Multiple Sclerosis (MS) is a chronic disease involving the central nervous system. The clinical course is generally relapsing at onset, and later progressive [1]. Both the inflammatory and neurodegenerative component is present from the early stages [1]. While the first aspect has been well documented and is monitored in clinical practice using conventional magnetic resonance imaging (MRI), the study of neurodegeneration is more difficult [2]. At present, an important biomarker of the neurodegenerative aspect in MS is the evaluation of brain volume by non-conventional techniques of MRI [3]. Indeed, the loss of brain volume appears to be related to the progression of the disease and the permanent accumulation of disability [3, 4]. One of the most important symptoms related to brain atrophy is the deficit of cognitive functions [5]. This is associated with several MRI markers, in particular grey matter pathology, both in terms of focal lesions and volume loss [5]. Cognitive impairment, which has variable severity and has been identified at all stages of the disease and clinical courses, affects between 40 and 70% of MS patients [6]. It appears to be associated with age, disease duration and disability [7]. Many diagnostic tools are able to explore cognitive functions in MS patients [8], although the majority of these are time consuming and require qualified personnel [8]. However, a recently developed neuropsychological assessment named “Brief International Cognitive Assessment for Multiple Sclerosis” (BICAMS) can be administered by non-specifically trained healthcare staff in around 15 min [8].

In the last decades, optic coherence tomography (OCT), a new instrument for evaluating the neurodegenerative aspect of MS, has been widely studied [9, 10]. OCT is able to evaluate the thickness of the retinal nerve fibre layer (RNFL), a defined marker of axonal loss. RNFL measurement could differentiate MS patients from healthy subjects, and it is clearly related to brain atrophy, disease duration, and disability [10–14]. The most involved RNFL region is the temporal quadrant [15]. The great majority of studies exploring RNFL in MS patients have been performed using a ‘time-domain’ OCT (considered the ‘old’ OCT model), which produces less accurate images than the ‘spectral domain’ OCT (SD-OCT), especially in longitudinal studies [16–19].

BICAMS and OCT share three important qualities for MS assessment in a clinical setting: brevity, ease of use, and low cost.

The present study entails a cross-sectional and longitudinal assessment of the use of SD-OCT and BICAMS in the evaluation of the neurodegenerative aspects of MS. These procedures are compared with MRI brain atrophy measurements, which are currently considered the gold standard among neurodegenerative markers in MS patients.

MS-relapsing patients consecutively referred to the MS Centre of the University of Cagliari/Binaghi Hospital from March to July 2015 were prospectively included in the study. A group of healthy controls working in the Binaghi hospital was also recruited over the same time period. All the subjects signed an informed consent. The inclusion criteria for MS patients were: 18-65 years of age; relapsing remitting course; no relapses and/or steroids intake in the 30 days before enrolment and follow-up evaluation. All patients started a disease-modifying drug (DMD) prior to enrolment. Psychoactive drugs or substances that might interfere with neuropsychological performance were forbidden for both patients and healthy controls.

The ethics committee of the University of Cagliari approved the study.

At baseline, after written informed consent was given, all the subjects underwent neuropsychological assessment, brain MRI, and OCT. These three evaluations were carried out within a maximum period of three months. The evaluations were repeated for a group of MS patients after 12 months ±3. The changes from baseline and follow-up (difference between follow-up and baseline) were calculated for all of the collected MRI, OCT and BICAMS parameters collected.

Gender and age were collected for the whole cohort. In the MS group, the following clinical data were also recorded from medical records by a neurologist with expertise in MS: age at onset; clinical course; expanded disability status scale score (EDSS) at baseline; all DMD from six months before baseline until the end of the study; presence of relapses and steroid therapy in the 30 days before baseline and before the follow-up evaluation; and optic neuritis in the clinical history with indication of the affected eye. Eyes affected by previous optic neuritis were excluded from the statistical analysis.

We included 66 MS subjects (female: 48; 72.7%) and 16 healthy controls (female: 9; 56.25%). Mean age at baseline was 43.4 years (SD: 12) in patients, and 46.8 years (SD: 9) in healthy controls. Mean education was 11.6 years (SD: 4.16) in patients, and 14.3 years (SD: 4.19) in healthy subjects. In the MS group, the mean duration of the disease was 10.8 years and the median EDSS was 2 (range: 0 - 7.5). Ten patients were not included in the OCT study due to poor compliance. In 4 other patients, the OCT evaluation was made only in 1 eye, as the other eye was blind for reasons other than MS.

The comparison between healthy controls and patients at baseline showed higher NBV (p = 0.006), higher NWV (p = 0.003), and higher RNFL thickness (p = 0.014) in the first group. The healthy subjects group also demonstrated better performance at SDMT (p = 0.037). No significant differences were found in NGV and p-NGV.

Cross sectional analysis in MS patients at baseline

In Table 2 were reported all baseline correlations. A significant correlation between: p-NGV and SDMT (p = 0.022); NGV and SDMT (p = 0.013); NGV and BVMT (p = 0.048); TEMP-RNFL and NBV (p = 0.007); TEMP-RNFL and NWV (p = 0.012); TEMP-RNFL and NGV (p = 0.048); TEMP-RNFL and p-NGV (p = 0.021); PMB-RNFL and NBV (p = 0.013); PMB-RNFL and NWV (p = 0.02); PMB-RNFL and NGV (p = 0.049); PMB-RNFL and p-NGV (p = 0.032). The correlations between SDMT and p-NGV are shown in Fig. 1, while the correlations between TEMP-RNFL and MRI parameters in Fig. 2.

Neurodegenerative assessment	Variables	r	p-value
SD-OCT
Average-RNFL	SDMT	0.01	0.96
	CVLT-II	0.20	0.18
	BVMT-R	0.08	0.60
	NBV	0.25	0.095
	NWV	0.16	0.28
	NGV	0.22	0.14
	p-NGV	0.29	0.06
TEMP-RNFL	SDMT	0.01	0.95
	CVLT-II	0.09	0.57
	BVMT-R	0.08	0.58
	NBV	0.40	0.007
	NWV	0.37	0.012
	NGV	0.30	0.048
	p-NGV	0.36	0.021
PMB-RNFL	SDMT	0.044	0.78
	CVLT-II	0.09	0.55
	BVMT-R	0.13	0.38
	NBV	0.37	0.013
	NWV	0.35	0.02
	NGV	0.30	0.049
	p-NGV	0.33	0.032
SDMT	NBV	0.20	0.10
	NWV	−0.07	0.56
	NGV	0.21	0.11
	p-NGV	0.31	0.022
CVLT-II	NBV	0.09	0.48
	NWV	0.06	0.66
	NGV	0.04	0.78
	p-NGV	0.01	0.97
BVMT-R	NBV	0.11	0.38
	NWV	0.05	0.68
	NGV	0.18	0.17
	p-NGV	0.22	0.11

Average-RNFL thickness displayed a correlation with both NBV (p = 0.10), and p-NGV (p = 0.06) even thought it did not reach the statistical significance. The OCT measures did not correlate with the results of BICAMS tests.

Since p-NGV correlated with both OCT and cognitive performances, a multivariable model was implemented. After selection, SDMT and TEMP-RNFL, together with age and disease duration, were significantly associated with p-NGV (Table 3). The multivariable model increased R² from 38.3% (with age) to 48.5% with the inclusion of SDMT (∆R² of 10.2%), and to 65.2% with the inclusion of TEMP-RNFL (∆R² of 16.7%).

	B (95% CI); p-value	∆R2
Model 1
SDMT (T-score)	1.20 (0.46-1.95); p = 0.002	+ 0.102
Model 2
SDMT (T-score)	1.20 (0.44-1.95); p = 0.003
TEMP-RNFL	1.26 (0.35-2.16); p = 0.008	+ 0.167

Both models include age and disease duration at baseline

We found that the evaluation of RNFL thickness by SD-OCT and cognitive functions by BICAMS are able to differentiate the group of MS patients from that of healthy controls, as previously shown by numerous studies [8–11], as well as being independent markers of neurodegeneration. These were compared with brain volume evaluation performed with MRI - the most commonly used instrument for MS follow-up and what is generally considered the gold standard marker for monitoring the degenerative component of the pathology.

Our data confirmed the association between both assessments and brain volume measures [5, 10–14]. In particular, we found that the BICAMS tests correlate with NGV, especially with cortical atrophy, which is notoriously associated with cognitive decline [5]. Moreover, we found a significant correlation between specific OCT parameters (temporal quadrant and PMB thickness) and brain volume measurements. This correlation appears to be stronger with global and white brain volume than grey matter volume. This result is in line with previous studies in which OCT measures were found to be associated with total brain volume and normalized white matter volume [12, 24], or both normalized white and grey matter volumes [11].

An innovative aspect of our work was the combined analysis of brain atrophy, RNFL thickness, and cognitive functions. Indeed, while the association between brain volume and the other two parameters has been clearly assessed previously, and confirmed by our data, there is less evidence of a possible correlation between RNFL thickness and cognitive impairment. Toledo et al. performed a unique study exploring this issue [25]. They found a significant correlation between SDMT and RNFL thickness, both for the average and the temporal quadrant. However, this result was not confirmed by our work, and we did not observe any correlation between RNFL (average, temporal and PMB) and BICAMS tests. It is worth noting that Toledo et al. performed their OCT analysis with a smaller sample and using a less accurate time-domain machine.

The lack of association between RNFL thickness and cognitive impairment, as for their correlation with different brain volume regions (only grey matter for BICAMS tests, both grey and white matter for TEMP-RNFL and PMB-RNFL), leads us to hypothesize that RNFL and cognitive impairment may represent different aspects of neurodegeneration in MS. Indeed, cognitive decline is secondary to axonal loss, dendritic transection and apoptosis [26]. Moreover, as well as the grey matter pathology, cognitive impairment is mainly associated with primitive degeneration present from the initial stages of the disease, as entirely independent degenerative process, and only partially it is secondary to inflammation. [6, 27]. Otherwise, RNFL thickness may equally be associated with both primitive degeneration and degeneration due to an inflammatory component. In particular, the thinning observed by OCT may be due to clinical and subclinical episodes of ON, primary neurodegeneration of the retinal ganglion cells and their axons, or retrograde transsynaptic degeneration of the ganglion cells and their axons due to MS lesions in the posterior visual pathways [28].

Longitudinally, no significant changes in BICAMS tests scores emerged. This was probably due to a well-known practice effect [20].

A reduction was observed in the mean percentage of brain volume, and in all the OCT parameters considered. However, the longitudinal trend of the two measures did not correlate, perhaps due to the fact that they do not reflect the same specific inflammatory and degenerative brain damage in MS pathology. When considering the brain atrophy measure, we also need to take into account the well-known phenomenon of so-called ‘pseudo-atrophy’ [4], which hinders the correct evaluation of the degenerative and inflammatory components. Thus, some patients could have a ‘false’ light reduction of brain volume over the course of one year due to the presence of inflammatory oedema. To note that all the patients were clinically stable between baseline and follow-up, and no active lesions in brain MRI were observed.

The study was also subject to some limitations.

First, follow-up may have been too short to observe significant changes in the degenerative component, which is present since the initial stages of the disease and is characterized by slow progression.

Second, while our cohort of participants is of a comparable size to that of previous cross-sectional studies, it may be too small to provide clear longitudinal results. Moreover, patients were heterogeneous for the duration of disease, DMD, and grade of inflammation at MRI. However, the great majority of them were clinically stable during the observational period, and no optic neuritis was recorded between baseline and follow-up.

Our study shows that the evaluation of brain volume, cognitive functions, and RNFL thickness are different continuous measures of neurodegeneration. Unconventional MRI, with its significant costs and requirement for qualified personnel, is more difficult to use in all MS settings. BICAMS and OCT, however, which can be performed at low cost after minimal training, may be suitable for use in clinical practice to assess different aspects of neurodegeneration.