Correlation Among Genotype, Phenotype, and Histology in Neuronal Ceroid Lipofuscinoses: An Individual Patient Data Meta-Analysis

Pediatric Neurology xxx (2016) 1e8

Contents lists available at ScienceDirect

Pediatric Neurology journal homepage: www.elsevier.com/locate/pnu

Original Article

Correlation Among Genotype, Phenotype, and Histology in Neuronal Ceroid Lipofuscinoses: An Individual Patient Data Meta-Analysis Gewalin Aungaroon MD a, *, Barbara Hallinan MD, PhD a, Puneet Jain MD, DM b, Paul S. Horn PhD a, c, Christine Spaeth MS, CGC a, d, Ravindra Arya MD, DM a a

Division of Neurology, Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio Pediatric Neurology Services, Department of Neonatal, Pediatric and Adolescent Medicine, BL Kapur Super Specialty Hospital, New Delhi, India c Division of Epidemiology and Biostatistics, Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio d Division of Human Genetics, Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio b

abstract BACKGROUND: Neuronal ceroid lipofuscinoses (NCL) are heterogeneous neurodegenerative disorders. A better understanding of genotypeephenotypeehistology correlation is expected to improve patient care and enhance understanding for phenotypic variability. This meta-analysis studies the correlation of NCL genotypes with clinical phenotypes, ages of onset, and pathologic findings. METHODS: A structured MEDLINE search was performed using search strings incorporating relevant Medical Subject Headings (MeSH) terms. Studies of NCL patients with genetic, clinical, and histologic data were included. Individual patient data were extracted. Chi-square statistic was used to test the genotype-wise differences in clinical phenotypes and histology. The distribution of age(s) of onset as a function of genotype was explored. Pairwise comparisons were performed with robust analysis of variance. RESULTS: Sixty-eight studies including a total of 440 individuals with NCL were analyzed. Genetic testing was performed on 395 patients, and a pathologic mutation was identified in 372 of 395 of them. A significant clustering of genotypes into juvenile-onset (only CLN3) and infantile-onset (all others) phenotypes was observed (P < 0.0001). However, the CLN6 genotype indicated a bimodal onset and included 14 of 17 subjects with the adultonset phenotype. The estimated age of onset was respectively lower for subjects with CLN1 mutation (3.01 years, 95% confidence interval [CI] ¼ 2.54 to 3.49) and higher for those with CLN6 mutation (16.33 years, 95% CI ¼ 15.68 to 16.98), compared with other genotypes (P < 0.05 for pairwise comparisons). There was a significant (P < 0.0001) clustering of genotype observed according to the sampled tissue types and electron microscopic findings. CONCLUSIONS: NCL genotypes significantly differ in terms of ages of onset and clinical phenotypes. There is a distinct segregation of genotypes and electron microscopic findings and high-yield tissue types for pathologic study. This information can possibly facilitate testing and diagnosis in resource-limited settings. Keywords: neuronal ceroid lipofuscinoses, neurodegenerative, genotype, phenotype, histology

Pediatr Neurol 2016; -: 1-8 Ó 2016 Elsevier Inc. All rights reserved.

Introduction Disclosures/conflict of interest statements: None of the authors have any disclosures/conflicts of interest pertinent to this article.

Article History: Received December 3, 2015; Accepted in final form March 30, 2016 * Communications should be addressed to: Dr. Aungaroon; Division of Neurology; Cincinnati Children’s Hospital and Medical Center; ML 2015; 3333 Burnet Avenue; Cincinnati, OH 45229. E-mail address: [email protected] 0887-8994/$ e see front matter Ó 2016 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.pediatrneurol.2016.03.018

Neuronal ceroid lipofuscinoses (NCL) are lysosomal storage disorders which constitute the most common neurodegenerative conditions in childhood. NCLs are clinically and genetically heterogeneous and are characterized by a variable combination of progressive cognitive and motor regression, seizures, vision loss, and decreased longevity.1 Historically, NCLs were classified mainly according to the age of onset of

2

G. Aungaroon et al. / Pediatric Neurology xxx (2016) 1e8

symptoms. With increasing recognition of underlying genetic defects, newer nomenclature and classification schemes have been developed.2 The underlying pathophysiologic mechanisms that are responsible for the variability in clinical presentation are still unclear but may be related to differential gene expression within the central nervous system, or differential vulnerability of various neuronal populations at different ages. A precise diagnosis is essential for prognostication and genetic counseling. Globally, the definitive diagnosis of these entities is limited because of the lack of accessible and costeffective genetic testing. Therefore a better understanding of genotypeephenotype correlation may improve patient care. In addition, it may facilitate understanding of pathophysiologic mechanisms responsible for phenotypic variability in this group of diseases. This is a systematic review of case series of NCL patients and individual patient data (IPD) meta-analysis to study correlation among genotype, phenotype, and histology. Materials and Methods Literature search and selection of studies A structured PubMed search was conducted using appropriate Medical Subject Headings (MeSH) terms obtained by exploding the tree

for “Neuronal Ceroid Lipofuscinoses” (Table e1). English language human NCL studies, published up to March 2013, were included. All abstracts were screened by three authors (G.A., R.A., and P.J.), and obviously unsuitable publications were rejected. Three authors (G.A., R.A., P.J.) then independently applied inclusion and exclusion criteria with any disagreements being resolved by consensus. Studies which did not provide sufficient clinical (age of onset, phenotype, or initial clinical features), genetic (gene, mutation), and pathologic (tissue, microscopic findings) data were excluded. Studies which did not describe a clear method for diagnosis were also excluded. Studies which provided only group data (n ¼ 4) and not IPD were excluded from analysis for genotypeephenotype or genotypeehistology correlation. Search results and study selection have been documented in a PRISMA diagram3 (Fig 1).

Data extraction and outcome measures For eligible studies, three of the authors (G.A., P.J., R.A.) extracted the data using predesigned data tables. Demographic details and clinical variables of interest including age at onset, sex, initial clinical feature(s), clinical phenotype, gene, mutation, tissue type studied, and electron microscopic (EM) findings were extracted for individual patients on a Microsoft Excel spreadsheet. The primary outcome measure was genotypeephenotype correlation as measured by segregation of clinical phenotypes according to genotype. For this outcome, we either used the clinical phenotype ascribed to a particular patient by the study author or used the age and initial clinical features described by the study author to obtain a recognized phenotype. The secondary outcomes included

FIGURE 1. The PRISMA diagram for the literature search and selection of studies. (The color version of this figure is available in the online edition.)

G. Aungaroon et al. / Pediatric Neurology xxx (2016) 1e8 correlation between genotype and distribution of age(s) of onset, tissue types studied, and EM findings.

Statistical analysis An IPD meta-analysis was performed.4 The relationship between genotypes (CLN1, CLN2, CLN3, CLN5, CLN6, and CLN7) and phenotypes (infantile and juvenile) was examined using chi-square test. One-way analysis of variance was conducted with age of onset as the dependent variable and genotype being the explanatory variable. Where age of onset was not available for individual patients, we used the reported mean, median, and midrange for the study, in that order. To resist the effect of possible outliers, a robust analysis of variance was used.5 Pairwise differences between the genotypes were computed, and to account for multiple testing, false discovery rate was used with a family-wise error rate set at 0.05. The correlation between genotypes and tissue used for pathologic study (rectum, skin, muscle, lymphocyte, nonspecific mucosa, cardiac, brain, conjunctiva, appendix and spleen) was examined using chi-square test. The relationship between genotypes and EM findings (curvilinear body [CVB], fingerprint profile [FPP], granular osmiophilic deposits [GRODs], dense bodies, rectilinear profiles and others [lipid droplets, autofluorescent inclusion, nonspecific inclusion and vacuoles]) was also examined by using chi-square test. Genotypes with 5 subjects (CLN10 or CTSD [n ¼ 3], CLN14 [n ¼ 2], and DNAJC5 [n ¼ 4]) were excluded from the statistical analysis. All analyses were conducted using SAS statistical software, version 9.3 (SAS Institute Inc, Cary, NC).

Results

A total of 68 studies with a total of 440 patients were identified (Table e2). Genetic testing was performed on 395 patients, and a pathologic mutation was identified in 372 of 395 patients (94.2%, Supplementary Figure). The clinical phenotypes were reported in a total of 337 patients. These 337 patients were categorized as congenital (n ¼ 2, 0.6%),

3

infantile (n ¼ 168, 49.8%), juvenile (n ¼ 137, 40.7%), and adult-onset NCL (n ¼ 30, 8.9%) as stated by different authors for their respective patients. The phenotype IPD was not reported in 10 studies.6-15 Among the 372 patients with pathologic genetic mutations, the clinical phenotype was reported in 272 of the patients. Patients were categorized as congenital (n ¼ 2, 0.5%), infantile (n ¼ 125, 33.3%), juvenile (n ¼ 124, 33.3%), adult-onset NCL (n ¼ 21, 5.6%), and not reported (n ¼ 100, 26.9%). The distribution of available clinical phenotypes according to different genotypes is summarized in Fig 2. All except seven studies12,16-21 provided data about age(s) of onset which ranged from first day of life9 to 62 years.6 The mean age of onset of symptoms was 5.9 years (8.0). The distribution of the age(s) of onset as a function of genotype is illustrated in Fig 3. Genotype and clinical phenotype

On first inspection of the data, it was observed that 14 of 17 (82.4%) patients with adult-onset NCL phenotype were reported to have a mutation in CLN6. Hence, the statistical analysis for the association between genotypes and clinical phenotype was limited to infantile and juvenile phenotypes. A significant difference in the clustering of genotypes was observed according to the clinical phenotype (P < 0.0001, Table 1). Although the CLN1, CLN2, CLN5, and CLN7 genotypes were significantly more likely to present as infantile NCL, CLN3 was the only genotype more likely to present as juvenile NCL (76 of 78 patients, 97.4%). In addition, CLN6 indicated a bimodal presentation with both infantile and adult-onset NCL. None of the patients with a mutation in CLN6 presented as juvenile NCL.

90 80

Number of paents

70

3

60 50

33 congenital adult

76

40

juvenile 30 20

14

7

3

37 5

24

10

2

0 CLN1

infanle

CLN2

CLN3

23

10 CLN5

19 8

CLN6

CLN7

1

1

2

4

CLN8 CLN10 CLN14 CTSD DNAJC5

Genotype FIGURE 2. The distribution of reported clinical phenotypes by genotype (N ¼ 272). (The color version of this figure is available in the online edition.)

4

G. Aungaroon et al. / Pediatric Neurology xxx (2016) 1e8

FIGURE 3. The age of onset in patients with neuronal ceroid lipofuscinoses by genotype. (Legend: genotypes with <5 subjects including; CLN10 [n ¼ 1], CLN14 [n ¼ 2], CTSD [n ¼ 2], and DNAJC5 [n ¼ 4] were not depicted). (The color version of this figure is available in the online edition.)

all other genotypes including CLN1, CLN2, CLN3, CLN5, and CLN7 (Table 2).

Genotype and age(s) of onset

We performed pairwise comparisons to analyze the association between age(s) of onset and different genotypes using robust methods as described previously. The estimated age(s) of onset from pooled data as a function of genotype is summarized in Table 2. Multiple pairwise comparisons indicated that the estimated age (years) of onset for the CLN1 mutation group (3.01  0.24) was significantly lower than other genotypes including CLN2 (3.85  0.28, P ¼ 0.023), CLN3 (4.91  0.28, P < 0.0001), CLN5 (4.54  0.44, P ¼ 0.002), and CLN6 (16.33  0.33, P < 0.0001). Although the age of onset for the CLN1 genotype was also lower than the CLN7 (3.70  0.29) genotype, the difference did not attain statistical significance (P ¼ 0.07). In addition, the estimated age of onset (years) for the CLN6 group (16.33  0.33) was significantly higher than TABLE 1. The Distribution of Infantile Neuronal Ceroid Lipofuscinoses (INCL) and Juvenile Neuronal Ceroid Lipofuscinoses (JNCL) by Genotype

Gene

INCL; N (%)

JNCL; N (%)

Total; N

CLN1 CLN2 CLN3 CLN5 CLN6 CLN7 Total

37 (52.86) 24 (77.42) 2 (2.56) 10 (66.67) 23 (100) 19 (86.36) 115

33 (47.14) 7 (22.58) 76 (97.44) 5 (33.33) 0 (0.00) 3 (13.64) 124

70 31 78 15 23 22 239

P value < 0.0001 for chi square (c2 ¼ 115.88, df ¼ 5).

Genotype and sampled tissue

At least one positive diagnostic EM finding on tissue study was observed in 264 patients. Skin was the most common tissue sampled (n ¼ 154 of 264, 58.3%), followed by lymphocytes (n ¼ 98 of 264, 37.1%), and rectal mucosa TABLE 2. The Correlation Between Age of Onset and Genotype in Patients With Neuronal Ceroid Lipofuscinoses and Genotype

Gene

CLN1 CLN2 CLN3 CLN5 CLN6 CLN7

CLN1 CLN2 CLN3 CLN5 CLN6

Onset Estimate

Standard Error

3.01 3.85 4.91 4.54 16.33 3.70

0.24 0.28 0.28 0.44 0.33 0.29

95% Confidence Interval Lower Limit

Upper Limit

2.54 3.31 4.35 3.68 15.68 3.13

3.49 4.39 5.46 5.40 16.98 4.27

CLN2

CLN3

CLN5

CLN6

CLN7

0.023*

<0.0001* 0.008*

0.002* 0.180 0.486

<0.0001* <0.0001* <0.0001* <0.0001*

0.070 0.715 0.003* 0.110 <0.0001*

* Indicates a statistically significant difference at the 0.05 after accounting for multiple testing.

G. Aungaroon et al. / Pediatric Neurology xxx (2016) 1e8

(n ¼ 68 of 264, 25.7%) with some patients having more than one tissue sampled. We tested the association between the tissue type studied and the different genotypes, which indicated a significant difference in the clustering of genotypes observed according to the types of tissue studied (P < 0.0001) including skin, lymphocytes, and rectal mucosa. For the CLN1 genotype, pathognomonic lysosomal inclusions were found on EM studies of skin and lymphocytes more often than those on samples of rectal mucosa. For the CLN2 and CLN7 genotypes, skin more frequently revealed a positive finding. For the CLN3 genotype, lymphocytes and rectal mucosa were most commonly positive for inclusions. For the CLN5 genotype, the number of biopsy samples was overall very small, but rectal mucosa and skin were found to have pathognomonic inclusions more frequently than lymphocytes. Finally, for the CLN6 genotype, skin and rectal mucosa were found to have similar frequencies of positive findings (Table 3). Genotype and EM finding(s)

TABLE 4. The Distribution of Genotype for Electron Microscopic Findings Among 264 Patients Who Had at Least One Pathognomonic Electron Microscopic Finding

Genotype

Curvilinear Body; N (%)

Fingerprint profile; N (%)

Granular osmiophilic deposits; N (%)

CLN1 CLN2 CLN3 CLN5 CLN6 CLN7 Test for overall difference using chi square

7 40 45 14 11 29

8 10 58 17 27 31

61 5 3 5 15 3

(2.7) (15.2) (17.0) (5.3) (4.2) (11.0) c2 ¼ 103.25, df ¼ 5; P value < 0.0001

(3.0) (3.8) (22.0) (6.4) (10.2) (11.7) c2 ¼ 111.11, df ¼ 5; P value < 0.0001

(23.1) (1.9) (1.1) (1.9) (5.7) (1.1) c2 ¼ 165.48, df ¼ 5; P value < 0.0001

Rectilinear profiles were identified in the CLN1, CLN3, and CLN7 genotypes in this study. Discussion

Overall, FPPs were found to be the most common EM finding, followed by CVBs and GRODs. Among a total of 264 patients with at least one positive EM finding on tissue study, CVBs were the most common EM finding for patients with mutations in the CLN3 or CLN2 genes (17.0% and 15.2%, respectively) followed by the CLN7 genotype (11.0%). FPPs were more frequently observed for the CLN3 genotype (22.0%), followed by CLN7 and CLN6 (11.7% and 10.2%, respectively). GRODs were by far the most common EM finding for the CLN1 genotype (Table 4). Analysis of the association between EM findings and different genotypes indicated a significant clustering of genotypes according to EM findings (P < 0.0001). For CLN1 and CLN2 genotypes, GRODs and CVBs were found to be the most common EM findings, respectively. For CLN3, CLN5, and CLN7 genotypes, CVBs were found to be slightly more common than FPPs. For CLN6, FPPs were found to be the most common EM finding. However, it should be observed that these three most common EM findings were observed in all genotypes described here. Other less common EM findings included dense bodies, identified with the CLN6 genotype. Lipid droplets were found in the CLN3, CLN5, and CLN6 genotypes. TABLE 3. The Distribution of Genotype for Skin, Lymphocytes, and Rectal Mucosa Pathologic Study Among 264 Patients Who Had At Least One Pathognomonic Electron Microscopic Finding

Genotype

Skin; N (%)

Lymphocytes; N (%)

Rectal Mucosa; N (%)

CLN1 CLN2 CLN3 CLN5 CLN6 CLN7 Test for overall difference using chi square

46 32 21 8 19 28

41 13 39 2 1 2

4 4 34 9 15 2

(17.4) (12.1) (8.0) (3.0) (7.2) (10.6) c2 ¼ 56.17, df ¼ 5; P value < 0.0001

5

(15.5) (4.9) (14.8) (0.8) (0.4) (0.8) c2 ¼ 47.95, df ¼ 5; P value < 0.0001

(1.5) (1.5) (12.9) (3.4) (5.7) (0.8) c2 ¼ 69.40, df ¼ 5; P value < 0.0001

This IPD meta-analysis supports a distinct segregation of genotypes and clinical phenotypes within the NCLs. Although the infantile phenotype is genetically heterogeneous, our meta-analysis of the published literature from over a decade and spanning multiple countries attests to a cosegregation of juvenile NCL with CLN3 and adult-onset NCL with CLN6 genotypes, respectively. In addition, this meta-analysis reveals that certain genotypes, particularly CLN1 and CLN6, can be distinguished clinically from other genotypes based on the age of onset of the initial symptom. This information has important clinical and research implications. Testing for all known NCL genes in a single panel is not widely available in resource-poor countries.22 Our meta-analysis reveals that for certain clinical phenotypes (juvenile and adult-onset NCL) and genotypes (CLN1 and CLN6), a genetic diagnosis can be suspected strongly even without the confirmatory evidence. This can help a clinician in a setting with limited resources to either counsel the family accordingly short of a confirmed genetic diagnosis or consider targeted testing which may be more cost effective. It should be clarified that although 97.4% (76 of 78) of patients with CLN3 mutations present with juvenile NCL, they represent only 61% (76 of 124) of all patients with juvenile NCL (Table 1). Similarly, although mutations in other genes including CLN1, CLN3, CLN4, CLN5, CTSD or CLN10, GRN or CLN11, and CTSF or CLN13 have been reported to present with an adult-onset NCL phenotype, they represent isolated reports which may be relevant in certain ethnic groups.23-25 Hence, a clinician using this information has to apply it in the appropriate demographic context. In spite of current knowledge of the genetics of the NCLs, the pathologic diagnosis may be desirable when genetic testing is limited or the clinical presentation is equivocal. The pathologic diagnosis of NCL is mainly based on ultrastructural abnormalities observed on electron microscopy. In this meta-analysis, the tissue types that most frequently indicated diagnostic pathologic findings were skin, rectal mucosa, and lymphocytes. The analysis revealed a distinct segregation of genotypes in EM studies of these three tissues. This meta-analysis also supports a statistically

6

G. Aungaroon et al. / Pediatric Neurology xxx (2016) 1e8

significant segregation of genotypes with specific EM findings of CVBs, FPPs, and GRODs. However, these three EM findings could be observed in all genotypes analyzed in this meta-analysis. Hence, these three distinctive EM findings may be suggestive but not pathognomonic for certain genotypes. Further, this meta-analysis reveals potential avenues of research. It is well known that the NCLs have genetic and phenotypic heterogeneity.2 Historically, these disorders were considered a common entity because of their neurodegenerative course and microscopic evidence of lysosomal accumulation of lipofuscin.1 However, there is lack of clarity about any underlying biological basis for grouping these disorders together under a single rubric. This meta-analysis supports the thought that juvenile (CLN3) and adult-onset (CLN6) NCL may possibly be biologically different from the group of infantile NCLs, although CLN6 also has an infantile phenotype. Future studies could attempt to elucidate the reasons for this genotypeephenotype correlation. One hypothesis that may be worth testing is whether differences in the spatial and temporal expression of the CLN3 and CLN6 gene products and their downstream effects result in distinct clinical manifestations. CLN6 is particularly interesting in this regard, with two nonoverlapping ages of onset and phenotypes. This study is based on reports which were dissimilar in terms of the reported data set and its content and format. Hence, the meta-analysis can only synthesize but not improve the quality of data. Historically, certain genotypes have been reported to be associated with “variant” phenotypes because they did not fit the classic descriptions of recognized phenotypes. For example, in certain populations, CLN6 mutations present with onset of symptoms around 5-7 years of age, feature small mixed storage bodies, and are associated with death in affected individuals in middle of third decade (Costa Rican variant).26,27 One of the limitations of this study is that we grouped these patients with variant late-infantile NCL with the “classic” infantile NCL to facilitate group-level comparisons. Because of the rarity of this disorder and the possible misdiagnosis due to the lack of a definitive genetic evaluation, certain genotypes were not included in this meta-analysis due to the limited number of subjects. However, we believe that this should not significantly affect the main findings in this study. In addition, in those subjects who had more than one type of tissue biopsied or more than one EM finding, the detailed findings for each tissue type were not always described. This prevented analysis of the association among genotypes, sampled tissues, and EM findings in a more statistically rigorous manner. Conclusion

This IPD meta-analysis of 440 patients with NCL indicates that the CLN3 genotype is significantly more likely to present with a juvenile phenotype compared with other genotypes. Second, a vast majority of the adult-onset NCL patients reported in literature have CLN6 mutations. In addition, signs begin significantly earlier in patients with CLN1 mutations compared with other genotypes. In addition, EM studies of skin more frequently revealed a diagnostic result in case of the CLN1, CLN2, and CLN7 genotypes,

lymphocytes did so in CLN1 and CLN3 genotypes, and rectal mucosa did so in CLN3 genotype. Specific EM findings are associated with certain genotypes; however, they are not pathognomonic. For patients with neurodegenerative signs consistent with NCL, the correlation between genotype and clinical phenotype, age of onset, high-yield tissue type for pathologic study, and EM findings may be useful for physicians practicing in resource-limited settings with lack of access to genetic testing. This study can help them optimize patient care, counsel families regarding prognosis, and prioritize molecular genetic testing for an individual patient in a cost-effective manner. Finally, the study also suggests a hypothesis for future investigation regarding age-related variation in expression of CLN genes and its relationship to the clinical phenotype. Sources of support: None. Authors’ contributions: study concept was provided by B.H. The study was designed by R.A. Data were extracted by G.A., P.J., and R.A. and analyzed by P.S.H. and R.A. The first draft of article was prepared by GA and critically reviewed by all authors. All authors approve of the final version.

References 1. Mole SE, Williams RE, Goebel HH. The neuronal ceroid lipofuscinoses (Batten disease). 2nd ed. Oxford: Oxford University Press; 2011. 2. Williams RE, Mole SE. New nomenclature and classification scheme for the neuronal ceroid lipofuscinoses. Neurology. 2012;79:183-191. 3. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. Ann Intern Med. 2009;151:W65-W94. 4. Simmonds MC, Higgins JP, Stewart LA, Tierney JF, Clarke MJ, Thompson SG. Meta-analysis of individual patient data from randomized trials: a review of methods used in practice. Clin Trials. 2005;2:209-217. 5. Hettmansperger TP, McKean JW. Robust nonparametric statistical methods. 2nd ed. Boca Raton, FL: CRC Press; 2011. 6. Arsov T, Smith KR, Damiano J, et al. Kufs disease, the major adult form of neuronal ceroid lipofuscinosis, caused by mutations in CLN6. Am J Hum Genet. 2011;88:566-573. 7. Elkay M, Silver K, Penn RD, Dalvi A. Dystonic storm due to Batten’s disease treated with pallidotomy and deep brain stimulation. Mov Disord. 2009;24:1048-1053. 8. Fealey ME, Edwards WD, Grogan M, Orszulak TA. Neuronal ceroid lipofuscinosis in a 31-year-old woman presenting as biventricular heart failure with restrictive features. Cardiovasc Pathol. 2009;18:44-48. 9. Fritchie K, Siintola E, Armao D, et al. Novel mutation and the first prenatal screening of cathepsin D deficiency (CLN10). Acta Neuropathol. 2009;117:201-208. 10. Kohan R, Carabelos MN, Xin W, et al. Neuronal ceroid lipofuscinosis type CLN2: a new rationale for the construction of phenotypic subgroups based on a survey of 25 cases in South America. Gene. 2013;516:114-121. 11. Kousi M, Siintola E, Dvorakova L, et al. Mutations in CLN7/MFSD8 are a common cause of variant late-infantile neuronal ceroid lipofuscinosis. Brain. 2009;132:810-819. 12. Moore SJ, Buckley DJ, MacMillan A, et al. The clinical and genetic epidemiology of neuronal ceroid lipofuscinosis in Newfoundland. Clin Genet. 2008;74:213-222. 13. de los Reyes E, Dyken PR, Phillips P, et al. Profound infantile neuroretinal dysfunction in a heterozygote for the CLN3 genetic defect. J Child Neurol. 2004;19:42-46. 14. Staropoli JF, Karaa A, Lim ET, et al. A homozygous mutation in KCTD7 links neuronal ceroid lipofuscinosis to the ubiquitinproteasome system. Am J Hum Genet. 2012;91:202-208. 15. Xin W, Mullen TE, Kiely R, et al. CLN5 mutations are frequent in juvenile and late-onset non-Finnish patients with NCL. Neurology. 2010;74:565-571.

G. Aungaroon et al. / Pediatric Neurology xxx (2016) 1e8 16. Hofmann SL, Das AK, Lu JY, Wisniewski KE, Gupta P. Infantile neuronal ceroid lipofuscinosis: no longer just a ‘Finnish’ disease. Eur J Paediatr Neurol. 2001;5 Suppl A:47-51. 17. Kirveskari E, Partinen M, Salmi T, et al. Sleep alterations in juvenile neuronal ceroid-lipofuscinosis. Pediatr Neurol. 2000;22:347-354. 18. Kohan R, Cismondi IA, Kremer RD, et al. An integrated strategy for the diagnosis of neuronal ceroid lipofuscinosis types 1 (CLN1) and 2 (CLN2) in eleven Latin American patients. Clin Genet. 2009;76: 372-382. 19. Lonnqvist T, Vanhanen SL, Vettenranta K, et al. Hematopoietic stem cell transplantation in infantile neuronal ceroid lipofuscinosis. Neurology. 2001;57:1411-1416. 20. Siintola E, Partanen S, Stromme P, et al. Cathepsin D deficiency underlies congenital human neuronal ceroid-lipofuscinosis. Brain. 2006;129:1438-1445. 21. Pineda-Trujillo N, Cornejo W, Carrizosa J, et al. A CLN5 mutation causing an atypical neuronal ceroid lipofuscinosis of juvenile onset. Neurology. 2005;64:740-742.

7

22. Jadav RH, Sinha S, Yasha TC, et al. Clinical, electrophysiological, imaging, and ultrastructural description in 68 patients with neuronal ceroid lipofuscinoses and its subtypes. Pediatr Neurol. 2014;50:85-95. 23. Ramadan H, Al-Din AS, Ismail A, et al. Adult neuronal ceroid lipofuscinosis caused by deficiency in palmitoyl protein thioesterase 1. Neurology. 2007;68:387-388. 24. Velinov M, Dolzhanskaya N, Gonzalez M, et al. Mutations in the gene DNAJC5 cause autosomal dominant Kufs disease in a proportion of cases: study of the Parry family and 8 other families. PLoS One. 2012;7:e29729. 25. van Diggelen OP, Thobois S, Tilikete C, et al. Adult neuronal ceroid lipofuscinosis with palmitoyl-protein thioesterase deficiency: first adultonset patients of a childhood disease. Ann Neurol. 2001;50:269-272. 26. Pena JA, Cardozo JJ, Montiel CM, Molina OM, Boustany R. Serial MRI findings in the Costa Rican variant of neuronal ceroid-lipofuscinosis. Pediatr Neurol. 2001;25:78-80. 27. Munroe PB, Mitchison HM, O’Rawe AM, et al. Spectrum of mutations in the Batten disease gene, CLN3. Am J Hum Genet. 1997;61:310-316.

8

G. Aungaroon et al. / Pediatric Neurology xxx (2016) 1e8

SUPPLEMENTARY TABLE 1. Literature Search and Selection of Studies

Database (Vendor)

Medline (PubMed)

Search strings

(“Neuronal Ceroid-Lipofuscinoses/blood”[Mesh] OR “Neuronal Ceroid-Lipofuscinoses/cerebrospinal fluid”[Mesh] OR 00 “Neuronal Ceroid-Lipofuscinoses/ diagnosis”[Mesh] OR 00 “Neuronal CeroidLipofuscinoses/immunology”[Mesh] OR 00 “Neuronal Ceroid-Lipofuscinoses/pathology”[Mesh] OR 00 “Neuronal Ceroid-Lipofuscinoses/urine”[Mesh])d obtained by exploding the tree for “Neuronal Ceroid-Lipofuscinoses”[Majr] (“Neuronal Ceroid-Lipofuscinoses”[Majr]) AND 00 “Pathology”[Mesh] (“Neuronal Ceroid-Lipofuscinoses”[Majr]) AND (“Pathology”[Mesh] OR 00 “Biopsy”[Mesh]) (“Neuronal Ceroid-Lipofuscinoses”[Majr]) AND 00 “ Biopsy ”[Mesh] (“Neuronal Ceroid-Lipofuscinoses”[Majr]) AND (“Genes”[Mesh] OR 00 “Gene-Environment Interaction”[Mesh] OR 00 “Gene Regulatory Networks”[Mesh]) (“Neuronal Ceroid-Lipofuscinoses”[Majr]) AND (“Genetic Association Studies”[Mesh] OR 00 “Genetic Loci”[Mesh] OR 00 “Epigenesis, Genetic”[Mesh] OR 00 “Genetic Processes”[Mesh]) Language: English, species: human No restrictions

Filters Period

00

G. Aungaroon et al. / Pediatric Neurology xxx (2016) 1e8

8.e1

SUPPLEMENTARY TABLE 2. Studies Included in This Meta-Analysis

Author, Year [ref#]

Country

Aberg, 2009 Aiello, 2009 Aldahmesh, 2009

Finland Italy, France Saudi Arabia

Arsov et al., 20116

Mixed

Barisic, 2003 Bensaoula, 2000 Bessa, 2006 Bessa, 2008 Bonsignore, 2006 Bras, 2012

Croatia USA Portugal Portugal Italian Belgium

Cannelli, 2006 Cannelli, 2007 Cannelli, 2009 Chang, 2012

Number of Subjects 1 9 7

15

Age at Onset (yr)

Clinical Phenotypes Represented

Genotype(s)

6 2-6 1-6.5

JNCL INCL INCL, JNCL

16-62

CLN3 CLN7 CLN6, CLN7 and unidentified genotype CLN6

1 1 1 1 2 4

4 3 3 3 2-2.5 8

ANCL, unidentified phenotype INCL JNCL INCL INCL INCL JNCL

Italy Italy Italy China

3 2 13 9

3.5-6 5-7 2-5 2-3.5

INCL JNCL INCL INCL

Das, 1998

Mixed

29

0.67-7.5

van Diggelen et al., 200125 Elkay et al., 20097

The Netherlands USA

2 2

Elleder, 2008 Fealey et al., 20098

Czech Republic USA

1 1

3 3

Fritchie et al., 20099

USA

1

0.003

Fukumura, 2011 Golberg-Stern, 2009 Harris, 2002

Japan Israel (Arab origin) Middle East

1 2 1

6 3 1.5

INCL, JNCL, unidentified phenotype ANCL Unidentified phenotype INCL Unidentified phenotype Unidentified phenotype JNCL INCL INCL

Harris, 2011

Unidentified

1

5

ANCL

Hofmann et al., 200116 Holmberg, 2000 Ivan, 2005

The Netherlands Finish USA (African-American)

3 8 1

Unidentified 4 26

JNCL INCL ANCL

Kalviainen, 2007 Kirveskari et al., 200017 Kohan et al., 200918

Finland Finland Brazil, Argentina, Chile

1 28 11

6 Unidentified 2.5-6

JNCL JNCL JNCL, JNCL

Kohan et al., 201310

Argentina, Chile

25

2-10

Koul, 2007

Oman

11

0.5-4

Unidentified phenotype INCL

Kousi et al., 200911

Mixed

25

1.5-11

Kwon, 2005 Lam, 2001 Lonnqvist et al., 200119 Mantel, 2004 Mazzei, 2002 Mohammad, 2009 Moore et al., 200812

USA China Finland Switzerland Italy Saudi Arabia Canada

1 1 3 2 1 4 43

5 2 Unidentified 3-4 4 2-4.5 Unidentified

54-56 4-5

Unidentified phenotype JNCL INCL INCL INCL JNCL INCL INCL, JNCL, unidentified phenotype

CLN2 CLN3 CLN5 CLN2 CLN1 Unidentified genotype CLN8 CLN5 CLN6 CLN2 and unidentified genotype CLN1

CLN1 CLN3 CLN2 Unidentified genotype CLN10 CLN2 CLN2 Unidentified genotype Unidentified genotype CLN3 CLN5 Unidentified genotype CLN1 CLN3 CLN1, CLN2 and unidentified genotype CLN2 CLN2 and unidentified genotype CLN7 CLN3 CLN2 CLN1 CLN3 CLN1 CLN6 CLN2, CLN3, CLN5, CLN6, and unidentified genotype (continued on next page)

8.e2

G. Aungaroon et al. / Pediatric Neurology xxx (2016) 1e8

SUPPLEMENTARY TABLE 2. (continued ) Author, Year [ref#]

Country

Clinical Phenotypes Represented

Genotype(s)

de los Reyes et al., 200413

USA

CLN3

3 2-8

Unidentified phenotype INCL JNCL

Pena et al., 200126 Perez Poyato, 2011

Venezuela Spain

1 24

Poyato, 2011

Spain

24

2-7

JNCL

Poyato, 2012

Spain

6

0.67-1.25

INCL

Puri, 2010 Ramadan et al., 200723 Reinhardt, 2009 Sarpong 2009 Siintola 2005 Siintola et al., 200620 Siintola 2007 Simonati (2), 2000 Simonati, 2009 Staropoli et al., 201214 Starpoli, 2012b Stogmann, 2009 Takano, 2008

USA UK Turkish Germany, Lebanon Turkey Pakistan, England Turkey Italy Italy Mexican Northern Europe Egypt Japan

1 1 4 7 2 2 5 1 6 2 1 5 1

2 22 2.5-4 5-9 2.5-4 Unidentified 1.8-3.5 0.3 2-4 0.67-0.75 0.08 5 1.2

INCL ANCL INCL JNCL INCL CNCL INCL INCL INCL INCL INCL INCL INCL

Teixeira, 2003

Portugal

26

0.67-4

INCL, JNCL

Tessa, 2000 Pineda-Trujillo et al., 200521 Velinov et al., 201224

Italy Colombia USA

3 3 11

2.5-3.5 unidentified 16.2-38

INCL JNCL ANCL

Vercammen, 2003 Wailiany, 1999 Wang, 2011 Wisniewski, 1999 Xin et al., 201015

Belgium USA China unidentified Swedish

1 8 1 4 10

3 1-8 3.5 3-8 4-17

Zelnik, 2007

Israel

1

5

JNCL INCL, JNCL INCL INCL, JNCL Unidentified phenotype INCL

Abbreviations: ANCL ¼ Adult neuronal ceroid lipofuscinoses CNCL ¼ Congenital neuronal ceroid lipofuscinoses INCL ¼ Infantile neuronal ceroid lipofuscinoses JNCL ¼ Juvenile neuronal ceroid lipofuscinoses

Number of Subjects 1

Age at Onset (yr) 0.08

CLN6 CLN1, CLN3, unidentified genotype CLN1, CLN3, unidentified genotype unidentified genotype unidentified CLN1 CLN8 CLN3 CLN6 CTSD CLN7 CLN2 CLN1 CLN14 CLN5 CLN7 Unidentified genotype CLN1, CLN2, CLN3, unidentified genotype CLN2 CLN5 DNAJC5, unidentified CLN3 CLN1 CLN2 CLN2 CLN5 CLN8

G. Aungaroon et al. / Pediatric Neurology xxx (2016) 1e8

8.e3

90 80

Number of paents

70

79

84

73

60 50 45

40

47

30 27

20 10

8

0 CLN1

CLN2

CLN3

CLN5

CLN6

CLN7

Genotypes SUPPLEMENTARY FIGURE 1. The distribution of patients with neuronal ceroid lipofuscinoses by genotype (N ¼ 372).

1

2

2

4

CLN8 CLN10 CLN14 CTSD DNAJC5