Focal granule cell bilamination of the dentate gyrus—its prevalence across the human age spectrum and review of the literature

Abstract

The prevalence of focal granule cell bilamination (FGCB) in the hippocampal dentate gyrus varies from 0% to 44%, depending on age and study population. FGCB is commonly thought to be a specific feature of temporal lobe epilepsy (TLE) but its prevalence in cases without TLE is unclear. Using formalin-fixed, paraffin-embedded hippocampal sections, this retrospective postmortem study evaluated the prevalence of FGCB and other granule cell pathologies in infants (1-12 months of age, n = 16), children (4-10 years, n = 6), and adults (28-91 years, n = 15) with no known history of epilepsy or seizures. We found FGCB in 6% of infants, 17% of children, and 27% of adults. We then compared our findings with those in published reports of sudden unexpected deaths in infancy (SUDI), childhood (SUDC), and epilepsy (SUDEP), and in surgical specimens from patients with TLE. The reported prevalence of FGCB in those studies was 6%-19% in infants, 0%-17% in children, and 0%-2% in adults in non-seizure-related cases and 9% in children and 3%-25% in adults with TLE. Our findings highlight the presence of FGCB in individuals with no known epilepsy/seizure-related histories in proportions similar to those reported in individuals with clinical epilepsy.

INTRODUCTION

Focal granule cell bilamination (FGCB) or duplication¹ of the granule cell layer (GCL) of the dentate gyrus (DG) is considered to be a variant of granule cell dispersion (GCD),² along with single and/or clustered ectopic granule cells (GCs).³ Historically, FGCB has been identified in surgical specimens from 7% to 35% of patients with Ammons’ horn sclerosis/temporal lobe epilepsy (TLE) (Table 1), and it has been considered to be a marker of seizure activity.^1,4 In more recent postmortem studies, however, the prevalence of FGCB was similar in seizure- and non-seizure-related deaths in both children,^5–7 and adults.^8,9 Consequently, the relationship between FGCB and seizures remains unclear.

Table 1.

Summary of retrospective studies on the prevalence of focal granule cell bilamination (FGCB).

Author and Country (reference no.)	Study protocol	Age range	Clinical context	Case cohort	Number of sections/cases, stain and thickness	Prevalence of FGCB
INFANT
Kinney et al., 2015. United States12	2 NP. Cell maturity taken into consideration.	2-358 d	SIDS	N = 153; (39 = Controls, 114 =uSUDI including SIDS)	1 H&E, 6 µm thick.	Total cohort = 32.7%. [7.7% Controls, 41.2% uSUDI].
Roy et al., 2020. United States6	1 observer. Did not consider cell maturity.	0-365 d (subset of a full pediatric spectrum)	SIDS	N = 91 SUDI cases (60 of these were neonates being <30 d old)	1-3 H&E, 5 µm thick.	13.2%. For 30-365 day age group = 19.4%
Kon et al., 2020. United Kingdom41	Pathology report analysis.a	3-330 d	SIDS	64 SIDS	2-4 H&E.	1.6% SIDS
Machaalani et al., 2022. Australia13	2 observers. Cell maturity taken into consideration.	30-330 d	SIDS	N = 90 (16 = Controls, 74 =uSUDI including SIDS)	3-19. H&E, Cresyl Violet &/or IHC for various markers. 7 µm thick.	Total cohort = 17.8%. [6.3% Controls, 20.3% uSUDI].
CHILD
Rodriguez et al., 2012. Australia42	1 NP	2.5 m-19 years (thus included infants)	Death in pediatrics	N = 25 (22 controls, 3 with seizure history)	1-2 H&E±LFB.	Total cohort = 4%
Hefti et al., 2016, United States27	2 NP	1-7 years	SUDC	N = 92 (23 Controls, 69 SUDC)	1-3 H&E. 6-10µm thick	Total cohort = 36% [13% Controls, 44% SUDC]
McGuone et al., 2020, United States5	2 NP	1-18 years	SUDC	20 SUDC (12 with seizure history)	2, H&E and LFB. Bilateral hippocampi at 3 levels including LGN	Total cohort = 15% [0% SUDC no seizure, 25% seizure history]
Roy et al., 2020, United States6	1 observer	1-19 years	SUDC	N = 56 (42 controls, 14 SUDC with seizure history)	1-3 H&E, 5 µm thick.	Total cohort = 13% [10% Controls, 21% seizure history]
Leitner et al., 2022, United States7	9 pathologists (3xNP, 3xFP, 3x NP/FP).	1-18 years	SUDC	N = 45 (26 Controls, 19 SUDC [9 with seizure history])	1-3 H&E, LGN level	Total cohort = 17% [16.8% Controls, 17% SUDC (regardless of seizure history)]
Gilani & Kleinschmidt-DeMasters, 2020, United States43	2 observers.	3-17 years	Rasmussen encephalitis (RE).	11 cases of RE (Surgical resections).	Average of 6. H&E & IHC for various markers.	9% of RE
ADULT
Houser 1990, United States2	1 NP	18-46 years	TLE	15 TLE (surgical resections), 6 autopsy controls	>10 from body of hippocampus, Cresyl Violet, 30 µm thick	Total cohort = 9.5% [0% Controls, 13.3% TLE]
Lurton et al., 1997, France11	2 observers	21-37 years	TLE	17 TLE (surgical resections), 4 autopsy controls	Several from body of hippocampus, toluidine blue, 10 µm thick	Total cohort = 9.5% [0% Controls, 11.8% TLE]
Thom et al., 2002, United Kingdom4	Not specified	15-58 years	TLE	260 but analysis on 183 TLE cases (surgical resections)	>3, H&E, CV, GFAP & others via IHC	10.3% TLE
Blümcke et al., 2009, Germany3	3 NP	Range not specified, mean age 38.2 ± 13.5 years	TLE	96 TLE (surgical resections)	1 H&E or NeuN. Mid hippocampus body. 4 µm thick	16% TLE
Marucci et al., 2010, Italy44	Not specified	19-54 years	TLE	14 Mesial temporal Sclerosis (surgical resections)	3 H&E, NeuN, Nissl	28.6% TLE
Calderon-Garciduenas et al., 2018, France45	2 NP	Range not specified 38.2 ± 10.9 years	TLE	247 TLE (surgical resections)	>3, H&E, GFAP, NeuN & other markers via IHC	10% TLE
Duarte et al., 2018, Brazil46	2 observers	17-62 years	TLE	77 TLE (surgical resections) & 12 autopsy controls	1× NeuN, mid hippocampus body, 5 µm thick	7.8% TLE Not specified for Controls
Bando et al., 2021, Brazil47	2 Pathologists	18-55 years	TLE	43 TLE (surgical resections)	4× CV, 4 levels of the hippocampus, 60 µm thick.	35% TLE
Jardim et al., 2021, Brazil48	2 observers	11-62 years (thus included children)	TLE	108 TLE (surgical resections) & 12 autopsy controls	1× NeuN, mid hippocampus body, 5 µm thick	6.5% TLE Not specified for Controls
Somani et al., 2019, United Kingdom8	1 observer	2-96 years; however, the control & epilepsy groups were adults (20-96 years)	SUDEP	N = 187 (53 non-epilepsy controls, 68 SUDEP, 66 epilepsy non-SUDEP)	1-4 CV & H&E, 5 levels of hippocampus. Thickness not stated.	Total cohort = 12.3% [1.8% Controls, 16.5% SUDEP, 17.1% Epilepsy]
Leitner et al., 2021, United States9	3 NP	2-66 years (thus included children)	SUDEP	N = 92 cases (61 SUDEP, 31 epilepsy)	1-3 H&E, LFB	Total cohort = 4.3% [3.3% SUDEP, 6.5% Epilepsy]

Abbreviations: d, days; FGCB, focal granule cell bilamination; FP, forensic pathologist; m, months; NP, neuropathologist; uSUDI, unexplained sudden unexpected deaths in infancy; IHC, immunohistochemistry; SUDC, sudden unexpected deaths in childhood; TLE, temporal lobe epilepsy.

^aThis was an autopsy report review to determine the frequency of DG features reported by one of the many pathologists over a time period; thus, microscopic analysis was not undertaken by the authors directly except for a few cases where a report was not found.

The prevalence of FGCB ranges from 0% to 44% with variation depending on the age of the population studied and the presence or absence of a history of seizures (Table 1). In non-seizure-related deaths the prevalence ranges from 6% to 19% in infants (0-12 months), 0%-17% in children (1-19 years), and 0%-2% in adults (18-96 years). However, direct comparisons between these studies are hindered by many methodological issues including the definition of FGCB used, the level or levels of the hippocampi examined, and the number and orientation of sections analyzed. These factors vary between studies and are often undocumented. The lack of specific defining criteria for FGCB results in subjective assessment; 1 study reported only 25% concordance between 2 neuropathologists for identifying FGCB in 20 cases of sudden unexplained death in children (SUDC).⁴ As a consequence, different morphologies of FGCB exist within the literature (Figure S1).

The adult studies mentioned above included surgical specimens from patients with TLE (Table 1), and 2 studies of postmortem specimens from patients with sudden unexpected death in epilepsy (SUDEP).^8,9 FGCB has also been reported in children and adults with autism with duplications of chromosome 15q11.2-q13(dup[15]) syndrome, the majority of whose deaths were diagnosed as SUDEP.¹⁰ However, in that study, the exact frequency was not provided, nor whether it was specific among the cases with a history of epilepsy. Of the 10 adult cohort studies listed in Table 1, only 3 included control cases,^2,8,11 2 of which were of ≤6 cases.^1,11

Therefore, further study to determine the prevalence of FGCB in adults without seizures is warranted. This retrospective study investigates the prevalence of FGCB at different ages in cases without a documented history of seizures or epilepsy. We also evaluated additional morphological features in the DG that have previously been associated with FGCB, including ectopic cells in the molecular layer, gaps, and thinning,^12,13 as well as the location of these features within the DG (ie, internal limb, medial part, or external limb), in an attempt to determine whether the features are a result of altered cell generation,¹⁴ maturation,¹⁵ or migration.^14,16

METHODS

Tissue acquisition

This retrospective study used archived formalin-fixed and paraffin-embedded pediatric and adult hippocampus sections that had been cut and stained for analysis in previous studies from our laboratory.^16–23 Thus, case selection was purposive for those previous studies, with inclusion criteria including a defined cause of death, no evidence of brain trauma or injury at autopsy, and availability of tissue from the brain regions of interest (brainstem medulla, hippocampus, and hypothalamus). For the present study, further inclusion criteria included the availability of 3 or more serial coronal sections from the body of the hippocampus that were stained, and that there was no history of a seizure or epilepsy in the clinical history extracted from the autopsy records for the pediatric cases or provided to the NSW Tissue Resource Centre (University of Sydney, Sydney, NSW, Australia), for the adult cases.

Sections of the pediatric hippocampus were cut from archival blocks obtained from NSW Health Pathology (Forensic and Analytical Science Service; formerly Department of Forensic Medicine, Sydney, NSW, Australia) from a cohort of infants (1-10 months; n = 16) and children (4-10 years; n = 6) who died between 1997 and 2012. Ethics approval came from the Human Ethics Committees of the University of Sydney (approval no. 3013/235) and the NSW Central Sydney Area Health Service (Protocol No 3593 and X13-0038). Hippocampal sections from a cohort of adults (28-91 years; n = 15) who died between 1998 and 2004, were obtained from the NSW Tissue Resource Centre. Given that no data currently exist associating dementia with FGCB and that the incidence of GCD does not differ between cases with dementia and non-demented controls,²⁴ cases with a history of dementia were not excluded. Tissue had been fixed in either 10% (cases prior to 2002) or 20% neutral buffered formalin and paraffin embedded. Six- to 7-µm serial sections were cut using a rotary microtome and mounted on 2% 3-aminopronopyltriethoxysilane-treated slides, dried overnight, and stored in a dust-free environment prior to staining with hematoxylin and eosin (H&E), or Cresyl violet, immunohistochemistry (IHC) counterstained with hematoxylin,^16,18–23 or non-radioactive in situ hybridization (ISH).¹⁷

Qualitative histological analysis

Details of analysis methods were recently described.¹³ Due to the retrospective nature of the present study, differently stained sections were analyzed (Table S1). Sections were analyzed by light microscopy (Olympus Upright BX51 Microscope, Olympus Optical Co., Tokyo, Japan) using 4×, 10×, and 20× objective lens; images were captured using an Olympus DP70 camera with the DP controller software (DP Controller, Olympus Optical Co.).

Sections from infant and child cases were analyzed by 2 observers (R.M. and A.V.), blinded to age group; 1 observer (R.M.) analyzed the adult sections unblinded to age group.

For each tissue section, morphological features were mapped on a hand-drawn sketch of the DG, and findings were localized to the internal limb, the external limb, or the intervening medial zone.^5,25 Features of the DG were noted only after ensuring that the GCs had similar levels of maturity (relevant for the infant cases) as determined by their size and shape. FGCB was documented using the definition, “A linear row of at least 8 ectopic GCs that is separated from the main DG in the supragranular (molecular layer) or subgranular (hilus/CA4 region) sites. There is an acellular, eosinophilic zone of neuropil between the line of ectopic cells and the main DG. The linear row may be one or more cells thick. There may be pyknotic nuclei in the subgranular zone, which may represent apoptosis” (Supplementary section¹²). The following additional DG morphological features were analyzed: single or clusters of ectopic GCs, GCD, gaps, thinning, hyperconvolution, heterotopia, blood vessel (BV) disruption, and BV cuffing.^12,13 The features of FGCB, ectopic GCs and GCD are considered to represent “cell gain” and classified as GC pathology (GCP) Type 2; thinning and gaps represent “cell loss” and were classified as GCP Type 1.⁵

Quantitative analysis of FGCB

For sections with FGCB, using the 20× eyepiece objective lens, the following quantitative measures were undertaken using the measuring scale tool of the DP controller software: FGCB thickness (measured as the distance from the external edge of the GCL row in the molecular layer to the internal edge of the GCL row in the hilus) (green line, Figure 1), distance of the gap between the 2 GCL rows (blue line, Figure 1), and thickness of each of the 2 GCL rows (red lines, Figure 1), with the continuing GCL being referred to as “main DG”¹² and the secondary non-intact row as “bilaminar row.” Measurements were taken at 3 different points along the FGCB per section.

Figure 1.

Representative micrographs of focal granule cell bilamination (FGCB) in which it was present (1/16 infants, 1/6 children, 4/15 adults [numbers 1-4]). For adult case number 1, the bilaminar layer was located in the hilus, whereas in all other cases, it was in the molecular layer (ML). Infant section stained by immunohistochemistry (IHC) for TUNEL, child by H&E, adult number 1 IHC for ALZ50, adult number 2 in situ hybridization for N-methyl-D-aspartate receptor 1 (NR1), adult number 3 IHC for nicotinic acetylcholine receptor subunit beta 2, adult number 4 IHC for NeuN (although staining was hampered as reported in²³). Examples of the quantitative measures for FGCB thickness (green lines), gap between the layers (blue lines), and thickness of each layer (red lines), are provided for 2 cases.

Statistical analyses

Each feature was scored as present (“1”) or absent (“0”), and the results were entered into Excel and exported to SPSS for Windows (V24, SPSS, Inc., Chicago, IL, United States) for analysis. The prevalence of each feature was reported as a percentage per age group; differences between the age groups were examined using chi square with the Fisher exact test. For quantitative measures related to FGCB thickness, the Student t-test was used to compare between pediatric and adult groups. A P-value of <.05 was taken as statistically significant.

RESULTS

Case inclusion

The majority of the cases were males with 63% in the infant, 100% in the child, and 64% in the adult, age groups. The male selection bias in the child group was due to that bias in our previous study of these cases.²⁶ The causes of death are listed in Table S1.

A mean of 11 (range 4-19) tissue sections per case were studied; for 56% of the cases, sectioning occurred at 2 time points (Table S1). For those cases, it is estimated that the 2 sets of serial sections were separated by ∼20-30 µm. For the majority of cases (81%), the sections were at the level of the lateral geniculate nucleus; sections from only 1 hippocampus hemisphere (93% left) were available for study (Table S1); thus, bilateral comparisons were not possible.

Prevalence of the DG features

Thinning, GCD, hyperconvolution, and BV disruption were each identified in ≥70% of cases independent of age, while BV cuffing was only identified in a single infant (Table 2). FGCB and clusters of ectopic cells were infrequent, being seen in 1 infant (6%), 1 child (17%), and 4 (27% FGCB) and 3 (20% cluster ectopics) adults (Table 2). Table S1 shows that both features were not necessarily seen in the same individual. There were no statistical differences in the prevalence of individual features at different ages, although there was a trend for gaps to be more common in children and adults compared to infants (P = .08, Table 2).

Table 2.

Percentage (%) distribution of dentate gyrus (DG) features for all cases and stratified among age groups.

	All Cases (n = 37), Sections (n = 397)	Infants (n = 16), Sections (n = 174)	Child (n = 6), Sections (n = 44)	Adults (n = 15), Sections (n = 179)	P-value across 3 groups
GCP Type 1
Gaps	35	19	50	47	.08
Thinning	76	75	67	80	.75
GCP Type 2
FGCB	16	6	17	27	.13
Single ectopic cells	62	56	67	67	.55
Cluster of ectopic cells	14	6	17	20	.25
GCD	84	94	67	80	.45
Other
Hyperconvolution	73	75	83	67	.92
Heterotopia	27	25	33	27	.91
BV disruption	86	75	100	93	.14
BV cuffing	3	6	0	0	.30

Abbreviations: BV, blood vessel; FGCB, focal granule cell bilamination; GCD, granule cell dispersion; GCP, granule cell pathology.

P-values are Fisher exact.

When all cases were combined, there was a moderate positive correlation between the presence of FGCB and clusters of ectopic GCs (r = .470, P = .003), between single ectopic GCs and BV disruption (r = .344, P = .04) and between single ectopic GCs and gaps (r = .341, P = .04) (Table S2A). However, when pediatric (infant and children) and adult cases were considered separately, a significant correlation between FGCB and cluster ectopics, and single ectopics and gaps was only seen in the pediatric group (Table S2B vs S2C).

Single and clusters of ectopic GCs were more common in the internal limb of the DG in all age groups (Table 3). Other features showed no consistent localization (Table 3; Table S1).

Table 3.

Prevalence (in %) of the dentate gyrus (DG) features for each age group according to the region of the DG.

		Infant				Child				Adult
DG feature	n (% out of n = 16)	Internal limb	Medial	External limb	n (% out of n = 6)	Internal limb	Medial	External limb	n (% out of n = 15)	Internal limb	Medial	External limb
GCP Type 1
Gaps	3 (19%)	2 (67%)	1 (33%)	0	3 (50%)	0	3 (100%)	0	7 (47%)	2 (29%)	4 (57%)	2 (29%)
Thinning	12 (75%)	6 (50%)	9 (75%)	5 (42%)	4 (67%)	2 (50%)	2 (50%)	1 (25%)	12 (80%)	6 (50%)	8 (67%)	4 (33%)
GCP Type 2
FGCB	1 (6%)	1(100%)	0	0	1 (17%)	1 (100%)	0	0	4 (27%)	3 (75%)	1 (25%)	0
Single ectopic cells	9 (56%)	7 (78%)	5 (56%)	1 (11%)	4 (67%)	2 (50%)	2 (50%)	0	10 (67%)	9 (90%)	3 (30%)	3 (30%)
Cluster ectopic cells	1 (6%)	1 (100%)	0	0	1 (17%)	1 (100%)	0	0	3 (20%)	2 (67%)	1 (33%)	0
GCD	15 (94%)	13 (87%)	13 (87%)	9 (60%)	4 (67%)	0	3 (75%)	1 (33%)	12 (80%)	9 (75%)	8 (67%)	5 (42%)
Other
Hyperconvolution	12 (75%)	12 (100%)	1 (8%)	2 (17%)	5 (83%)	2 (40%)	3 (60%)	0	10 (67%)	9 (90%)	1 (10%)	1 (10%)
Heterotopia	4 (25%)	2 (50%)	3 (75%)	0	2 (33%)	0	2 (100%)	0	4 (27%)	2 (50%)	1 (25%)	1 (25%)
BV disruption	12 (75%)	8 (67%)	8 (67%)	1 (8%)	6 (100%)	3 (50%)	5 (83%)	0	14 (93%)	3 (21%)	9 (64%)	3 (21%)
BV cuffing	1 (6%)	0	1 (100%)	0	0 (0%)	0	0	0	0 (0%)	0	0	0

Abbreviations: BV, blood vessel; FGCB, focal granule cell bilamination; GCD, granule cell dispersion; GCP, granule cell pathology.

Focal GC bilamination

In the 6 cases with FGCB, the length of the cell row, the distance between the bilaminar rows, and individual cell shapes varied among the cases (Figure 1). Other than the child case, row length exceeded the minimum of 8 continuous cells as per definition. In the infant case, FGCB was confined to the internal limb (Figure 1; Table 3) and was identified in 5 of 8 (63%) of serial sections (extending approximately 35 µm anteroposteriorly). The size and shape (round) of the cells in the 2 rows were similar, although ovoid-shaped cells of the bilaminar layer closest to the hilus were additionally seen (Figure 1). In the child case, FGCB was confined to the internal limb in 1/5 (20%) serial sections, but not identified in a second set of 5 serial sections taken 20-30µm from the original set, nor in single sections from 3 additional blocks of this hippocampus (Figure 2). The cell shapes (round) and sizes were similar between the 2 layers (Figure 2). The DG of this particular case was irregular, not having the typical “C”-shaped structure. The FGCB was evident on a second curvature formed in the internal limb (Figure 2B, B1a) while the first curvature had single ectopic cells present (Figure 2B1b). The single ectopic cells on the first curvature were also noted in a similar position in 3 additional anteroposterior blocks (Figure 2A1-D1).

Figure 2.

Child case with focal granule cell bilamination (FGCB) found in 1 out of 4 hippocampal blocks represented as A-D in the anteroposterior plane (distance between the tissue sections unknown). Block B had 2 sets of 5 serial sections; FGCB was found in 1 section of the first set in the second curvature formed at the internal limb (B1a, red equal sign). A1-D1 shows magnification of the internal limb with the main curvature having the presence of single ectopic cells (B1b, diamond). Red arrow in C points to a blood vessel that has resulted in a gap of the DG GCL. H&E staining is presented in black and white to allow for visual contrast.

For the adult cases, 1 had FGCB on the medial part (Case No. 1, Lewy body dementia) and 3 on the internal limb (cases 2-4, Figure 1, bottom panel). For adult case number 1, it was present in 5/8 (63%) tissue sections and was a row of internal GCs (within the hilus), that in subsequent sections appeared as a focal split (as defined by Kinney et al.¹² and appears at the arrow in Figure 3B and C). This case also only had mild GCD on the internal limb (not shown). For the other 3 cases, the bilaminar row of cells was in the molecular layer and was found in 2/11 (18%) serial sections in case number 2 (unknown cause of death but normal brain), 2/9 (22%) first serial set and 0/5 second serial set of sections of case number 3 (ischemic heart disease), and 1/4 (25%) first serial set and 0/8 second serial set of case number 4 (cardiac cause of death involving coronary artery). For adult case number 2, distinguishing FGCB from single and cluster of ectopic cells required using the 20× objective to ensure it fits the definition of a continuous row of 8 cells (Figure 4A1). For cases 2-4, the GCs in the bilaminar layer were predominantly round, compared to the mix of round and ovoid shapes in the main DG (Figure 1).

Figure 3.

Adult case number 1 with bilaminar layer of the focal granule cell bilamination (FGCB) in the CA4/hilus region in 5/8 serial sections. Representative examples are indicated red equal signs in A and B. In the remaining 3 sections of this series, there are remnants at the tail end (black arrows in B and C), appearing as the possible start of the focal split. Black unequal signs indicate the end of the FGCB. Sections stained by immunohistochemistry for ALZ50, counterstained with hematoxylin, and shown in black and white to allow for visual contrast.

Figure 4.

Adult case number 2 was stained by non-radioactive in situ hybridization for NR1. At 4× magnification, focal granule cell bilamination (FGCB) is not appreciated in the boxed area (A), but only visible at ×20 (A1; red equal sign; demonstrating a row of 8 cells). It co-exists with single and clustered ectopic cells (black diamond). In subsequent sections (B1-E1), FGCB no is longer evident and is replaced by a row of ectopic cells (single and clusters). Red arrow in (A) points to a blood vessel disrupting the DG GCL on the internal limb.

Apart from adult case number 1 in which the bilaminar layer was in the hilus, the width of the FGCB varied between 93 and 200 µm and was due to the variation in the main DG thickness (range 40-124 µm) and not the bilaminar row (range 20-48 µm), with the thickness of the bilaminar row being on average half (pediatric) and one-third (adult) thinner than the main DG (Table 4). No measured parameter differed statistically between pediatric and adult age groups (Table 4).

Table 4.

Quantitative measurements of focal granule cell bilamination (FGCB) of the 6 cases in which it was identified.

	Location of FGCB, and dispersed direction	GCL thickness proximal to FGCB	FGCB Total Thickness	Distance of Gap	Basal layer thickness	Dispersed layer thickness
Infant (n = 5 sections)	Internal limb, Molecular layer	111 ± 25	128 ± 21	40 ± 13	52 ± 15	34 ± 11
Child (n = 1 section)	Internal limb, Molecular layer	93 ± 14	165 ± 15	48 ± 10	72 ± 8	30 ± 6
Adult case no. 1 (n = 5 sections)	Medial part, hilus	115 ± 14	313 ± 31	131 ± 24	123 ± 11	78 ± 8
Adult case no. 2 (n = 2 sections)	Internal limb, Molecular layer	120 ± 11	181 ± 27	41 ± 2	124 ± 1	35 ± 6
Adult case no. 3 (n = 2 sections)	Internal limb, Molecular layer	121 ± 2	137 ± 6	33 ± 4	70 ± 14	31 ± 15
Adult case no. 4 (n = 1 section)	Internal limb, Molecular layer	121 ± 22	186 ± 35	25 ± 10	124 ± 12	32 ± 8
Pediatric averagea		102 ± 13	147 ± 26	44 ± 6	62 ± 14	32 ± 3
Adult averageb		121 ± 1	168 ± 27	33 ± 8	106 ± 31	33 ± 2
P-value**		.07	.44	.20	.17	.78

Data presented as the average of 3 measurements along the FGCB per section ±SD. Units in micrometers (µm).

^aPediatric (average of the infant and child).

^bAdult average of cases 2, 3, and 4. Case 1 was excluded given dispersed layer was in the hilus and an outlier.

** P-value calculated by t-test comparing averages of pediatrics and adults.

Comparing cases with (n = 6) and without (n = 31) FGCB regardless of age, cluster ectopic cells were more common in cases with FGCB (50% vs 7%, P = .02), and although a similar trend was evident when separated according to pediatric (50% vs 5%) and adult (50% vs 9%) age, this did not reach statistical significance, most likely due to low case numbers (Table S3).

Comparison to literature prevalence

Published studies of infant and child cohorts that include “control” cases can be directly compared with our study. However, this is not possible for many adult studies that did not include control cases (Table 1). Moreover, not all features were studied in the literature. Thus, our comparison for these features is limited to GCP Type 1 (thinning and gaps) and 2 (FGCB, ectopic GCs, and GCD) (Table S4).

The heterogeneity of the appearance of FGCB identified in our cases is similar to that described in the literature (Figure S1). In this study, the prevalence of FGCB in infants (6%) is similar to the 8% prevalence reported by Kinney et al.,¹² but lower than the 19% reported by Roy et al.⁶ In children, our prevalence (17%) is comparable to the range of 10%-17% of the 3 studies available^6,7,27 (Table 1; Table S4). However, in adults, our FGCB prevalence (27%) was at the upper end of the range 3%-35% reported in TLE and SUDEP (Table 1; Table S4).

In our study, single ectopic cells in infants were twice as common as that reported in another infant cohort,¹² while in childhood and adulthood, prevalence was in agreement with 1 study per age group (child⁹, adult⁵), but 3 times more than other studies (Table S4). Our identification of GCD was higher than all other studies for each age group (Table S4).

For the adult cohort, the location of FGCB and clusters of ectopic cells within the internal limb and medial part of the DG was similar to that described by Blumcke et al.³ However, we did not identify FGCB or clusters of ectopic cells in the external limb, whereas Blümcke et al. identified them in 43% and 68%, respectively.³ There were also substantial differences in the distribution of thinning, gaps, GCD, and single ectopic cells (Table S5); this could be due to all sections in Ref.⁵ having been from patients with TLE,⁵ in contrast to our control cohort.⁵

DISCUSSION

The main findings of this study are that FGCB was identified in 16% of individuals who died with no clinically reported history of seizures, predominantly in the internal limb, and that the presence of FGCB and clusters of ectopic GCs was highly correlated. The prevalence of FGCB appeared to increase with age, being identified in 6% of infants, 17% of children, and 27% of adults. However, analysis in a greater number of cases, particularly in the childhood ages is required for statistical verification. While the prevalence in infants and children was similar to that reported in other studies, in adults it was much higher than the 0%-2% reported in control cohorts and was similar to the prevalence in the upper limit range reported for TLE and SUDEP.

Prevalence of FGCB

The similar prevalence of FGCB between studies in the infant and child age ranges is most likely due to the fact that these studies are relatively recent (ie, since 2012), and applied similar defining criteria. However, within this age group, immature GCs persist in the hilus,²⁸ and may mimic FGCB, resulting in inflated prevalence of FGCB if not taken into consideration.^6,13 We only identified FGCB in 1 of 6 children and only in 1 of 10 sections (1 of 5 serial sections and 0 in the second set of serial sections). Although limited by low case numbers, our data in this age range are unique in that the DG was sampled more widely (4 or more sections examined from each case, covering an anteroposterior distance of 28-100 µm), than in other studies in children.^6,7,27 Based on this and previously published studies, the likely prevalence of FGCB in pediatrics (1 month-12 years of age) with no known clinical history of seizures is approximately 10%.

The low prevalence of 0%-2% of FGCB in adult control cohorts in the literature,^2,8,11 compared to our study of 27%, may be due to lower case numbers, fewer sections being examined from each case, and differences in defining criteria. Lack of standardized definition results in subjective assessment and interobserver discordance,^4,7 especially if FGCB is subtle when compared to observations in TLE or SUDEP samples.⁸ Similar disparities between diagnostic groups due to subtlety in features have been reported in relation to the cornu ammonis pyramidal cell layer in adult humans; Sloviter et al.²⁹ showed that although controls had irregularities, they were not as prominent as those seen in TLE. Tangential sectioning of the GCL or proximity to a BV has also been discussed as related to the appearance of subtle FGCB.⁸ However, in our experience, this does not seem to hold true given the normal appearance of the GCL adjacent to FGCB, and that BVs were not observed in the same location of the FGCB in adjacent serial sections; this was verified by correlation analyses finding no association with BV disruption (Table S2).

Given the focal nature of FGCB, with each focus extending an average of 21 µm anteroposteriorly as we recently identified in infants,¹³ the likelihood of identifying a focus would increase with the number of adjacent serial sections examined. We identified FGCB in 18%-63% of adjacent serial section sets of 4-11 tissue sections, with no specific relationship between number of serial sections and percentage. Moreover, FGCB was not consistently identified in serial sets 20-50 µm apart. Thus, even with our higher sampling rate as compared to studies summarized in Table 1, it is possible our identification of FGCB is underestimated. An extensive study with a defined systematic sampling of the hippocampus in the coronal plane, initially focused at the level of the LGN for standardization, would be required to provide normalization of the probability to identify FGCB in a single coronal section and the minimum number of serial sections required to definitively identify its presence.

Morphological appearance and characteristics of FGCB

The appearance of FGCB is heterogeneous, with variations in length, thickness, the distance between the 2 layers/laminae, and cell shapes of the 2 layers (Figure 1; Figure S1). Quantitative analyses by Silva et al.³⁰ in adult TLE sections indicated that although the GCs of the bilaminar layer were round in profile (compared to ovoid in the main GCL), they were of similar dimension. Consistent with Silva et al., we also found that in our 3 adults where the FGCB was in the molecular layer, the GCs of the main GCL were predominantly ovoid while those of the bilaminar layer were round.

Regarding thickness, 1 study reported a range of 100-210 µm⁸ while another reported (100-350µm),⁶ with no difference between their cases and controls (SUDEP,⁸ SUDC with vs without epilepsy⁶). Consistent with Roy et al.,⁶ the FGCB thickness range for our sections was wide (93-360 µm), yet this was affected by the 1 adult case in which the FGBC was in the hilus as opposed to the other 5 cases in which it was in the molecular layer. Furthermore, we found no effect of age on FGCB thickness, consistent with the report by Roy et al.⁶ for their control cohort.

Based on the above findings, the variation in the appearance of FGCB between cases is not likely to be age-related and preliminary data suggests it may not be due to epilepsy status^6,8; rather, it may be related to the pathogenesis of the FGCB, with the 2 main hypotheses being altered migration and altered maturation.³¹ To study these hypotheses, identifying the chemical/molecular signature of the cell types in the DG features would be required and was beyond the scope of this study. Markers related to the maturation hypothesis include mature neuronal markers of NeuN, MAP2, calbindin, and calretinin, while those for the migration hypothesis include reelin, and some Ras homolog family member (Rho) GTPases.³² Of these, recent data indicate that the GCs in the bilaminar layer have higher expression of calbindin⁶ and RhoA³² compared to GCs in the main-DG.

A correlation between the presence of FGCB and clusters of ectopic cells was evident across the 3 age groups in this study and has been previously reported in infants,¹² and in adults with TLE,⁵ suggesting a possible shared pathogenesis.

Relationship among DG features and contribution of age

The 10 features studied were predominantly located in the internal limb and medial part of the DG, as opposed to the external limb, thus suggesting higher vulnerability of GCs in the internal limb and medial DG. This is consistent with the trajectory of new cell generation and migration pattern in the development of the DG, whereby GCs of the external limb position and mature before the internal limb^14,15 and could be linked to the proposed ongoing neurogenesis in the adult DG,^33–35 although adult human neurogenesis remains debated.^36,37 Higher vulnerability of the GCs in the internal limb has been demonstrated in mice³⁸ where GCL thickening (dispersion) was doubled in the internal limb (in mice referred to as lower blade) than the external limb (upper blade), after kainic acid-induced epliptogenesis. This was evident in the DG of coronal sections in the septal coordinates, but not the temporal coordinates, where the thickening was restricted to the medial DG.³⁸ Thus, differences in the prevalence of DG GCL features along the septotemporal axis seem to be evident, hence supporting the need for increased sampling rate along the DG axis.

DG features of both loss (GCP Type 1) and gain (GCP Type 2) were evident in the same case, within the same tissue section and/or along the series of sections, supporting the migration hypothesis and indicating compensatory mechanisms could be occurring for the altered/disrupted cell migration. These were particularly evident for cases with cardiac-related causes of death and is consistent with reports of structural differences in the DG in sudden cardiac deaths in adults²⁹ and cardiac-related causes of death in pediatrics.⁶ However, our adult cases who had no major cause of death other than aging and dementia, also had some features present, albeit to a lesser extent compared to those with cardiac-related causes, and these features tended to be specifically GCD, single ectopics, and thinning. That DG features were identified in these individuals is not surprising when considering that other hippocampal changes are evident, for example, the prevalence of hippocampal sclerosis is 30%-40% in epilepsy and 0.4%-26% in dementia and aging,³⁹ with GCP Type 1 and Type 2 of the DG taken into consideration when classifying the type of hippocampal sclerosis.⁴⁰

The prevalence of gaps was 2.5 times higher in childhood and adulthood when compared to infancy. The significance of this finding is unclear; however, it is interesting that a correlation between gaps and single ectopic cells was found specific to the pediatric age group, suggesting possible compensatory mechanisms between these 2 opposing types of features (cell loss vs gain) in pediatrics.

Limitations

The absence of epilepsy or seizures in our cases was based on information extracted from the limited clinical information provided to the forensic department during autopsy for the pediatric cases, and to the NSW brain bank in adult cases. Consequently, it is possible that some cases with undocumented epilepsy or seizures were included.

Given the retrospective nature of the study, analyses were on differently stained sections over a 25-year period. However, counterstaining of the IHC sections was sufficient to allow for visual differentiation of the DG features. Moreover, the precise anteroposterior localizations of our sections were not standardized; despite studying a minimum of 4 adjacent serial sections within the body of the hippocampus, this is still only a minute snapshot of the entire DG. Thus, a more systematic analysis of sets of serial sections along intervals of the hippocampus in the anteroposterior axis would be necessary to verify our data, particularly in the child age range for which we only had 6 individuals.

Conclusion

Across the age span of 1 month to 91 years, FGCB is identified in 16% of individuals with no known history of seizures. The prevalence of FGCB increases with age, being 6% in infants, 17% in children, and 27% in adults, and it is predominantly located to the internal limb of the DG and is highly correlated with the presence of clusters of ectopic GCs, suggesting a possible shared pathogenesis.

SUPPLEMENTARY MATERIAL

Supplementary material is available at academic.oup.com/jnen.

FUNDING

The adult tissues were received from the New South Wales Brain Tissue Resource Centre at the University of Sydney, which is supported by the University of Sydney. Research reported in this publication was supported by the National Institute of Alcohol Abuse and Alcoholism of the National Institutes of Health under Award Number R28AA012725. The content is solely the responsibility of the authors and does not represent the official views of the National Institutes of Health.