem. ¹³ Formalin‐fixed paraffin‐embedded 5 μm‐thick tissue sections were routinely stained with hematoxylin and eosin (H&E) and Toluidine blue. Staining against ASA was performed using affinity‐purified antibodies targeting human ASA (1:100), as previously described. ¹³ Double stains against ASA and the pan‐neuronal marker NeuN (1:500, ab177487, Abcam) were also performed for one transplanted patient (MLD‐53). Light microscopy pictures were taken with a Leica DM6000B microscope (Leica microsystems). The ImageJ plugin Color Deconvolution ¹⁸ was used to quantify ASA positive pixels. This quantification was performed for all patients where histopathology was performed, except for three untreated patients. Thalamus tissue was only available for two of the untreated patients.

Results

Index subjects

MLD‐4 had a medical history of mild gross motor delay, with difficulties in running first noted at 2 years of age. At the age of 3, running was possible, though parents still found his movement clumsy. He presented at the age of 4.5 years with ataxia and peripheral neuropathy. Cognitive function was normal using the “Wechsler Preschool and Primary Scales of Intelligence—Revised (WPPSI‐R)” (Total Intelligence Quotient (TIQ) = 111). Brain MRI showed subtle signal abnormalities in the deep and periventricular white matter without involvement of the corpus callosum. Diagnosis of early‐juvenile MLD was confirmed genetically and biochemically (Table 1). One month after diagnosis, at the age of 4.9 years, HSCT was performed with HLA‐A and HLA‐B (4/6) mismatched unrelated cord blood from a female donor. Three months after HSCT, he experienced post‐transplant lymphoproliferative disorder (PTLD) in lung and liver due to Epstein–Barr virus (EBV) reactivation, treated with aciclovir, rituximab, two cycles of cyclophosphamide, vincristine, prednisolone (COP), and intrathecal chemotherapy. Annual brain MRI for the first 5 years post‐HSCT showed slight enlargement of the lateral ventricles without evidence of progression of the white matter abnormalities. Repeated cognitive assessment with the Dutch Wechsler Intelligence Scales for Children (WISC‐III‐NL) showed a slow decline of verbal (VIQ) and performance (PIQ) IQ: VIQ = 98, PIQ = 77 (1 year after HSCT); VIQ = 92, PIQ = 67 (2 years after HSCT); VIQ = 85, PIQ = 61 (5 years after HSCT); VIQ = 67, PIQ < 55 (8 years after HSCT); VIQ = 55, PIQ < 55 (10 years after HSCT). Six years after HSCT, he developed partial epileptic seizures that were controlled with levetiracetam, which could be discontinued after 4 years. GMFC‐MLD was 2 for the latest follow‐ups: 13 years after HSCT, walking was only possible with support of one hand; 3 years later only with strong support of both sides. Gait was spastic ataxic, with bilateral foot drop.

Table 1.

Characteristics of the three transplanted patients.

	MLD‐4	MLD‐37	MLD‐45
Sex	Male	Female	Female
MLD type	Early‐Juvenile	Early‐Juvenile	Late‐infantile
Age at
Presentation	4.5 years	Asymptomatic	1.4 years
Diagnosis	4.8 years	2.2 years	1.9 years
ARSA mutations
Allele 1	c.251C>T p.(Pro84Leu)	c.1073T>C p.(Leu358Pro)	c.465+1G>A (r.0)
Allele 2	c.1150G>A p.(Glu384Lys)	c.1283C>T p.(Pro428Leu)	c.830_831delTCinAA (p.Ile277Lys)
Residual ASA activity (in blood leukocytes)	32 nmol/h/mg (55–200)	9 nmol/17 h/mg (81–300)	4 nmol/17 h/mg (>120)
Urinary sulfatide excretion	Elevated	225 nmol/creatinine	N.A.
Clinical symptoms before HSCT	Ataxia and peripheral polyneuropathy	Presymptomatic	Ataxia and peripheral polyneuropathy
Brain MRI before HSCT	Subtle signal abnormalities in the deep and periventricular white matter	Subtle ill‐defined signal abnormalities in the periventricular white matter	Mildly delayed myelination and nonspecific white matter abnormalities
Age at HSCT
First	4.9 years	2.8 years	2.1 years
Second		3.3 years
HSCT conditioning dose
Fludarabine		160 mg/m2 2	160 mg/m2
Cyclophosphamide (targeted)	150–250 μg/L
Busulfan (targeted)	90 mg·h/L	90 mg·h/L 2	90 mg·h/L
ATG	10 mg/kg	Campath 0.9 mg/kg 2	10 mg/kg
GvHD prophylaxis
Prednisone	1 mg/kg	1 mg/kg 2	1 mg/kg
Cyclosporine (targeted)	200–250 mg/L	200–250 mg/L 2	200–250 mg/L
Donor details
Source	Cord blood (unrelated)	Cord blood (unrelated)	Cord blood (unrelated)
Match	4 out of 6	5 out of 6	6 out of 6
Engraftment
Days	27 days post‐HSCT	24 days post‐HSCT	17 days post‐HSCT
Donor chimerism after HSCT* (in blood)
1 month	100%	100%	97%
1 year	98%	100%	97%
Latest follow‐up	97% (12 years)	98% (7 years)	100% (3 years)
ASA activity after HSCT (in blood leukocytes)
2 months	107 nmol/17 h/mg (81–300)	199 nmol/17 h/mg (81–300)	128 nmol/17 h/mg (81–300)
1 year	127 nmol/17 h/mg (81–300)	141 nmol/17 h/mg (81–300)	N.A.
Latest follow‐up	194 nmol/17 h/mg (10 years)	206 nmol/17 h/mg (8 years)	198 nmol/17 h/mg (2 years)
Urinary sulfatide excretion after HSCT at latest follow‐up	140 nmol/creatinine (12 years)	481 nmol/creatinine (15 years)	659 nmol/creatinine (4 years)
HSCT outcomes/notes	Post‐transplant lymphoproliferative disorder (Epstein Barr Virus reactivation)	Graft rejection and retransplant	No complications
Brain MRI after HSCT at latest follow‐up	Slowly progressive global atrophy with stable white matter abnormalities	Slowly progressive cerebral and cerebellar atrophy with stable white matter abnormalities	Severe global atrophy without evidence for progression of her white matter abnormalities
Time since HSCT	13 years	15 years	9 years
Brain tissue available for pathology	No	No

	MLD‐64	MLD‐50 (17)	MLD‐53 (17)
Sex	Female	Male	Female
MLD type	Late‐Juvenile	Late‐infantile	Early‐Juvenile
Age at
Presentation	Asymptomatic	2.0 years	5.5 years
Diagnosis	7.4 years	2.1 years	7.1
ARSA mutations
Allele 1	c.1277C>T p.(Pro426Leu)	c.251C>T p.(Pro84Leu)	c.836_837delTCinsAA p.(Ile279Lys)
Allele 2	c.1130C>T p.(Pro377Leu)	c.1174C>T p.(Arg392Trp)	c.1283C>T p.(Pro428Leu)
Residual ASA activity (in blood leukocytes)	8 nmol/17 h/mg (45–260)	Undetectable (30–90)	1 nmol/17 h/mg (81–262)
Urinary sulfatide excretion	377 nmol/creatinine	N.A.	N.A.
Clinical symptoms before HSCT	Presymptomatic	Peripheral polyneuropathy	Ataxia, mild spasticity and peripheral polyneuropathy
Brain MRI before HSCT	Subtle signal abnormalities in the corpus callosum and periventricular white matter.	Subtle myelination delay	Extensive white matter abnormalities
Age at HSCT
First	7.6 years	2.2	7.2
Second	12.2
HSCT conditioning dose
Fludarabine	160 mg/m2	160 mg/m2 2	160 mg/m2 2
Cyclophosphamide (targeted)
Busulfan (targeted)	90 mg·h/L	90 mg·h/L 2	90 mg·h/L 2
ATG	ATG 5 mg/kg 1 ; Campath 0.6 mg/kg 2	10 mg/kg 2	10 mg/kg 2
GvHD prophylaxis
Prednisone	1 mg/kg 1	1 mg/kg	1 mg/kg
Cyclosporine (targeted)	200‐250 mg/L 1	200–250 mg/L	200–250 mg/L
Tacrolimus	10–15 mg/L 2
Methotrexate	10 mg/m2 2
Donor details
Source	Cord blood (unrelated) 1 ; bone marrow (MUD) 2	Cord blood (unrelated)	Cord blood (unrelated)
Match	6 out of 6 1 ; 10 out of 10 2	5 out of 6	6 out of 6
Engraftment
Days	19 days post‐HSCT 1 ; 20 days post‐HSCT 2	14 days post‐HSCT	17 days post‐HSCT
Donor chimerism after HSCT* (in blood)
1 month	97% 1 ; 99% 2	97%	88%
1 year	100% 1 ; N.A. 2	N.A.	N.A.
Latest follow‐up	100% (3 years)	97% (7 months)	99% (2 months)
ASA activity after HSCT (in blood leukocytes)
2 months	119 nmol/17 h/mg (81–300) 1	87 nmol/17 h/mg (81–300)	113 nmol/17 h/mg (2 months)
1 year	207 (81–300) 1	N.A.	N.A.
Latest follow‐up	190 nmol/17 h/mg (4 years)	87 nmol/17 h/mg (81–300) (7 months)	113 nmol/17 h/mg (2 months)
Urinary sulfatide excretion after HSCT at latest follow‐up	328 nmol/creatinine (3 years)	N.A.	N.A.
HSCT outcomes/notes	Severe GvHD.	No complications. Rapid disease progression after HSCT.	No complications. Rapid disease progression after HSCT.
Brain MRI after HSCT at latest follow‐up	Mild cortical atrophy, subtle improvement of WM abnormalities	Extensive WM abnormalities and severe atrophy	Extensive WM abnormalities and atrophy
Time since HSCT	4 years after first HSCT; Deceased 2 months after second HSCT	Deceased 1 year after HSCT	Deceased 0.8 years after HSCT
Brain tissue available for pathology	Yes	Yes	Yes

ATG was substituted with Campath in the HSCT conditioning regimen of the second HSCT in MLD‐37 and MLD‐64.

ATG, anti‐thymocyte globulin; GvHD, Graft versus Host Disease; HSCT, hematopoietic cell transplantation; HLA, human leukocyte antigen; MRI, magnetic resonance imaging; N.A., not available.

¹ Data on first HSCT.

² Data on second HSCT.

MLD‐37 was diagnosed with MLD at the age of 2.2 years (Table 1), after the diagnosis in an older sibling who presented with motor and cognitive involvement along with extensive WM abnormalities on brain MRI. Neurological examination was normal, and her cognitive function age‐adequate, evaluated with the Bayley Scales of Infant Development (BSID‐II‐NL, Mental Developmental Index (MDI) = 116). Brain MRI showed subtle ill‐defined signal abnormalities in the periventricular white matter. At 2.8 years, HSCT was performed with an HLA‐DR locus (5/6) mismatched unrelated cord blood from a male donor. Post‐transplant course was complicated by graft rejection, leading to a second HSCT at 3.3 years (HLA‐DR locus (5/6) mismatched unrelated cord blood, female donor). After treatment, cognitive assessment with WISC‐III‐NL showed a slow decline of verbal and performance IQ: VIQ = 118, PIQ = 98 (3 years after HSCT); VIQ = 103, PIQ = 86 (5 years after HSCT); VIQ = 89, PIQ = 80 (9 years after HSCT). Walking deteriorated, initially due to progressive peripheral neuropathy and ataxia, but later (starting 5 years after HSCT) due to slowly progressive spasticity. Loss of unsupported walking, corresponding to a progression of the GMFC‐MLD score from 1 to 2, was approximately 8 years after HSCT.

MLD‐45 was born slightly preterm and presented at the age of 1.4 years with ataxia and peripheral neuropathy. She never achieved independent walking. Cognitive function was age‐adequate (BSID‐II‐NL; MDI = 101). Brain MRI showed mildly delayed myelination and subtle nonspecific white matter abnormalities. Diagnosis of late‐infantile MLD was confirmed at age 1.9 years months (Table 1). HSCT was performed at 2.1 years (6/6 matched unrelated cord blood, female donor). Six and 18 months after HSCT, brain MRI did not show disease progression. There was, however, severe progression of peripheral neuropathy hampering her from talking or performing voluntary movements, making a formal cognitive assessment impossible. GMFC‐MLD progressed from a score of 2 to 6 throughout follow‐up. The girl remained alert and interactive. She suddenly died at age 13 years, probably from an acute respiratory problem with an already severely compromised respiratory state from her motor deficits.

Subjects with available brain tissue data

MLD‐50 presented at age 24 months with peripheral polyneuropathy and mildly delayed myelination on brain MRI, and late‐infantile MLD was confirmed. The disease progressed despite uncomplicated HSCT. He died 1 year after HSCT. ¹³

MLD‐53 presented at 5.5 years with mild spasticity and peripheral polyneuropathy. MRI showed extensive white matter abnormalities. Early‐juvenile MLD was confirmed at 7.1 years. The disease rapidly progressed despite uncomplicated HSCT. She died 1 year after diagnosis. ¹³

MLD‐64 was diagnosed at the age of 7.4 years, after MLD had been diagnosed in her older sibling. Neurological examination and cognition were normal (IQ: VIQ = 109, PIQ = 107). EMG indicated a mild demyelinating neuropathy. MRI showed subtle signal abnormalities in the corpus callosum and deep white matter. One month after diagnosis, at age 7.6 years, HSCT was performed with 6/6 matched unrelated cord blood from a male donor. Approximately 6 months after transplantation, graft‐versus‐host disease developed followed by severe hemolytic anemia. Although she received intensive GvHD treatment including a broad range of immunomodulatory drugs, regular transfusions and hospital admissions, yearly neurological examination after transplantation remained normal. Because of refractory hemolytic anemia, a second transplantation with another unrelated male donor was performed 4.6 years after the first HSCT with a 10/10 match. The patient developed severe refractory GvHD and died from pulmonary aspergillosis two months later.

MRI analyses

Assessment of patient MRI scans (Fig. 1) showed little to no WM involvement at 3 years after treatment, and no progression during follow‐up for three patients; MLD‐4, MLD‐37, and MLD‐45 had a stable MLD MRI severity score of 4, 2, and 8 points, respectively. Slowly progressive atrophy of the cerebral hemispheres was seen in all three patients. Quantitative volumetric analyses showed decreasing cerebral volume and increasing total intracranial CSF volume for all three patients over time (Fig. 2A+B). Peripheral and ventricular CSF both showed a similar increase when analyzed separately (not shown). Volume reduction following HSCT in the thalamus was seen in all patients (Fig. 2C). This reduction was progressive. MLD‐4 and MLD‐45 showed a volume reduction of the cerebellum (Fig. 2D).

Figure 1.

MRI evolution in the four transplanted patients. Time after HSCT is indicated in years (y, rounded values). Axial images are all T2‐weighted. Sagittal images are all T1‐weighted, except for the MLD‐45 + 8y scan, which shows a sagittal T2‐weighted image. The reference patient is shown as an example of typical WM lesions in untreated MLD. The top three subjects did not have pronounced WM lesions 3 years after treatment, which remained unchanged in their follow‐up scans. MLD‐64 had white matter lesions 1 year after transplantation and a clear improvement in the follow‐up scans. In all four subjects, enlargement of sulci in all cerebral regions was observed in the follow‐up scans, indicating cerebral atrophy. The +8y images of MLD‐45 show some movement artifacts as sedation was not possible due to severe weakness and respiratory problems. *Time after transplantation for MLD‐64 refers to first transplantation. Latest follow‐up (+5y) was just before the second transplantation. The last row shows the MRI of a patient with classic MLD changes (age 7 years, disease too advanced for HSCT at diagnosis) for comparison.

Figure 2.

Longitudinal Brain volumes as percentage of intracranial volume in the four transplanted patients. Data of each patient are denoted by a different color. Control subjects are shown as individual gray dots (see legend). Vertical lines through the x‐axis shows the age at which HSCT was performed for the corresponding patient (as denoted by the same color). All four subjects show a decline of cerebral and thalamus volume over time (A, C) as well as an increase of total CSF (B). Two subjects show a decline of cerebellar volume (D).

In addition to these three patients, we also studied the MRIs of MLD‐64. White matter abnormalities were present 1 year after transplantation. Those had improved at approximately 5 years post‐treatment; this patient scored 4 points before HSCT and showed a peak of 11 points 1 year after treatment and a subsequent decrease to 6 points at the latest follow‐up. Atrophy of the cerebral hemispheres was less pronounced than in the three index patients. Quantitative volumetric analyses showed similar findings as in the three index patients, although thalamic volume loss was least pronounced in this patient.

Histopathology

In the three treated patients for whom histopathological data were available (MLD‐50, MLD‐53, and MLD‐64), ASA staining showed a strong presence of the enzyme in cortical neurons of controls, but virtually no ASA immunoreactivity in the cortical neurons of both untreated and transplanted MLD patients (Fig. 3). Double staining against ASA and NeuN confirmed strongly decreased intraneuronal ASA in the cortical neurons of a transplanted patient (MLD‐53) compared to control (Fig. 4). In none of the transplanted patients, ASA‐expressing macrophages could be found in the cortical GM. By contrast, ASA‐expressing macrophages were clearly present in the WM of the transplanted MLD patients. WM of untreated patients showed very little ASA positivity. Transplanted patient thalamic tissue showed ASA immunoreactivity in neurons, as well as ASA‐expressing macrophages. Untreated patients also showed some ASA immunoreactivity in thalamic neurons, though less than controls (Fig. 3).

Figure 3.

ASA expression in the frontal white matter, frontal cortex and thalamus of (HSCT treated and untreated) MLD patients and controls. ASA staining shows absence of immunoreactivity in the white matter of an untreated patient (A), while the white matter of a transplanted patient shows clear ASA presence (B). The gray matter of both the transplanted and untreated patient shows nearly no immunoreactivity (D+E). The control subject shows ASA‐presence in both the white and gray matter (C+F). Both untreated and transplanted patients show ASA immunopositivity in the thalamus (G+H), though less than the control (I).

Figure 4.

ASA/NeuN co‐staining in the cortex of a HSCT treated patient and control. ASA/NeuN co‐staining shows virtually absent intraneuronal ASA in the cortex of a transplanted patient (MLD‐53, A) compared to a control (B). CON, control; +HSCT, transplanted patient.

ASA presence in the tissues was also analyzed by quantifying the amount of immunopositive pixels. Quantification of ASA positive pixels showed higher values for transplanted patients than for untreated patients for all tissue types (Fig. 5). ASA positive pixels in the cortical GM were almost absent in MLD‐50, more present in patient MLD‐53 and most in patient MLD‐64. The ASA positive pixels in the cortical GM of MLD‐64 were mostly detected in the deep cortical layers adjacent to the WM. The highest amount of ASA positive pixels in the WM were seen in MLD‐53, the lowest amount in MLD‐50.

Figure 5.

Quantification of ASA positivity per tissue type for patients and controls. The amount of ASA positive pixels is higher for transplanted patients compared to untreated patients in all regions, though consistently lower than control values (A). Average percentages of ASA positive neuron surface area show nearly no immunopositivity for cortical neurons in untreated patients and strongly decreased values for transplanted patients compared to controls (B). Untreated patients show a similar percentage immunopositivity of neuron surface area in the thalamus as treated patients. Error bars show standard deviation. CON, control (n = 2); −HSCT, untreated patients (n = 3); +HSCT = Transplanted patients (n = 3).

Calculations of surface area percentage ASA immunopositivity per neuron showed nearly no immunopositivity for cortical neurons in untreated patients and strongly decreased immunopositivity for the transplanted patients compared to controls. Patient MLD‐50 showed almost no cortical neuron immunopositivity, and patient MLD‐64 showed the most. For thalamic neurons, MLD‐50 showed the most immunopositivity while MLD‐64 showed the least. On average, thalamic neurons showed the same low ASA immunopositivity for transplanted and untreated patients compared to controls.

Discussion

MLD is considered a treatable leukodystrophy, depending on subtype and disease stage, and HSCT is generally expected to give good results when performed before the onset of symptoms or very early in the disease course. ³ , ⁴ That additional awareness regarding expectations and prognosis is needed is illustrated by three MLD patients transplanted early, either before clinical onset or with peripheral neuropathy as the only manifestation. After HSCT, those patients slowly deteriorated with respect to cognitive and motor abilities, while showing stable, mild and nonspecific WM abnormalities, with normal plasma enzyme levels. The absence of progressive WM involvement typical of MLD in these patients argues for a good effect of HSCT on the leukodystrophy, and thereby sufficient ASA levels in the WM. By contrast, the unexpected progressive deterioration, with spastic ataxia and cognitive decline, in combination with slowly progressive brain atrophy, points toward GM degeneration.

Volumetric MR analyses confirmed visually assessed atrophy and showed a gradual decrease of cerebral and thalamus volume with concomitant increase of CSF volume in the three index patients, despite only minor WM abnormalities. Two patients additionally showed cerebellar atrophy. The decrease in cerebral, cerebellar, and thalamus volumes is in line with existing literature demonstrating their involvement in MLD already early in the disease course ¹⁰ , ¹¹ , ¹² and also with previous findings of histopathological GM involvement ⁸ , ¹⁹ ; none of these studies, however, have analyzed these structures in the context of HSCT.

We wondered whether treatment effects on WM and GM were similar and therefore examined these tissues in both transplanted and non‐transplanted MLD patients, respectively. We previously demonstrated a strong presence of ASA‐expressing macrophages in the WM of transplanted patients. ¹³ The current study shows that the cortical GM of transplanted patients contained substantially less ASA staining without ASA‐expressing macrophages. In the patient with the least neurological symptoms and longest interval between HSCT and autopsy, there was some evidence of cortical GM ASA presence, but still restricted mainly to the deep cortical layers adjacent to the WM.

That GM is less amenable to HSCT effects, may be due to the fact that it harbors less microglia than WM, both in the healthy brain ²⁰ and in disease conditions such as multiple sclerosis (MS). In MS, GM lesions typically show substantially fewer innate immune cells than WM lesions, probably due to the relatively anti‐inflammatory environment of GM. ²¹ Myelin debris, a hallmark of WM damage and abundantly present in MLD, is a powerful attractor of microglia and scarcely present in the less myelin‐containing GM. ²¹ This is another possible explanation for the relative absence of (donor) microglia in GM. The poor accessibility of GM to donor macrophages might also explain why HSCT does not have significant clinical effects in GM lysosomal disorders as infantile neuronal ceroid lipofuscinosis and Sanfilippo disease. ²² , ²³

Thalamic involvement on brain imaging has been described in (untreated) MLD. ¹⁰ , ¹¹ , ¹² The current results showed a higher residual ASA staining in thalamic neurons than in cortical neurons for the untreated patients, though still much less than for controls. The relatively high amount of residual ASA in untreated patients is noteworthy. It seems unlikely that this is due to nonspecific binding of the antibody, to neurons for example, as that would have resulted in a similar signal increase in the other analyzed tissues in this patient group. A possible explanation might be intrinsic regional differences of neuronal ASA concentrations within the brain, as controls also showed a higher thalamus neuronal ASA presence compared to the cortex. In transplanted patients, the overall number of ASA positive pixels in the thalamic tissue was higher than in untreated patients (still less than controls), while ASA staining per thalamic neuron was similar to untreated patients. This indicates that transplantation does not seem to increase intraneuronal ASA in the thalamus, and that the overall increase of positive pixels after transplantation is due to more ASA localized outside of neurons. This supports our theory that enzymatic cross‐correction is not the driving therapeutic mechanism of HSCT in MLD. ¹³

These findings illustrate that HSCT probably has different effects on WM and GM. This raises the question whether alternatives to HSCT might be more effective at tackling neuronal sulfatide storage. HSC‐GT, which has already been shown to be an effective MLD treatment, ²⁴ , ²⁵ , ²⁶ has the benefit of higher macrophage ASA expression than HSCT. Yet, this treatment option also relies on the migration of transplanted cells. Possibly, the higher ASA expression in HSC‐GT treated patients leads to higher GM ASA levels compared to allogeneic HSCT and thereby mitigates neuronal sulfatide storage. In another treatment strategy, non‐cellular enzyme replacement therapy (ERT), ²⁷ enzyme is injected directly into the CSF, making enzyme distribution independent from cellular migration. Still, the effectivity of enzyme transport from CSF to specific cell types including neurons is not known. ASA variants combining increased activity and stability have been recently constructed by genetic engineering and proved highly effective in ERT of an MLD mouse model ( ²⁸ ; Matzner et al., unpublished). ERT has also been proved effective in late‐infantile neuronal ceroid lipofuscinosis, a lysosomal GM disorder. ²⁹ Proper posttranslational modification of ASA produced under bioreactor conditions might also facilitate endocytosis into neurons and glial cells, which is unlikely with macrophage‐produced ASA. ³⁰

Limitations

MLD is a rare disease; systematic studies on outcome after different treatment modalities are only just beginning. The availability of homogenous long‐term follow‐up data is restricted. Brain tissue of treated patients to analyses treatment effects is scarce. These challenges constrain systematic analysis of clinical questions. For example, statistical tests would be of limited value when applied to our quantitative analyses due to the paucity of histopathological data. However, the consistency of trends seen even using this limited dataset (e.g., ASA positivity for transplanted versus untreated patients) supports our interpretations.

Other limitations are due to more trivial reasons such as MR data acquisition on 4 different scanners with 2 field strengths in this study spanning more than a decade. This is generally unavoidable when following patients for extended periods of time. These variations affect volumetric analyses, although expressing volumes relative to the intracranial volume mitigates some of the effects.

Medications used for HSCT conditioning and immunosuppression also induce brain atrophy. This treatment‐induced atrophy is expected to occur relatively shortly after treatment and does not persist in the long term. ³¹ , ³² This phenomenon might partially explain the dip or peak in cerebral and CSF volumes, respectively, in the first scans after HSCT for MLD‐45 and MLD‐64. However, the fact that the atrophy seen in the three index patients progresses many years after the cessation of HSCT related drugs and indicates that it is likely caused by a process independent from treatment.

Although our findings illustrate the potential clinical relevance of the GM component of MLD, definitively proving causation remains difficult with the limited available data. Additional research with larger and more homogeneous patient groups are needed to strengthen this hypothesis. Additionally, the three patients with slow clinical deterioration over time had early‐onset (late infantile or early juvenile) MLD. Whether patients with later onset forms develop comparable long‐term problems still needs to be established.

Conclusion

Our findings show that GM involvement in MLD can progress despite stable WM pathology after HSCT, possibly due to the inability of donor macrophages to sufficiently reach the GM. The clinical deterioration of the three MLD patients with long follow‐up after HSCT triggering this study illustrates the potential clinical relevance of GM pathology. Effective MLD treatments might need to combine the benefits of multiple modalities, to target different brain compartments. ¹³ Future studies need to address long‐term outcomes, incorporate larger patient groups, and include studies of brain tissue to further elucidate the exact mechanisms and clinical implications of WM and GM involvement and its response to different treatments.

Conflict of Interest

There were no conflicts of interest. This cross‐sectional study was carried out at the Amsterdam Leukodystrophy Center and approved by the Institutional Review Board Commission of the VU University Medical Center. Parents gave written informed consent to participate.

Author Contributions

Murtadha Al‐Saady: Data collection/analysis/interpretation; writing. Shanice Beerepoot: Data collection/analysis/interpretation; writing. Bonnie C. Plug: Data collection/analysis/interpretation; writing. Marjolein Breur: Data collection/analysis/interpretation; writing. Hristina Galabova: Data collection/analysis. Petra J. W. Pouwels: Supervision; data collection/analysis/interpretation; writing. Jaap‐Jan Boelens: data collection, review & editing. Caroline Lindemans: data collection, review & editing. Peter M. van Hasselt: data collection, review & editing. Ulrich Matzner: Review & editing. Adeline Vanderver: Review & editing. Marianna Bugiani: Supervision; data collection/analysis/interpretation, writing. Marjo S. van der Knaap: supervision; data collection; writing. Nicole I. Wolf: Study design; supervision; data collection/analysis/interpretation; writing.

Funding Statement

This work was funded by Dutch Brain Foundation grant WS2015‐02.