Multiple sclerosis (MS) is a neurological autoimmune disease characterized by axonal demyelination leading to neurodegeneration. It affects over 2.5 million people worldwide, representing the primary cause of neurological disability in young adults (1). MS usually debuts as a clinically isolated syndrome (CIS) and shows varied clinical courses (2, 3), traditionally classified as relapsing-remitting MS (RRMS), secondary progressive MS (SPMS, as an evolution of the RRMS), and primary progressive MS (PPMS, only 10-15% of overall cases).

Early diagnosis is crucial in managing MS, yet remains challenging due to lack of specific biomarkers (4). Diagnosis typically involves magnetic resonance imaging (MRI) to detect and measure brain lesions, along with analysis of oligoclonal IgG bands in sera and cerebrospinal fluid (CSF) and evoked potentials (5). However, these techniques are invasive and costly, underscoring the need for new biomarkers in MS. MS misdiagnosis is common and a recent report found that near 20% patients with an established diagnosis of MS did not fulfill contemporary McDonald Criteria and had a more likely alternate diagnosis (6).

Osteopontin (OPN) emerges as a potential biomarker candidate. This multifunctional cytokine and adhesion molecule is involved in bone remodeling, wound healing, and immune cell activation (7, 8). OPN also plays a role in promoting a proinflammatory cytokine balance while inhibiting anti-inflammatory responses (9). OPN, produced by immune cells as well as neurons and glia, is therefore relevant in neuroinflammatory diseases like MS (10). Studies indicate that OPN levels are elevated in MS plaques dissected from patients’ brains, suggesting its involvement in disease pathogenesis (11). Additionally, OPN can suppress autoreactive T cells, potentially influencing immune-mediated diseases (12).

In the past two decades, over 25 therapies both injectable and oral were approved for RRMS, some with indications for progressive disease. However, patients with MS still experience progression of disability via neurodegeneration (13). Two alternative strategies have been proposed for MS treatment, namely escalation and early, or in some cases first-line treatment with high-efficacy agents. Escalation involves initiation of treatment with a drug of low or moderate efficacy, evaluation of treatment response and, if needed, switching to other medication. Safety concerns supported this approach, but the development of therapies with better safety profiles and the implementation of risk mitigation actions to limit side effects foster the use of highly effective therapies for early treatment (14). Patients initiating early treatment, regardless of prognostic factors and MRI burden at baseline, show significantly reduced disability progression (15). In fact, Natalizumab (NTZ), a therapy used in several diseases such as MS and Crohn’s disease (16), is a humanized monoclonal antibody targeted to the α4 integrin that diminishes T-cell trafficking to the CNS and reduces relapse rate. However, the effect of NTZ on OPN levels is not completely elucidated yet.

In summary, both early diagnosis and therapeutic intervention delay long-term disease progression, and the identification of diagnostic and response biomarkers is an unmet need. Given these observations, OPN seems to hold promise as a valuable biomarker for MS. Our aim is to conclusively evaluate the utility of OPN as a diagnostic biomarker in MS through systematic review of the literature and meta-analysis, and to assess the OPN potential aiding in NTZ treatment evaluation.

We initially identified 605 articles using our search strategy. After eliminating 87 duplicate entries, the first phase of selection involved screening articles based on their titles and abstracts. A total of 454 articles were excluded as they did not address our objectives, leaving 64 articles for a more detailed evaluation. Subsequently, they were selected for systematic review and meta-analysis. Among these, nine were discarded as they were not performed in humans and 26 lacked sufficient information for inclusion. Four out of the 29 selected articles allowed the evaluation of NTZ efficacy (Flow diagram as shown in Supplementary Figure 1 ).

Data were extracted and compiled in Tables 1 – 3 . Due to the similarity between HC and NOND groups, both were analyzed together. In the meta-analyses, levels of OPN in CSF, plasma, and serum samples were compared in MS patients with all types of controls. However, not all comparisons were feasible due to the limited availability of studies, particularly in serum samples. In addition, stratification by clinical forms and the effects of NTZ treatment on OPN levels were assessed. The results, including Chi-squared (χ²), I², and Standard Mean Difference (SMD) were documented in Figures 1 – 9 , as follows.

In the comparisons between overall MS and control groups, significant differences in OPN levels in CSF samples were observed only when comparing MS patients vs. HC and NOND ( Figure 1C , p = 0.0007), with OPN levels being higher in MS patients. Similar trends were found when comparing each clinical form to HC and NOND, with significant differences found in CSF in most cases (p = 0.0002 for RRMS, p = 0.02 for PPMS, and p < 0.0001 for SPMS, Figures 2C , 3B , 4C , respectively). Even though the heterogeneity among studies was very high in the comparison of PPMS ( Figure 3B , I² = 82%), patients consistently showed increased OPN levels in all the studies included. The difference was also evident when comparing CIS with HC and NOND in serum samples ( Figure 5C , p = 0.05), with a similar trend in CSF ( Figure 5B , p= 0.07). However, these comparisons could not be evaluated in plasma samples, due to substantial heterogeneity among studies ( Figures 1D , 2E , 4D ). Regarding individuals with OND, the differences emerged in the comparison to each MS clinical form independently. Comparing plasma samples, RRMS patients displayed lower levels than controls ( Figure 2B , p= 0.002), and similar trends were found when CSF samples of OND were compared to PPMS ( Figure 3A ,p= 0.0004) and CIS ( Figure 5A , p = 0.004). Significant differences were not detected when overall MS were considered ( Figures 1A, B in CSF and plasma, respectively). Again, the high heterogeneity in CSF (I² = 73%) hampered this evaluation, although three out the four studies included presented higher CSF levels in MS patients ( Figure 1A ).

In the comparison among clinical forms, significant differences were observed between RRMS and both PPMS ( Figures 6B, C : p= 0.02 in plasma and p= 0.0004 in serum) and SPMS ( Figures 6D,E : p= 0.03 in CSF and p= 0.0004 in plasma, with high heterogeneity but consistently displaying increased values in SPMS in the latter). No significant differences were found between the progressive forms ( Figure 7 ). Lower levels of OPN were detected in CIS compared to MS clinical forms ( Figure 8 ), with significant differences observed in the comparisons with RRMS ( Figure 8A : p= 0.003 in CSF) and PPMS ( Figures 8C, D : p= 0.01 in CSF and p < 0.0001 in plasma). An exception was noticed in the comparison between OPN levels in serum of CIS and RRMS patients, with higher levels in people with CIS compared to RRMS ( Figure 8B : p= 0.0007).

Finally, a meta-analysis focused on Natalizumab (NTZ) treatment in MS patients was conducted ( Figure 9 ), specifically in plasma samples. When comparing OPN levels before and after a minimum of 6 months of NTZ treatment, a significant reduction in OPN levels was observed with the presence of NTZ ( Figure 9C : p = 0.002). After the 6 month-treatment, the comparison between OPN levels in RRMS patients and HC was not significant ( Figure 9B ), by decreasing the high OPN levels in pre-treatment patients and therefore canceling the significant difference initially observed with HC ( Figure 9A , p= 0.001).

OPN is a pleiotropic protein able to regulate numerous pathways by interacting with multiple receptors. Osuka et al. (49) showed that thrombin cleavage of the full-length secreted OPN generates C- and N-terminal fragments, which exhibit distinct biological activities derived of the exposure of regions not accessible in the full-length molecule. The overexpression of OPN correlate with autoimmune disease severity (50). Nonetheless in MS, a multifactorial condition with elevated OPN levels, evidence also supports its neuroprotective effects (51, 52). The present study conducted a systematic review and meta-analysis of OPN levels as biomarker in MS. A prior revision in 2018 (53) reported that OPN could serve as a diagnostic and prognostic biomarker with clinical utility. However, the authors did not address the influence of treatments such as NTZ on OPN levels. Additionally, from 2018 several new studies on OPN levels were performed and, therefore, our study aimed to evaluate the potential use of OPN as diagnostic and treatment biomarker in MS. Our main purpose with this work was to restrict the analyses to an accessible and well-established OPN quantification method (ELISA), which could be easily incorporated to the clinical setting when validated as diagnostic marker. The reviewed studies employed CSF, plasma, and serum as sample sources. Most of them analyzed two types of samples to provide complementary results. While CSF collection is more invasive as compared to plasma and serum, sometimes it yields a more sensitive detection and therefore, studies typically employed CSF alongside alternations between serum and plasma. The studies exhibited significant variability and heterogeneity, posing a considerable challenge to extract firm conclusions. Upon closer examination of individual articles, disparities in the study design and sample processing surfaced, emphasizing these critical considerations for potential biomarker development.

Given the inherent diversity in clinical symptomatology and disease courses in MS, our meta-analyses used a random-effects model, which postulates that the actual impact of one variable may vary across studies owing to their heterogeneity (54). Certain comparisons excluded some studies. For instance, although MS is more prevalent in women (3:1), the study by Kariya et al. (26) was omitted as it presents higher percentage of women than the other studies included in the comparison of CSF between MS and other neurological disorders (OND) and, even then, the heterogeneity remained high (changed: from 92% to 62%). On the contrary, the study by De Fino et al. (44) was excluded due to a low percentage of women in the comparison between CIS and RRMS, decreasing heterogeneity from 62% to 30% in CSF and from 71% to 0% in serum. Furthermore, in the comparison of CSF between MS and the combination of HC and NOND, the study by Edwards et al. (23) was disregarded due to the use of a different ELISA kit and heterogeneity fell from 88% to 0%. Nonetheless, the cause of heterogeneity among articles could not be ascertained in some cases. For example, the study of Wen et al. (33) was disqualified in the comparison of CSF between RRMS and the combination of HC and NOND, resulting in a change in heterogeneity from 90% to 54%. Additionally, the article by Atlintas et al. (31) was excluded from the comparison of plasma between PPMS and HC and NOND, reducing heterogeneity from 75% to 0%. Finally, the study by Comabella et al. (30) in plasma was eliminated in the comparisons between RRMS and PPMS (heterogeneity changed from 92% to 52%) and PPMS vs. SPMS (heterogeneity changed from 87% to 0%). Significant difference in CSF levels of OPN exists between individuals with MS and the combined group of HC and NOND, showing higher levels in the MS group ( Figure 1C ); however, when comparing OPN levels between MS and patients with OND, this significant difference disappears ( Figure 1A ). Even after careful analysis, in this case the heterogeneity could not be resolved ( Figure 1A , I² = 73%) as the four studies presented analogous weight, all but the latter were performed in British population (the last one in North American population), and the two last ones used the same ELISA kit. Therefore, no clear reason exists to eliminate any of them. Although the mentioned heterogeneity hampers a conclusive interpretation and in the three studies with higher size out of the four studies included the OPN levels are increased in MS patients, the overall results would suggest that higher OPN levels might be a shared characteristic in various neurological disorders, irrespective of the specific condition. Notably, neurological disorders like Alzheimer’s and Parkinson’s diseases exhibit altered OPN levels (55–58). Consequently, as comparison of OPN levels between people with MS and OND rendered no differences, this biomarker would seem to lack enough specificity for a differential diagnosis of MS when other neurological conditions are considered, and OPN should be combined with additional tests to gain specificity for MS diagnosis. In this sense, Kodosaki et al. (28) claimed that combinations of biomarkers were considerably better than single biomarker predictions and specifically the combination of CSF [chitinase-3-like-1 + TNF-receptor-1 + CD27] and serum [osteopontin + MCP-1] presented an area under the curve of 0.97 for MS diagnosis. Interestingly, when stratifying by clinical form, significant differences become evident between OND and PPMS patients but not with RRMS or SPMS ( Figures 3A , 2A , 4A , respectively), although these comparisons show the previously discussed heterogeneity among studies. Nonetheless, OPN levels would still act as a crucial diagnostic biomarker if able to distinguish the first episode of MS (CIS) from other neurodegenerative diseases. Our study suggests that high OPN levels could serve as potential biomarker for MS diagnosis, as they are higher in CIS CSF compared to OND individuals and to HC+NOND ( Figures 5A,C ), with two different studies included in each comparison. To advance in the validation of OPN as a potential biomarker for diagnosing MS at onset, next steps should emphasize the use of ELISA as a uniform methodology to ensure reliable comparisons and allow the definition of sensitivity and specificity for OPN detection.

Regarding the clinical forms, higher levels of OPN were detected in PPMS patients compared to RRMS in both serum and plasma samples ( Figures 6B, C ) and, even with high heterogeneity, SPMS showed higher plasma levels than RRMS ( Figure 6E ), suggesting that circulating OPN was elevated in both progressive MS forms, while RRMS showed higher OPN levels than SPMS in CSF ( Figure 6D ), as previously observed by Agah et al. (53). No significant differences were found between the progressive forms, PPMS and SPMS ( Figure 7 ). No actual MS treatment stops neurodegeneration (59), which accumulates as the disease progresses, and circulating OPN seems indicative of this process. In a study not considered in the present meta-analysis as it did not include a control group and used a multiplex assay, Nowak-Kiczmer et al. reported significant differences in serum content of OPN between PPMS and either SPMS or RRMS, with the highest values present in SPMS and the lowest in PPMS (60). These differences underscore the importance of considering not only the specific clinical forms when exploring biomarkers for MS, but also the sample source and OPN determination method. In summary, the difference in OPN levels between MS clinical forms may aid in their accurate distinction, something important to ascertain underlying mechanistic processes (61, 62) and therefore, to suggest therapeutic strategies. When compared to the specific clinical forms, OPN levels in CSF of CIS individuals with a first episode were significantly lower than those found in RRMS, SPMS or PPMS patients ( Figures 8A, F, C ), replicating results previously observed by Agah and cols (53). In fact, factors including aging and OPN have been reported to cause microglia to lose their protective states and become injurious (63, 64).

Finally, our study aimed to evaluate the effect of NTZ on OPN levels. OPN synergistically interacts with Vascular Cell Adhesion protein 1 (VCAM-1) and facilitates lymphocyte homing to the inflamed brain in a process driven by the α4β1 integrin, which binds both VCAM-1 and OPN (10). The blockade of this integrin with the humanized monoclonal antibody, Natalizumab, reduces the relapse rate in MS. All studies included in the meta-analysis used plasma samples, except for the one with CSF samples conducted by Romme-Christensen et al. (34), which was consequently excluded from our meta-analysis. Before NTZ treatment, MS patients showed significantly higher OPN levels in plasma than HC ( Figure 9A ), and MS patients normalized OPN levels to those found in HC after NTZ treatment (between 6 and 12 months, Figure 9B ). Notable differences emerged in MS patients following the administration of NTZ ( Figure 9C ), specifically after a minimum of 6 months of treatment, and our findings suggest that OPN could serve as a biomarker for evaluating the effectiveness of treatment, as expected provided the direct functional link of OPN with this MS therapy. In other MS therapeutic approaches, no sufficient number of studies associated OPN and treatment efficacy, although some isolated examples in certain therapies can be found (i.e (65, 66).

One of the caveats regarding the OPN evidence as a biomarker (as with many others), is the diversity of methodological approaches in the different studies. The present systematic review and meta-analysis considered the method of OPN determination, and only included studies that evaluated OPN levels through a widespread used method as ELISA. Despite this, up to seven different commercial and one in-house ELISA kits were counted in the reviewed studies, and a remarkable variability of reported levels was common among them. Before any possible implementation in the clinical setting, efforts should focus in the evaluation of the technique and laboratory tests that most accurately measure OPN levels in the diverse sample types, as discrepancies between OPN sample sources have been already described (61). To foster the use of OPN as a potential biomarker for MS, additional data are required. For instance, expanding the scope of the studies to include diverse ethnicities would enlarge the availability of study populations for comparison and facilitate the extrapolation of results beyond the Caucasian population.

Our work contributes with valuable insights to the ongoing efforts in identifying robust and clinically relevant biomarkers for MS, emphasizing the nuanced nature of the disease and the potential utility of OPN as additional discriminating factor for diagnosis. Further research to conclusively elucidate the specific roles and implications of OPN in the various clinical forms, will ultimately advance our understanding of MS pathophysiology and refine diagnostic approaches.