Original Research Paper Multiple Sclerosis Journal— Retinal inner nuclear layer volume reflects Experimental, Translational and Clinical inflammatory disease activity in multiple July–Septemebr 2019, 1–11 DOI: 10.1177/ sclerosis; a longitudinal OCT study 2055217319871582 ! The Author(s), 2019. Article reuse guidelines: sagepub.com/journals- Lisanne J Balk , Danko Coric, Benjamin Knier, Hanna G Zimmermann, Raed Behbehani, permissions Raed Alroughani , Elena H Martinez-Lapiscina, Alexander U Brandt , Bernardo Sánchez-Dalmau, Angela Vidal-Jordana , Philipp Albrecht, Valeria Koska, Joachim Havla, Marco Pisa, Rachel C Nolan, Letizia Leocani , Friedemann Paul, Orhan Aktas , Xavier Montalban, Laura J Balcer, Pablo Villoslada, Olivier Outteryck, Thomas Korn and Axel Petzold; on behalf of the IMSVISUAL consortium Abstract Background: The association of peripapillary retinal nerve fibre layer (pRNFL) and ganglion cell-inner plexiform layer (GCIPL) thickness with neurodegeneration in multiple sclerosis (MS) is well estab- lished. The relationship of the adjoining inner nuclear layer (INL) with inflammatory disease activity is less well understood. Objective: The objective of this paper is to investigate the relationship of INL volume changes with inflammatory disease activity in MS. Methods In this longitudinal, multi-centre study, optical coherence tomography (OCT) and clinical data (disability status, relapses and MS optic neuritis (MSON)) were collected in 785 patients with MS (68.3% female) and 92 healthy controls (63.4% female) from 11 MS centres between 2010 and 2017 and pooled retrospectively. Data on pRNFL, GCIPL and INL were obtained at each centre. Results: There was a significant increase in INL volume in eyes with new MSON during the study (N ¼ 61/1562, b ¼ 0.01 mm3, p < .001). Clinical relapses (other than MSON) were significantly associ- ated with increased INL volume (b ¼ 0.005, p ¼ .025). INL volume was independent of disease pro- gression (b ¼ 0.002 mm3, p ¼ .474). Conclusion: Our data demonstrate that an increase in INL volume is associated with MSON and the occurrence of clinical relapses. Therefore, INL volume changes may be useful as an outcome marker for inflammatory disease activity in MSON and MS treatment trials. Keywords: Inflammation, inner nuclear layer, multiple sclerosis, optical coherence tomography Date received: 15 April 2019; Revised received 18 July 2019; accepted: 20 July 2019 Introduction axons) and the combined thickness of the ganglion Correspondence to: LJ Balk, Thinning of the inner retinal layers, as observed with cell layer and inner plexiform layer (GCIPL, consist- Department of Neurology, the use of optical coherence tomography (OCT), is a ing of mainly ganglion cells) have shown to be asso- Amsterdam UMC, MS Centre, Mailbox 7057, 1007 common finding in multiple sclerosis (MS) ciated with grey and white matter atrophy in patients MB, Amsterdam, the patients.1 Retinal OCT has been suggested as a with MS.2–6 Although the association of pRNFL and Netherlands. l.balk@amsterdamumc.nl structural imaging biomarker for neuroaxonal degen- GCIPL thickness with neurodegeneration in MS is Lisanne J Balk, eration, as reduced thickness of both the peripapil- well established, a more complex situation is lary retinal nerve fibre layer (pRNFL, consisting of observed for the adjoining inner nuclear layer Creative Commons Non Commercial CC BY-NC: This article is distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 License (http://www.creativecommons.org/licenses/by-nc/4.0/) which permits non-commercial use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage). Multiple Sclerosis Journal—Experimental, Translational and Clinical Danko Coric, (INL). A histological study demonstrated that the Data were collected between 2010 and 2017. MS Department of Neurology, Amsterdam UMC, Vrije INL, representing a neuronal network of bipolar, patients were included if they were between ages Universiteit Amsterdam, the amacrine and horizontal cells, shows signs of atro- 16 and 80 years, and had a diagnosis of clinically Netherlands phy but the presence of inflammatory cells was also isolated syndrome or MS (including relapsing– Benjamin Knier, Department of Neurology, described.7 Nevertheless, the INL seems not to be remitting (RR), secondary progressive (SP) and pri- Klinikum rechts der Isar, susceptible to retrograde degeneration caused by mary progressive (PP) subtypes) according to the Technische Universit€at München, Germany MS-related optic neuritis (MSON), as it does not revised 2010 McDonald Criteria.12 HCs were includ- Hanna G Zimmermann, show the extensive neuro-axonal injury in eyes ed if they were between ages 18 and 80 years and Charité – Universit€atsmedizin with MSON as observed in the pRNFL and had no history of any neurological disease or oph- Berlin, corporate member of Freie Universit€at Berlin, GCIPL.1,8 Rather than reflecting neurodegeneration thalmologic reason for retinal pathology. Regarding Humboldt-Universit€at zu like the innermost pRNFL and GCIPL, the INL may the OCT assessments, individuals were included if Berlin, and Berlin Institute of Health, NeuroCure Clinical rather be a biomarker for inflammatory processes. In they had at least two OCT measurements (baseline Research Center, Germany 2012, Gelfand et al. first described the presence of and at least one follow-up) with INL volume avail- Raed Behbehani, microcystic macular oedema (MMO) in the INL and able for at least one eye (minimum follow-up period Al-Bahar Ophthalmology Centre, Ibn Sina the relationship with disability.9 Furthermore, a ret- of six months). Patients were excluded if they had Hospital, Kuwait experienced symptomatic MSON within six months rospective study by Saidha and colleagues reported Raed Alroughani, preceding the OCT assessment (baseline or follow- Division of Neurology, that increased thickness of the combined INL and Department of Medicine, up), or if history of MSON was ambiguous or outer plexiform layer (OPL) was associated with dis- Amiri Hospital, Kuwait unknown. Inclusion and exclusion of individuals is ease activity in MS.10 More recently, Knier et al. Elena H Martinez- shown in the flowchart in Figure 1. Lapiscina, reported that successful use of disease-modifying IDIBAPS – Hospital Clinic, University of treatment (DMT) is associated with sustained reduc- OCT Barcelona, Spain tion of INL volume,11 suggesting that the INL could Retinal OCT was performed at each centre by use of Alexander U Brandt, serve as a biomarker to monitor central nervous spectral-domain OCT with Spectralis (Heidelberg Charité – Universit€atsmedizin Berlin, corporate member of system inflammation. Therefore, the aim of this Engineering, Heidelberg, Germany, N ¼ 10) or Freie Universit€at Berlin, Humboldt-Universit€at zu study was to investigate the relationship of INL OCT-2000 (Topcon Corp, Itabashi, Japan, N ¼ 1). Berlin, and Berlin Institute of volume changes over time with local and global Data on the INL and GCIPL volume (mm3) in the Health, NeuroCure Clinical inflammatory disease activity in a large cohort of Research Center, Germany macular area were acquired using a macular volume Bernardo Sánchez-Dalmau, patients with MS. scan centred on the fovea, using a 6 mm ring area. IDIBAPS – Hospital Clinic, Data on global pRNFL thickness (mm) were obtained University of Barcelona, Spain Methods using a 12-degree ring scan (corresponding to a Angela Vidal-Jordana, 3.4 mm diameter) manually placed around the optic Multiple Sclerosis Centre of Study design and participants disc. At each centre, automated segmentation of Catalonia, Neurology- Neuroinmunology Unit, Vall We used longitudinal data from the International OCT scans and quality control (including the assess- d’Hebron Hospital, Spain Multiple Sclerosis Visual System Consortium ment whether eyes had signs of MMO) were per- Philipp Albrecht, (IMSVISUAL) database (www.imsvisual.org). formed.13,14 Importantly, the scanning device and Valeria Koska, Department of Neurology, Patients were recruited from 11 centres in the protocols were kept identical for all longitudinal Medical Faculty, Heinrich Netherlands (Amsterdam N ¼ 165), Germany (Berlin measurements within each centre.15 Heine University, Germany N ¼ 81, Düsseldorf N ¼ 15, Munich (Universit€at Joachim Havla, Clinical and ophthalmological outcome measures Institute of Clinical München) N ¼ 11, Munich (Technische Universit€at Demographic data included data on sex, age at base- Neuroimmunology, Ludwig- Maximilians Universitaet München) N ¼ 169), Kuwait (Kuwait-City N ¼ 98), line and disease duration (from disease onset). Muenchen, Germany Spain (Barcelona CEMCAT N ¼ 39, Barcelona Clinical data were collected longitudinally and Marco Pisa, IDIBAPS N ¼ 69), Italy (Milan N ¼ 56), the United included MS subtype, occurrence of relapses Vita-Salute San Raffaele University, Italy States (New York City N ¼ 10) and France (Lille between visits, Expanded Disability Status Scale Rachel C Nolan, N ¼ 72). Healthy control individuals (HCs) were (EDSS) score, history of previous MSON and occur- New York University School of Medicine, USA recruited from three centres (Amsterdam N ¼ 41, rence of new episodes of MSON between visits, Letizia Leocani, Munich N ¼ 17 and Berlin N ¼ 34). All patients presence of MMO and use of DMT. The assessment Vita-Salute San Raffaele and HCs participated in local observational studies of history of symptomatic MSON (based on medical University, Italy and provided written informed consent for participa- history, according to a standard protocol),16 EDSS Friedemann Paul, Charité – Universit€atsmedizin tion in their respective studies. Data were pooled score and data on clinical relapses were provided by Berlin, corporate member of retrospectively. All data from reported and ongoing the individual centres. Freie Universit€at Berlin, Humboldt-Universit€at zu cohort studies at MS centres were stored in the Berlin, and Berlin Institute of IMSVISUAL repository. The raw dataset is avail- Importantly, given the longitudinal design of this Health, NeuroCure Clinical able from IMSVISUAL on request. study, we made a clear distinction between episodes 2 www.sagepub.com/msjetc Balk et al. Research Center, Germany Experimental and Clinical Research Center, Max Delbrueck Center for Molecular Medicine and Charité – Universit€atsmedizin Berlin, Germany Orhan Aktas, Department of Neurology, Medical Faculty, Heinrich Heine University, Germany Xavier Montalban, Multiple Sclerosis Centre of Catalonia, Neurology- Neuroinmunology Unit, Vall d’Hebron Hospital, Spain Division of Neurology, Department of Medicine, St. Michael’s Hospital, University of Toronto, Canada Laura J Balcer, New York University School of Medicine, USA Pablo Villoslada, IDIBAPS – Hospital Clinic, University of Figure 1. Flowchart of study design. Of the 903 individuals in the initial database, 785 patients and 92 healthy controls Barcelona, Spain were included in this study. All participants had at least two visits, and a subset also had a third or fourth visit. Olivier Outteryck, Longitudinal data on inner nuclear layer (INL) and ganglion cell-inner plexiform layer (GCIPL) volume (at least two Université Lille (Inserm U1171), France visits, minimum follow-up >6 months) was available for all included individuals and peripapillary retinal nerve fibre layer (pRNFL) for 765 patients and 91 healthy controls. Thomas Korn and Department of Neurology, IMSVISUAL: International Multiple Sclerosis Visual System Consortium; MS: multiple sclerosis; MSON: multiple Klinikum rechts der Isar, sclerosis optic neuritis. Technische Universit€at München, Germany Munich Cluster for Systems of MSON before the study (referred to as ‘pre-study as indicated. Figures showing longitudinal changes Neurology (SyNergy), Germany MSON’) and episodes of MSON during the follow- in retinal layer thickness were produced using rela- Axel Petzold, up of the study (referred to as ‘MSON during tive annualised change scores (i.e. baseline was set Department of Neurology, follow-up’). as 100%). Amsterdam UMC, Vrije Universiteit Amsterdam, the Netherlands EDSS assessment was performed by a certified Regarding the associations between annualised Moorfields Eye Hospital & The National Hospital for examiner and in the absence of acute relapses. change in retinal thickness and the occurrence of Neurology and Disability progression was defined by an increase relapses or disease progression, all eyes with a Neurosurgery, UK in EDSS score of 1.0 point in case EDSS score history of MSON were excluded. Short-term on behalf of the IMSVISUAL consortium was less than 5.5 at baseline, or an increase of 0.5 effects (clinical event and retinal change assessed if EDSS score was 5.5 or greater at baseline. This within same follow-up period) as well as long- approach is consistent with previous IMSVISUAL term effects (time-lag analyses, clinical event collaborative projects.17 between baseline and first follow-up visit and change in retinal layer thickness between first Statistical analyses and second follow-up visit) were investigated. Annualised changes in retinal layer thickness or Consequently, only individuals with at least three volume were calculated for every follow-up period. visits were included in these analyses. All analyses Subsequently, the annualised change scores were were adjusted for their respective baseline retinal averaged over the complete observation period, layer thickness. resulting in one average annualised rate of change for every eye. All analyses were therefore performed Correlations between the different layers were cal- on eye level, using generalised estimation equation culated with standardised regression coefficients in (GEE) models with a correlation matrix structure GEE models and are therefore also adjusted for that treats the eye measurements as exchangeable inter-eye dependency. Statistical analyses were per- to adjust for intra-subject inter-eye dependency.15 formed using SPSS V.22.0 (IBM Corp, Armonk, All GEE models were additionally adjusted for rel- NY, USA) and Stata V.14.1 (StataCorp LP, evant confounders (baseline OCT value, pre-study College Station, TX, USA) with a two-sided statis- episodes of MSON, disease duration, use of DMT) tical significance level of .05. www.sagepub.com/msjetc 3 Multiple Sclerosis Journal—Experimental, Translational and Clinical Results At baseline, MS patients showed significantly higher INL volume compared with HCs (difference of Baseline 0.02 mm3, p ¼ .018) and lower GCIPL volume and In total, 1570 eyes from 785 MS patients (68.3% pRNFL thickness (difference of –0.18 mm3 and –4.4 female) and 184 eyes from 92 HCs (63.4% mm, respectively, p < .001 for both comparisons). female) were included (Figure 1). MS patients had Eyes with pre-study episodes of MSON showed a a median disease duration of 6.4 years (interquartile higher INL volume compared with eyes without range (IQR) 1.9–15.0). The majority of patients (0.99  0.08 mm3 and 0.97  0.08 mm3, respective- (80.3%) had an RR disease course. More than half ly, p ¼ .001), whereas GCIPL volume and of all patients (N ¼ 419, 53.4%) had never experi- pRNFL thickness alike were lower in eyes with enced a clinically confirmed MSON before baseline. pre-study MSON compared with eyes without Of all patients with a history of at least one con- (Table 2). firmed episode of pre-study MSON (N ¼ 366), 281 (77%) patients had unilateral MSON and 85 (23%) Change over time in INL, GCIPL and pRNFL a history of MSON in both eyes (not necessarily thickness and the effect of MSON simultaneously). MMO was present in 2.4% of The median follow-up duration was 2.1 years (range, patients (15/638) and in 1.4% (18/1275) of eyes. 0.5 to 5.2 years) for MS patients and 2.0 years An overview of the baseline characteristics is (range, 0.6 to 4.6 years) for HCs. When all eyes of shown in Table 1. MS patients were analysed together, the INL showed Table 1. Baseline characteristics. All participants Healthy controls N ¼ 785 (N ¼ 92) Sex (female, N, %) 536 (68.3%) 59 (63.4%) Age (y) 41.0 (12.6) 43.4 (11.5) Disease duration (y, median (IQR)) 6.4 (1.9–15.0) EDSS (median (IQR)) 2.0 (1.0–3.0) Disease type CIS 45 (5.7%) RRMS 630 (80.3%) SPMS 74 (9.4%) PPMS 36 (4.6%) MSON before baseline, N (%) No previous MSON 419 (53.4%) MSON Unilateral MSON 281 (35.8%) Bilateral MSON 85 (10.8%) MMO before baseline (N ¼ 638) MMO– 623 (97.6%) MMOþ 15 (2.4%) Disease-modifying treatment at moment of baseline (N ¼ 743) None 343 (46.2%) Interferon beta 172 (23.2%) Glatiramer acetate 72 (9.7%) Natalizumab 61 (8.3%) Fingolimod 53 (7.1%) Dimethyl fumarate 20 (2.7%) Othera 21 (2.8%) CIS: clinically isolated syndrome; EDSS: Expanded Disability Status Scale; IQR: interquartile range, reported as 25th and 75th percentile; MMO: microcystic macular oedema; MSON: multiple sclerosis–related optic neuritis; PP: pri- mary progressive; RR: relapsing remitting; SP: secondary progressive. a Rituximab, teriflunomide, azathioprine, mitoxantrone, cyclophosphamide, alemtuzumab and mycophenolate mofetil. 4 www.sagepub.com/msjetc Balk et al. Table 2. Retinal layer thickness at baseline. MSON No MSON p valuea p valuea p valuea All eyes before BL before BL HCs MSON No MSON No MSON N ¼ 1570 N ¼ 451 N ¼ 1119 N ¼ 184 vs HC vs HCs vs MSON INL (mm3) 0.98 (0.08) 0.99 (0.08) 0.97 (0.08) 0.96 (0.09) .001 .066 .001 GCIPL (mm3) 1.79 (0.26) 1.62 (0.25) 1.86 (0.23) 1.97 (0.19) <.001 <.001 <.001 pRNFL (mm) 91.4 (15.8) 81.4 (17.5) 95.2 (13.2) 95.8 (9.1) <.001 .106 <.001 BL: baseline visit; HC: healthy controls; GCIPL: ganglion cell-inner plexiform layer; INL: inner nuclear layer; MSON: multiple sclerosis– related optic neuritis; pRNFL: peripapillary retinal nerve fibre layer. a Generalised estimation equation analyses, unadjusted. Figure 2. Relative change in retinal layer thickness with 95% confidence interval (based on generalised estimation equation model) for (a) all multiple sclerosis (MS) and healthy control (HC) eyes and (b) stratified by multiple sclerosis optic neuritis (MSON). GCIPL: ganglion cell-inner plexiform layer; INL: inner nuclear layer; pRNFL: peripapillary retinal nerve fibre layer. a non-significant average annualised rate of change (b ¼ 0.001, p ¼ .219). In contrast, any episode of of –0.0003 mm3 (p ¼ .553, Figure 2(a)). HCs also MSON during the observation period strongly showed no significant change (annualised rate of affected INL volume. In eyes with MSON during change 0.006 mm3, p ¼ .245). follow-up (N ¼ 61/1562), INL volume showed a sig- nificant annualised increase of 0.01 mm3 (p < .001). Regarding the effect of MSON, there was a clear In contrast, in eyes without MSON during the obser- difference between pre-study MSON (i.e. before vation period, no significant annualised change in the baseline OCT assessment) and MSON occurring INL was observed (b ¼ –0.001 mm3, p ¼ .203, during the follow-up period. Pre-study MSON did Figure 2(b)). Exclusion of patients with a progres- not affect the rate of change in INL thickness signif- sive disease type, or adjustments for use of DMT, icantly. Eyes with and without pre-study MSON disease duration or participating centre, did not showed similar rates of change of the INL change these results (data not shown). www.sagepub.com/msjetc 5 Multiple Sclerosis Journal—Experimental, Translational and Clinical Table 3. (a) Short- and (b) long-term effects of MSON, clinical relapses (other than MSON) and disability progression on annualised change in INL and GCIPL volume and pRNFL thickness. (a) b (95% CI) short term p valuea MSON (N ¼ 26 eyes) vs no MSON (N ¼ 1039 eyes) INL 0.01 (0.006 to 0.020) <.001 GCIPL –0.13 (–0.18 to –0.08) <.001 pRNFL –7.61(–10.8 to –4.3) <.001 Relapse (N ¼ 214 eyes) vs no relapse (N ¼ 789 eyes) INL 0.000 (–0.004 to 0.004) .868 GCIPL –0.10 (–0.18 to –0.002) .012 pRNFL –0.54 (–1.14 to 0.07) .082 Progression (N ¼ 223 eyes) vs no progression (N ¼ 673 eyes) INL 0.001 (–0.004 to 0.005) .774 GCIPL 0.001 (–0.006 to 0.008) .764 pRNFL –0.13 (–0.66 to 0.41) .646 (b) b (95% CI) long term (time-lag model) p valuea MSON (N ¼ 11 eyes) vs no MSON (N ¼ 581 eyes) INL –0.006 (–0.026 to 0.013) .535 GCIPL 0.023 (–0.065 to 0.112) .604 pRNFL –1.124 (–3.78 to 1.53) .406 Relapse (N ¼ 148 eyes) vs no relapse (N ¼ 440 eyes) INL 0.005 (0.001 to 0.01) .025 GCIPL –0.005 (–0.015 to 0.005) .307 pRNFL –0.40 (–1.57 to 0.77) .501 Progression (N ¼ 97 eyes) vs no progression (N ¼ 409 eyes) INL 0.001 (–0.004 to 0.007) .609 GCIPL –0.006 (–0.02 to 0.006) .329 pRNFL –0.65 (–0.69 to 1.99) .342 b ¼ regression coefficient; CI ¼ confidence interval; HC: healthy controls; GCIPL: ganglion cell-inner plexiform layer; INL: inner nuclear layer; MSON: multiple sclerosis–related optic neuritis; pRNFL: peripapillary retinal nerve fibre layer. a Generalised estimation equation model adjusted for inter-eye dependency and baseline retinal thickness. The annualised rate of change for GCIPL in MS long-term (time-lag analyses, clinical event between patients was –0.012 mm3 (p < .001), which was sig- t0 and t1 and change in retinal layer thickness nificantly more than observed in HCs (–0.004 mm3, between t1 and t2, Table 3(b)) effects are reported. p ¼ 0.116, p value for comparison .008). The pRNFL The median duration of t0–t1 was 1.1 year (IQR 1.0– showed significantly more thinning in MS patients 1.9) and for t1–t2 the median duration was 1.0 year (–0.97 mm, p < .001) compared with HCs (–0.42 mm, (IQR 1.0–1.7). p < .001, p value for comparison .001, Figure 2(a)). For both layers, eyes with episodes of MSON during Clinical episodes of MSON during follow-up the follow-up period showed significantly more thin- demonstrated only a short-term effect on INL (thick- ning than unaffected eyes (Figure 2(b)). ening) and GCIPL and pRNFL (thinning). In the time-lag analyses investigating the long-term effects, Short- and long-term effects of clinical disease these effects disappeared. Exclusion of patients with activity on retinal layer thickness a progressive disease course did not change the sta- Table 3 demonstrates the effects of new episodes of tistical findings. MSON during follow-up, other clinical relapses and disease progression on annualised change in INL and Clinical relapses (other than MSON) during follow- GCIPL volume and pRNFL thickness. The short-term up were present in 24.4% of patients. The occur- (clinical event and retinal change assessed within the rence of clinical relapses during the first follow-up same follow-up period, Table 3(a)) as well as the was not related to change in INL within the same 6 www.sagepub.com/msjetc Balk et al. period (median 1.1 years from baseline, b ¼ 0.000, p ¼ .006, black line in Supplementary Figure 1(a)) 95% confidence interval (CI) –0.004 to 0.004, and to a lesser extent (although not statistically sig- p ¼ .868) but was significantly associated with an nificant) to a decrease in pRNFL (standardised increase in INL volume in the subsequent follow- b ¼ –0.15, p ¼ .148, black line in Supplementary up (median 2.2 years from baseline, b ¼ 0.005, Figure 1(b)). In eyes without new MSON, no signif- 95% CI 0.001 to 0.01, p ¼ .025). This effect was icant association with change in INL volume was similar when only patients with a relapsing disease observed (grey lines in Supplementary Figures 1(a) course were included (N ¼ 508 eyes, b ¼ 0.005 and (b)). In contrast, GCIPL and pRNFL show posi- (95% CI 0.00 to 0.01, p ¼ .049)). For GCIPL tive correlations both in eyes with new MSON (stand- volume and pRNFL thickness, this effect was more ardised b ¼ 0.41, p < .001) and without MSON pronounced in the short term (i.e. relapse and retinal (standardised b ¼ 0.26, p ¼ .004, see Supplementary volume change within the same follow-up period, Figure 1(c)). Table 3(a)). Discussion Disability progression was observed in 17.2% This longitudinal, multi-centre study demonstrates (during the entire follow-up period). Annualised that thickening of the INL as measured with change in INL volume was independent of disability spectral-domain OCT reflects adjacent inflammation progression both in the short and long term. of the optic nerve. Besides this association with local Likewise, disability progression was not significant- inflammation, the INL seems to reflect some degree ly associated with annualised changes in GCIPL or of global disease activity, as the occurrence of clin- pRNFL (Table 3(a) and 3(b)). ical relapses in any functional system was signifi- Effect of MMO, disease type and DMT on cantly associated with a subsequent increase in INL retinal changes volume. However, this effect was relatively small In the 1.4% of eyes with MMO before or during the (difference of 0.005 mm3) and should be interpreted study (18/1275 eyes), the INL volume at the last visit with caution, as the sensitivity might be limited. was 0.07 mm3 higher compared with eyes without MMO (p ¼ .006, adjusted for new episodes of These findings build on previous findings from other MSON). Likewise, the average annualised rate of studies, demonstrating the relationship between change of INL volume was significantly higher in thickening of the INL and physical disability9 and eyes with MMO compared with eyes without disease activity.10 Saidha et al.10 reported that INL/ (b ¼ 0.01, p ¼ .011, adjusted for baseline INL and OPL thickening at baseline was predictive of clinical episodes of MSON during follow-up), showing a relapses, new T2 and contrast-enhancing lesions on significant annualised increase over time in MMO magnetic resonance imaging and disability progres- eyes (0.01  0.02 mm3), but no change in eyes with- sion during follow-up. In contrast, in the present out (–0.0002  0.02 mm3). study we did not observe any predictive value of INL thickening on clinical relapses or disability pro- Just more than half the patients (53.8%) used DMT gression. Our data demonstrated an association during the study. Although the annualised change in between INL volume and clinical relapses only in INL volume was not influenced by use of DMT, the the time-lag model, in which clinical relapses pre- absolute INL volume was significantly higher in ceded INL thickening. This would suggest that INL patients using fingolimod compared with RRMS thickening occurred subsequent to inflammatory dis- patients who did not use any DMT, independent of ease activity. A predictive effect of baseline INL history of pre-study MSON and MMO, EDSS at volume on the occurrence of relapses or disability baseline and disease duration (difference 0.03 mm3, progression, as described by Saidha and col- p ¼ .004). Other therapies did not show significant leagues,10 was not observed in the present study. differences in INL volume. When thickening of the INL was first described, it Interrelationship between layers was directly linked to the presence of MMO.9 All analyses regarding the interrelationships between Although MMO is present in MS and is related to the layers demonstrated effect modification by pres- increased disability,9,10,18 it is not specific for MS ence of a new episode of MSON during follow-up and and may vary over time in individual patients.19,20 are therefore stratified. In eyes with MSON during Importantly, we have previously demonstrated that follow-up, an increase in INL volume was related to MMO was transient in 84% of cases.19 In the pre- a decrease in GCIPL volume (standardised b ¼ –0.42, sent study, MMO was present in 2.4% of patients, www.sagepub.com/msjetc 7 Multiple Sclerosis Journal—Experimental, Translational and Clinical which is consistent with previous findings.9–11,21 In neurodegeneration, whereas INL volume may be a the present study we observed increased INL volume valuable parameter for reflecting inflammato- in eyes with MMO, which is also consistent with ry activity. existing literature.1 Moreover, eyes with MMO also showed a significant increase in INL volume A recent study by Knier et al. reported that effective over time. Nevertheless, the findings of the present treatment with DMT in patients with MS is associ- study did not change when MMO eyes were exclud- ated with a sustained reduction in INL volume in the ed, suggesting that thickening of the INL can occur absence of MSON, and they suggested that INL in the absence of visually detectable MMO. volume may be a response marker for successful treatment of inflammation.11 Building on these find- The underlying mechanism responsible for thicken- ings, we investigated the effect of DMT on INL ing of the INL remains unknown. The findings of volume changes. Although the absolute INL this study would imply that previously suggested volume was significantly higher in patients using mechanisms such as inflammation-related dynamic fingolimod compared with RRMS patients who did fluid shifts and Müller cell dysfunction are more not use any DMT, which corroborates previous find- likely than other non-inflammatory mechanisms ings on retinal effects of this drug,27,28 the annual- such as traction and retrograde trans-synaptic degen- ised change in INL volume was not influenced by eration.21,22 Dynamic retinal layer volume changes use or type of DMT. However, it should be noted can be explained by fluid shifts due to a combination that DMT data were available only at baseline for of osmotic and hydrostatic gradients, the retinal the majority of patients, and that data on the exact glymphatic system.23–26 The INL is embedded duration of treatment or previous DMTs were not between the superficial vascular plexus and the available. This lack of detailed information did not deep capillary plexus, which can be clearly visual- permit a thorough investigation of direct effects of ised on OCT angiography.24 Typically, fluid reaches DMT or replication of previous results. the retina through the internal limiting membrane and both plexuses, whereas both plexuses and Another limitation of the study was that the data did Müller cells can absorb the interstitial fluid. In not permit the determination of how acute new epi- case of inflammation, diffusion of fluid from the sodes of MSON were. A systematic assessment of retinal blood vessels increases, leading to an increase the early time course of acute MSON would be in INL volume. Another suggested mechanism is extremely valuable, and there will need to be a con- pathology of Müller cells, which would impair the sensus as to what will be defined as onset of an acute absorption of interstitial fluid, also resulting in episode of MSON. Furthermore, disease activity was increased INL volume. Other suggested non- recorded only by clinical relapse activity. Data on inflammatory mechanisms such as traction or retro- radiological disease activity (new T2 and/or grade trans-synaptic degeneration are also plausible gadolinium-enhancing lesions during follow-up) but less well supported by our data, given the clear were not available. Therefore, no conclusions and direct increase in INL volume following MSON. could be made regarding the relationship of INL One approach to further elucidate the underlying volume and radiological disease activity. A mechanism would be the investigation of the retinal common limitation of multi-centre studies is the dif- vessels using OCT angiography. ference in methodology among the participating centres.29,30 The OCT device and software were Previously, we and others have described the limited the same for all centres (Spectralis) but one susceptibility of the INL to retrograde degeneration (Topcon). The data on retinal layer thickness of caused by MSON.1 This is in line with our current this particular centre were not significantly different findings, in which the pRNFL and GCIPL clearly from the other centres’, and additional adjustment showed significant thinning in eyes with previous for a potential centre effect did not change any of episodes of MSON, whereas for the INL no thinning the results (data not shown). but rather an increase in volume was observed. The opposing effects of local inflammation on the INL In summary, our data demonstrate that an increase in on the one hand and pRNFL/GCIPL on the other are INL volume is strongly associated with inflamma- clearly demonstrated by the negative correlation tion of the optic nerve, and to a lesser degree with between the layers. This further substantiates the other clinical relapses. Therefore, INL volume may potential of pRNFL/GCIPL as a measure for be a valuable parameter for capturing inflammatory 8 www.sagepub.com/msjetc Balk et al. disease activity and may be considered as an out- P. Albrecht reports grants and non-financial support from come measure for MS and MSON treatment trials. Biogen; grants, personal fees and non-financial support from Allergan; personal fees and non-financial Acknowledgements support from Bayer; personal fees and non-financial sup- port from Merck; grants, personal fees and non-financial LJ Balk contributed to the conception and design of the support from Merz Pharmaceuticals; grants, personal fees study, acquisition and analysis of data, and drafting of the and non-financial support from Novartis; grants, personal manuscript. DC contributed to the acquisition of data and fees and non-financial support from Roche; grants, person- drafting of the manuscript. BK, HGZ, RB, RA, EHML, al fees and non-financial support from Teva; and grants, AUB, BSD, AVJ, PA, VK, JH, MP, RCN, LL, FP, OA, personal fees and non-financial support from Ipsen, out- XM, LJ Balcer, PV, OO and TK contributed to the acqui- side the submitted work. sition and preparation of data and critically revised the V. Koska has nothing to declare. manuscript. AP contributed to the conception and design J. Havla has received speaker honoraria, travel expenses, of the study, analysis of data, and drafting of the manu- and personal compensations from Merck, Roche, Santhera, script. All authors have given final approval of the submit- Biogen, Bayer Healthcare, Sanofi Genzyme and ted version of the manuscript. Novartis Pharma. M. Pisa has nothing to declare. Conflict of interests R.C. Nolan has nothing to declare. L.J. Balk has received an institutional research grant from L. Leocani reports grants, personal fees and non-financial Teva and is a researcher for the OCTIMS study, an obser- support from Merck; personal fees and non-financial sup- vational study for validating OCT as a marker for MS port from Roche; personal fees from Teva; grants, personal sponsored by Novartis. fees and non-financial support from Biogen; non-financial D. Coric has nothing to declare. support from Almirall; and grants and non-financial sup- B. Knier has received research grants from Novartis and port from Novartis, outside the submitted work. travel expenses from Novartis, Merck and Bayer. F. Paul served on the steering committee for the Novartis H.G. Zimmermann has received a research grant from OCTIMS study and MedImmune; has received speaker Novartis and speaking fees from Teva. honoraria and travel funding from Bayer, Novartis, R. Behbehani has nothing to declare. Biogen Idec, Teva, Sanofi-Aventis/Genzyme, Merck R. Alroughani has received honoraria as a speaker and for Serono, Alexion, Chugai and MedImmune; is an academic serving on scientific advisory boards for Bayer, Biogen, editor for PLoS One; is an associate editor for Neurology: Neuroimmunology & Neuroinflammation; has consulted GSK, Merck, Novartis, Sanofi-Genzyme and Roche; and for Sanofi Genzyme, Biogen Idec and MedImmune; and research grants from Biogen, Novartis and has received research support from Bayer, Novartis, Sanofi-Genzyme. Biogen Idec, Teva, Sanofi-Aventis/Genzyme, Alexion, E.H. Martinez-Lapiscina reports personal fees from Merck Serono, German Research Council, Werth Biogen, Genzyme, and Novartis; travel reimbursement Stiftung of the City of Cologne, German Ministry of from Roche; grants from Instituto de Salud Carlos III – Education and Research (BMBF Competence Network (CM13/00150; MV15/00012; JR16/0006; MV17/00021; Multiple Sclerosis) and Arthur Arnstein Stiftung Berlin. PI17/01228), Fundacio Marato TV3 (20142030), GMSI O. Aktas has received grants from the German Research (2016), Fundacio Cellex Barcelona, Sanofi-Genzyme, Foundation (DFG), Eugène Devic European Network (EU- Novartis, Sanofi Genzyme and Roche, outside the submit- FP7), German Ministry of Education and Research, and ted work; is a member of the working group of the Schaufler Foundation; honoraria for lectures from IMSVISUAL; and is a researcher for the OCTIMS study. Almirall, Novartis, Bayer, Genzyme, Teva, Merck A.U. Brandt is named as inventor on several patent appli- Serono, Biogen, Roche and Medimmune; and received cations describing serum biomarkers in MS, motor func- travel/accommodation/meeting expenses from Novartis, tion assessments using perceptive visual computing, and Bayer Schering and Merck Serono. methods for analysing retinal images. He is cofounder of X. Montalban has received personal compensation for shareholder of the companies Motognosis and Nocturne, activities with Biogen, Celgene, Sanofi Genzyme, the latter of which has an interest in using OCT for Merck, Novartis, Roche and Teva Pharmaceuticals. MS diagnosis. L.J. Balcer has nothing to declare. B. Sánchez-Dalmau has received compensation for con- P. Villoslada is currently an employee of Genentech; this sulting services or speaking honoraria from Novartis and work was conducted before this affiliation and indepen- received research grants from Sanofi-Aventis; and is a dently of the company. He holds stocks in Bionure Inc, researcher for the OCTIMS study. Spire Bioventures, Mintlabs and Health Engineering; is A. Vidal-Jordana receives support for Juan Rodés con- academic editor of Multiple Sclerosis and Demyelinating tracts (JR16/00024) from Fondo de Investigaciones Diseases; and is on the executive committee of the Sanitarias, Instituto de Salud Carlos III, Spain, and has European Association of Systems Medicine. received compensation for consulting services or speaking O. Outteryck reports grant for research from Novartis; honoraria from Novartis, Roche and Sanofi-Aventis. grants and personal fees from Biogen-Idec; and funding www.sagepub.com/msjetc 9 Multiple Sclerosis Journal—Experimental, Translational and Clinical for travel from Biogen-Idec, Genzyme-Sanofi, Merck- predicting disease activity in MS. Neurology 2007; Serono, Novartis and Teva Pharmaceutical Industries, out- 68: 1488–1494. side the submitted work. 6. Balk LJ, Steenwijk MD, Tewarie P, et al. T. Korn reports compensation for travel expenses from Bidirectional trans-synaptic axonal degeneration in Merck and speaker honoraria from Novartis. the visual pathway in multiple sclerosis. J Neurol A. Petzold reports that the VUmc Multiple Sclerosis Neurosurg Psychiatry 2015; 86: 419–424. Center Amsterdam participated in the OCTIMS study 7. Green AJ, McQuaid S, Hauser SL, et al. Ocular and the PASSOS study, which were sponsored by pathology in multiple sclerosis: Retinal atrophy and Novartis, and the centre has received research support inflammation irrespective of disease duration. Brain for OCT projects from the Dutch Multiple Sclerosis 2010; 133: 1591–1601. Society. AP’s research was supported by the National 8. Balk LJ, Twisk JW, Steenwijk MD, et al. A dam for Institute for Health Research (NIHR) Biomedical retrograde axonal degeneration in multiple sclerosis? Research Centre based at Moorfields Eye Hospital J Neurol Neurosurg Psychiatry 2014; 85: 782–789. National Health Service (NHS) Foundation Trust and 9. Gelfand JM, Nolan R, Schwartz DM, et al. University College London Institute of Ophthalmology. Microcystic macular oedema in multiple sclerosis is The views expressed are those of the authors and not nec- associated with disease severity. Brain 2012; essarily those of the NHS, the NIHR, or the Department 135: 1786–1793. of Health. 10. Saidha S, Sotirchos ES, Ibrahim MA, et al. Microcystic macular oedema, thickness of the inner nuclear layer of the retina, and disease characteristics Funding in multiple sclerosis: A retrospective study. Lancet This research received no specific grant from any funding Neurol 2012; 11: 963–972. agency in the public, commercial, or not-for-profit sectors. 11. Knier B, Schmidt P, Aly L, et al. Retinal inner nuclear layer volume reflects response to immunotherapy in ORCID iD multiple sclerosis. Brain 2016; 139: 2855–2863. Lisanne J Balk https://orcid.org/0000-0002-3915-0052 12. Polman CH, Reingold SC, Banwell B, et al. Raed Alroughani https://orcid.org/0000-0001-5436- Diagnostic criteria for multiple sclerosis: 2010 revi- 5804 sions to the McDonald criteria. Ann Neurol 2011; Alexander U Brandt https://orcid.org/0000-0002-9768- 69: 292–302. 13. Schippling S, Balk LJ, Costello F, et al. Quality con- 014X trol for retinal OCT in multiple sclerosis: Validation Angela Vidal-Jordana https://orcid.org/0000-0002- of the OSCAR-IB criteria. Mult Scler 2015; 7270-5507 21: 163–170. Letizia Leocani https://orcid.org/0000-0001-9326-6753 14. Tewarie P, Balk L, Costello F, et al. The OSCAR-IB Orhan Aktas https://orcid.org/0000-0002-2020-9210 consensus criteria for retinal OCT quality assessment. PLoS One 2012; 7: e34823. 15. Cruz-Herranz A, Balk LJ, Oberwahrenbrock T, et al. Supplemental Material The APOSTEL recommendations for reporting quan- Supplemental material for this article is available online. titative optical coherence tomography studies. Neurology 2016; 86: 2303–2309. 16. Petzold A, Wattjes MP, Costello F, et al. The investi- References gation of acute optic neuritis: A review and proposed 1. Petzold A, Balcer LJ, Calabresi PA, et al. Retinal layer protocol. Nat Rev Neurol 2014; 10: 447–458. segmentation in multiple sclerosis: A systematic 17. Martinez-Lapiscina EH, Arnow S, Wilson JA, et al. review and meta-analysis. Lancet Neurol 2017; Retinal thickness measured with optical coherence 16: 797–812. tomography and risk of disability worsening in multi- 2. Saidha S, Al-Louzi O, Ratchford JN, et al. Optical ple sclerosis: A cohort study. Lancet Neurol 2016; coherence tomography reflects brain atrophy in multi- 15: 574–584. ple sclerosis: A four-year study. Ann Neurol 2015; 18. Kaufhold F, Zimmermann H, Schneider E, et al. Optic 78: 801–813. neuritis is associated with inner nuclear layer thicken- 3. Dorr J, Wernecke KD, Bock M, et al. Association of ing and microcystic macular edema independently of retinal and macular damage with brain atrophy in mul- multiple sclerosis. PLoS One 2013; 8: e71145. tiple sclerosis. PLoS One 2011; 6: e18132. 19. Burggraaff MC, Trieu J, de Vries-Knoppert WA, et al. 4. Gordon-Lipkin E, Chodkowski B, Reich DS, et al. The clinical spectrum of microcystic macular edema. Retinal nerve fiber layer is associated with brain atro- Invest Ophthalmol Vis Sci 2014; 55: 952–961. phy in multiple sclerosis. Neurology 2007; 20. Brandt AU, Oberwahrenbrock T, Kadas EM, et al. 69: 1603–1609. Dynamic formation of macular microcysts indepen- 5. Sepulcre J, Murie-Fernandez M, Salinas-Alaman A, dent of vitreous traction changes. Neurology 2014; et al. Diagnostic accuracy of retinal abnormalities in 83: 73–77. 10 www.sagepub.com/msjetc Balk et al. 21. Balk L, Killestein J, Polman C, et al. Microcystic mac- 26. Denniston AK, Keane PA, Aojula A, et al. The ocular ular oedema confirmed, but not specific for multiple glymphatic system and idiopathic intracranial hyper- sclerosis. Brain 2012; 135(pt 12): e226; author tension: Author response to “Hypodense Holes and the reply e227. Ocular Glymphatic System”. Invest Ophthalmol Vis 22. Lujan BJ and Horton JC. Microcysts in the inner Sci 2017; 58: 1134–1136. nuclear layer from optic atrophy are caused by retro- 27. Nolan R, Gelfand JM and Green AJ. Fingolimod treat- grade trans-synaptic degeneration combined with vit- ment in multiple sclerosis leads to increased macular reous traction on the retinal surface. Brain 2013; volume. Neurology 2013; 80: 139–144. 136: e260. 28. Dinkin M and Paul F. Higher macular volume in 23. Petzold A. Retinal glymphatic system: An explanation patients with MS receiving fingolimod: Positive out- for transient retinal layer volume changes? Brain come or side effect? Neurology 2013; 80: 128–129. 2016; 139: 2816–2819. 29. Warner CV, Syc SB, Stankiewicz AM, et al. The 24. Spaide RF. Retinal vascular cystoid macular edema: impact of utilizing different optical coherence tomog- Review and new theory. Retina 2016; 36: 1823–1842. raphy devices for clinical purposes and in multiple 25. Wostyn P, De Groot V, Van Dam D, et al. Age-related sclerosis trials. PLoS One 2011; 6: e22947. macular degeneration, glaucoma and Alzheimer’s dis- 30. Coric D, Petzold A, Uitdehaag BMJ, et al. P1084 ease: Amyloidogenic diseases with the same glym- Updates in OCT segmentation software influence lon- phatic background? Cell Mol Life Sci 2016; gitudinal assessment of retinal atrophy. Mult Scler 73: 4299–4301. 2017; 23(3 suppl): 563–564. www.sagepub.com/msjetc 11