Introduction

Multiple sclerosis (MS) is the most common chronic inflammatory demyelinating disease affecting the central nervous system (CNS) of young adults in Western countries leading, in most cases, to severe and irreversible clinical disability. Magnetic resonance imaging (MRI) has an high sensitivity in detecting macroscopic tissue abnormalities in patients with MS. Conventional MR sequences (dual-echo, fluid-attenuated inversion recovery and T1-weighted imaging) provide important pieces of information for diagnosing MS [1–3], understanding its natural history, and assessing treatment efficacy [4, 5]. Unfortunately and despite this, standardization of MR procedures outside the setting of clinical trials (with regard to scanning protocol and frequency, sequence parameters, and outcome measures) is still lacking. Furthermore, in patients with established MS, the strength of the associations between conventional MRI findings and the clinical manifestations of the disease remains modest, at best. This is likely due to the low specificity of conventional MRI in the evaluation of the heterogeneous pathological substrates of the disease, its inability to provide an estimate of such a damage outside focal lesions, and the fact that it does not give information on the mechanisms through which the CNS recovers after tissue injury has occurred.

The identification of clinical, neuropsychological and imaging biomarkers characteristic of the entire spectrum of MS, and the definition of standardized methods for their collection and analysis are central not only to improve the understanding of disease pathophysiology and evolution, but also to generate research hypotheses, monitor treatment, and increase cost-effectiveness and power of clinical trials. The application of modern structural and functional MRI techniques to the study of MS patients is improving the understanding of the mechanisms responsible for the accumulation of irreversible clinical deficits in this disease. While structural MR techniques have allowed to quantify in vivo the extent and severity of disease-related damage in the different CNS compartments [6–10], the use of functional imaging techniques has highlighted that the presence and efficiency of brain plasticity might have a role in limiting the clinical consequences of MS-related tissue damage, at least at some stages of the disease [11, 12].

Despite having provided important pieces of information, the studies conducted so far in MS have several drawbacks, including the small samples of patients enrolled (which are representative of a limited range of clinical phenotypes), and the recruitment of selected groups of patients (for instance, without overt clinical impairment of the investigated systems). As a consequence, it remains to be established whether their results are robust enough to be considered representative of what really occurs in MS as a whole. Furthermore, advanced MRI techniques still require careful standardization, monitoring of scanner stability over time, and normative values as a reference. Therefore, despite the extensive use of these techniques in the research setting of highly specialized centers, their application in the assessment of MS patients in routine clinical practice has yet to be realized. In addition, a standardization of advanced MRI across different centers remains challenging. Should this be achieved, it would be possible to collect large MRI data sets of MS patients, who would enable generating and testing specific hypotheses.

Against this background, we promoted the Italian Neuroimaging Network Initiative (INNI), which involves centers and investigators with an International recognized expertise, with the major goal to determine and validate novel MRI biomarkers to be utilized as predictors and/or outcomes in future MS studies. In addition, INNI aims also to guide the application of MRI in MS at a national level.

The first two goals of this initiative were: (1) the creation of a web-based system with available clinical, neuropsychological and MRI data at the participating centers, to allow data sharing; (2) the use of such data to perform large-scale studies to define the role of clinical, neuropsychological and advanced imaging biomarkers in understanding MS pathophysiology. Subsequently, the INNI initiative will help to define standardized MRI and clinical protocols for the evaluation of patients with MS at a national level in Italy, allowing to integrate a large amount of data obtained from different centers. Responsible data sharing is in the public interest, however, it raises complex challenges. To help addressing these challenges and to answer people with MS ‘call to action’, the Italian MS Society Foundation, in line with its the Research Strategy Map [13], has promoted a data sharing research initiative. Within this framework, INNI will be instrumental towards an increasing uptake of personalized interventions for people with MS.

Here, we present the project, the centers involved, the structure and rules governing the initiative and the web-based system of clinical, neuropsychological and MRI data that has been implemented to allow data sharing.

Methods

The INNI project has been promoted by the Neuroimaging Study Group of the Italian Society of Neurology and is financially supported by a research Grant from the Fondazione Italiana Sclerosi Multipla (FISM 2013/S/1). FISM is the owner of the database, according to the Italian law on copyright. INNI currently involves four MS centers in Italy (Milan, Neuroimaging Research Unit, San Raffaele Scientific Institute; Rome, Department of Neurology and Psychiatry, Sapienza University; Naples, Department of Neurological Sciences, Second University of Naples/Neurological Institute for Diagnosis and Care ‘‘Hermitage Capodimonte”; Siena, Department of Medicine, Surgery and Neuroscience, University of Siena).

The first phase of the project was dedicated to legal issues and to obtaining approval from local Ethical committees at the founding sites. A Steering Committee (SC) was appointed, including representatives of the four promoting centers and of FISM. The SC ensures that the INNI project adheres to the study design and methodology laid out in the Grant submission. In the first project phase, the SC took care of: (1) defining inclusion/exclusion criteria and creating standard forms for the collection of clinical, neuropsychological and MRI data; (2) supporting the creation of the online database where such data have to be uploaded; and (3) defining guidelines to regulate the access levels to the online database.

INNI database

Main inclusion/exclusion criteria

To be included in the database, subjects have to be healthy controls (i.e., subjects with no previous history of neurological, psychiatric, or medical disorders, and a normal neurological exam), or patients with clinically definite MS, a clinically isolated syndrome (CIS) suggestive of MS, or a radiologically isolated syndrome (RIS). For patients, a neurological evaluation within 6 months from the MRI scan is also required. Only adult subjects (i.e., age ≥18 years) are included in the database.

Neurological evaluation

The variables chosen for inclusion in the INNI neurological evaluation form are reported in Table 1. Such form includes a section with the main information about disease history (e.g., disease onset date, type of onset, date of evolution to progressive phenotypes, information about previous relapses, etc.) and one section with global and specific clinical disability scales. Disease-modifying and symptomatic treatments (with treatment start and end dates) can also be recorded into the system.

Table 1 Neurological information collected in the INNI database
Full size table

Neuropsychological assessment

The variables chosen for inclusion in the INNI neuropsychological evaluation form are reported in Table 2. In this form, it is possible to collect scores from the Brief Repeatable Battery of Neuropsychological Tests (BRB-N) [14], Brief International Cognitive Assessment of Multiple Sclerosis (BICAMS) [15], Wisconsin card sorting test (WCST) [16], Stroop test [17], MS quality of life-54 items (MSQoL-54) questionnaire [18], as well as information about fatigue [19, 20] and depression [21–23].

Table 2 Neuropsychological information collected in the INNI database
Full size table

The INNI online platform includes optional forms, where results of blood, evoked potentials and cerebrospinal fluid examinations can be entered.

MRI examination

Since advanced MR techniques [in particular, diffusion tensor (DT) MRI and resting state (RS) fMRI] particularly benefit from the use of high-field magnets, it was decided to include in the online database only MRI scans acquired at 3.0 T (at least in the initial phase of the project).

Based on surveys collected at the participating sites and of the cumulative experience of the SC members, it was decided that the MRI protocol had to include (minimum requirements): (1) sequences for lesion quantification [dual echo (DE) or T2-weighted/fluid-attenuated inversion recovery (FLAIR) scans] acquired with axial orientation and a slice thickness of no more than 3 mm; (2) sequences for atrophy quantification acquired using high-resolution 3D T1-weighted scans [24]; (3) DT MRI sequences acquired with ~30 diffusion-weighted direction and a nearly isotropic spatial resolution [25]; (4) RS fMRI sequences covering all brain, with at least 140 scans and an acquisition session at least 5 min long [26].

Additional advanced MRI sequences (such as double inversion recovery, magnetization transfer MRI, susceptibility weighted imaging, etc.) can be uploaded, whenever available.

Creation of the INNI online database

The INNI online database was developed in collaboration with the consortium GARR, the Italian network for research and education (http://www.garr.it). The INNI platform allows the central collection of subjects’ clinical, neuropsychological and MRI data by means of user-friendly interfaces. When uploaded into the system, DICOM files are automatically split into the different sequences composing the examination, and a semi-automatic assignment of each sequence to the appropriate label (e.g., DE, FLAIR, 3D T1, DT MRI, RS fMRI, etc.) is performed. No identifying patients’ information is stored in the INNI platform: patient data are assigned to a unique identification code (ID). To ensure subjects’ privacy, any personal information is also deleted from the DICOM files.

The online INNI database is available at: https://database.inni-ms.org. It was developed by the GARR consortium in 2015 and formally tested in November 2016. The database content is available for authorized users only, who received appropriate login and password. For each participating center, two profiles can access the INNI database: (1) the Site Administrator (site responsible and main contact person), and (2) the Data Manager, who is in charge for data upload.

Guidelines for the access to the INNI online data

The SC defined different profiles for centers who are willing to join the INNI initiative. According to their category (research or profit institutions) and according to the number of patients shared in the online platform, the center will be defined as “Research User”, “Research Contributor”, “Profit User” or “Profit Contributor”, with different restrictions on visibility of database content and access to the data. More details about these profiles can be found on the website http://www.inni-ms.org (Fig. 1).

Fig. 1
figure 1

Home page of the INNI online platform. Database functions are available only upon signing in with appropriate login and password

Full size image

Data analysis

On 13th January 2017, a query was run on database content, and an Excel sheet including all uploaded MRI scans was produced. To be included in this search, MRI data had to be coupled with a valid demographic/neurological assessment. We performed some descriptive analyses on this population using SPSS software version 23.0. The main subjects’ characteristics were reported as means ± standard deviations (SD) or frequencies for continuous and categorical variables, respectively.

Results

Neurological evaluation

Data from 1310 subjects with a baseline neurological evaluation have been uploaded in the INNI database. There are 908 patients with MS (304/604 males/females, mean age = 39.8 years, SD = 11.4 years) and 402 healthy controls (183/219 males/females, mean age = 40 years, SD = 15.3 years). The main demographic and clinical characteristics of these subjects are shown in Table 3. In details, there are 590 patients (65%) with relapsing-remitting (RR) MS, 139 (15%) secondary progressive (SP) MS, 58 (6%) benign MS (defined as a disease duration ≥15 years and an EDSS ≤3.0), 62 (7%) primary progressive (PP) MS, 53 (6%) CIS and 6 (1%) RIS patients. Figure 2 shows the frequency of compilation of the main neurological variables included in the online form. Of the 908 MS patients included in the database, 90 (10%) have only a baseline neurological evaluation, while 817 (90%) have at least one follow-up neurological evaluation (total number of follow-up examinations = 838, median follow-up time = 1.04 years, range = 14 days–7.5 years).

Table 3 Main demographic and clinical information of healthy subjects and patients with multiple sclerosis (MS) collected in the INNI database
Full size table
Fig. 2
figure 2

Frequency of compilation of the main neurological (a) and neuropsychological (b) scores of the INNI database in patients with multiple sclerosis (MS) and healthy controls. EDSS Expanded Disability Status Scale score, FS functional systems, AI ambulation index, FT finger tapping, 9HPT nine hole peg test, T25FW timed 25-foot walk, FSS fatigue severity scale, MFIS modified fatigue impact scale, MADRS Montgomery–Asberg depression scale, CMDI Chicago multiscale depression inventory, BDI Beck depression inventory; STAI state-trait anxiety inventory, SRT selective reminding test, LTS long term storage, CLTR consistent long term retrieval, SPART spatial recall test, SDMT symbol digit modalities test, PASAT paced auditory serial addition test, WLG word list generation, MSQoL MS Quality of life—54 items (p = physical; m = mental), CVLT California verbal learning test, BVMT-R brief visuospatial memory test—revised, WCST Wisconsin card sorting test

Full size image

Neuropsychological assessment

Two-hundred and two healthy controls (50.2%) and 865 MS patients (95.2%) underwent at least one neuropsychological evaluation. The frequency of compilation of the main neuropsychological variables in the online form is reported in Fig. 2. Most frequently collected tests were the paced auditory serial addition test (PASAT) and symbol digit modalities test (SDMT) scores of the BRB-N battery, and information about fatigue. Depression scores were collected in the majority of MS patients, using different depression scales across sites. Ninety-five healthy controls (47%) and 810 MS patients (93%) have at least one follow-up neuropsychological evaluation (total number of follow-up examinations = 652, median follow-up time = 1.18 years, range = 4 days–7.5 years).

MRI examination

MRI scans were all acquired using 3.0 T scanners (Milan and Siena: Intera and Achieva, respectively, Philips Medical Systems, Best, The Netherlands; Rome: Magnetom Verio, Siemens, Erlangen, Germany; Naples: Signa HDxt, GE Healthcare, Milwaukee, USA). The parameters of the main MRI sequences used at each site are summarized in Table 4.

Table 4 Parameters used for the acquisition of the main MRI sequences included in the INNI database
Full size table

Of the 1310 subjects with a baseline MRI scan, 116 healthy controls (29%) and 467 MS patients (51%) have at least one follow-up examination (total number of follow-up scans = 1087, median follow-up time = 1 year, range = 2 days–7.5 years). All main MRI sequences were acquired in the large majority of subjects, as shown in Fig. 3.

Fig. 3
figure 3

Frequency of acquisition of the main MRI sequences of the INNI database in patients with multiple sclerosis (MS) and healthy controls, at baseline and follow-up examinations. DE dual echo, DT MRI diffusion tensor MRI, RS fMRI resting state functional MRI, DIR double inversion recovery

Full size image

Discussion

In this paper, we present the INNI initiative, a network created by four centres leading the neuroimaging research of MS in Italy, with the major goal to define standardized methods of collection and analysis of advanced MR imaging techniques, and to identify clinical, neuropsychological and imaging biomarkers characteristic of the entire spectrum of MS. The online platform for the central collection of data is now ready, and includes a rather large population of control subjects and MS patients. These subjects have a homogeneous neurological and neuropsychological evaluation, and the MRI acquisition was performed with a similar strategy across sites. Major scanner upgrades have been codified into the database, to perform proper adjustments of the MRI analysis produced by using INNI data. Therefore, future analysis run on subjects from the INNI database are likely to be powerful, accurate and representative of the general MS population. The large number of subjects included in the database will also allow to select samples with homogeneous and particular characteristics (to address specific research questions) or to perform studies on rare disease phenotypes (which are usually underrepresented in single center studies).

The next step of the INNI initiative will be to use the data available in the database for specific research projects. Planned, short-term research projects include: (1) investigation of functional abnormalities within RS networks related to cognition and correlation with structural damage within cognitive-related tracts and cognitive performance; (2) exploration of abnormalities of the sensorimotor RS network and their association with clinical and MRI variables; (3) assessment of RS functional connectivity alterations in the main brain functional networks according to T2 lesion volume (high vs low) and physical/cognitive disability scores (high vs low) to explore the functional substrates of the extremes of the clinical-MRI paradox observed in MS; (4) quantification of global and regional distribution of white matter and gray matter lesions, with the creation of lesion probability maps of white matter and cortical lesions, and correlation between lesion location in the different white matter/gray matter structures and dysfunction of selected brain networks at rest.

In addition to performing specific research studies, future aims of the INNI network will be the creation of a standardized protocol of acquisition of advanced structural and functional MR techniques, to be applied for the study of patients with MS. This will allow to homogenize the approach to MS patients at a national level. This will require the circulation of a questionnaire to all the centers which have an interest to the project to collect information concerning scanners and coils, as well MRI parameters currently employed for the study of patients with MS. Then, after considering the sequences that are at present used at each center, a common acquisition protocol will be defined. In a first phase, only sequences for the quantification of T2 lesions and atrophy (DE, FLAIR, 3D T1-weighted MRI) will be standardized. Advanced MRI sequences (DT MRI, RS fMRI, DIR and MT MRI) might be applied initially only at sites predisposed to or currently using these techniques, and also for these sequences a common protocol will be designed. The common standardized protocol will include some suggestions for periodical quality assessment (QA) of MRI images, to monitor MRI scanners performances and avoid image quality deterioration with time. If required, peripheral centers will be helped in the set-up of the protocol, through a dummy run procedure. The data acquired with the standardized protocol will be made available in the INNI online database for future projects. These dataset will represent an invaluable source for future studies of predictors of the disease.

Another long-term goal of the INNI initiative will be instructing neurologists involved in the cure of MS patients in the use, evaluation and interpretation of information derived from advanced MRI techniques. This will imply to design standardized and centralized procedures for daily-life implementation of advanced MRI measures. Providing harmonization of procedures and allowing selection for evidence-based therapeutic strategies, will play a key role for a better management of the disease.