Introduction: The Evolution of Prognostication in Anti-NMDAR Encephalitis
Anti-N-methyl-d-aspartate receptor (anti-NMDAR) encephalitis remains one of the most significant discoveries in neuroimmunology of the 21st century. While the disease is often devastating—characterized by psychiatric symptoms, cognitive decline, seizures, and autonomic instability—it is also remarkably responsive to immunotherapy. However, the clinical course is notoriously heterogeneous. Some patients achieve full recovery within months, while others face prolonged intensive care stays and permanent disability.
For years, clinicians relied on the original NEOS (anti-NMDAR Encephalitis One-year functional Status) score to estimate recovery. While useful, the clinical landscape has evolved, demanding more nuanced tools that can predict not just one-year disability, but also immediate response to treatment and long-term reintegration into society. The recently published NEOS2 study, an international cohort analysis, provides a sophisticated update to this prognostic framework, offering a validated model for the modern era of autoimmune neurology.
The Clinical Challenge: Predicting the Unpredictable
One of the primary difficulties in managing anti-NMDAR encephalitis is the timing of treatment escalation. First-line therapies, including corticosteroids, intravenous immunoglobulins, and plasma exchange, are standard. Yet, a significant subset of patients fails to respond, requiring second-line agents like rituximab or cyclophosphamide. The ability to identify these high-risk patients at the time of diagnosis—rather than after weeks of failed therapy—is a holy grail of precision medicine in this field.
Furthermore, patients and families often ask more than just “will they survive?” They want to know when they can return to their lives. The NEOS2 study addresses this by incorporating “return to school or work” as a primary long-term endpoint, moving beyond the crude measures of the modified Rankin Scale (mRS).
The NEOS2 Study: Methodology and Design
This international cohort study involved a massive collaboration across France, Germany, Japan, the Netherlands, and Spain. It included 702 patients with a definite diagnosis of anti-NMDAR encephalitis, confirmed via clinical criteria and cerebrospinal fluid (CSF) antibody testing.
The researchers used logistic regression to develop and validate multivariable models based on variables available at the time of diagnosis. The study was uniquely designed to predict outcomes across three distinct timeframes:
1. Short-term: Improvement in mRS two weeks after starting first-line treatment (NEOS2-T).
2. Mid-term: Functional outcome (mRS ≤2) at one year.
3. Long-term: Resumption of school or work within three years (NEOS2-W).
Identifying the Five Pillars of Prognosis
The NEOS2 score is built upon five independent predictors that were found to be statistically significant across the cohort. These variables are easily accessible to clinicians at the point of diagnosis:
1. Age at Diagnosis
Increasing age was associated with poorer outcomes (OR 0.35). While the disease primarily affects younger individuals, older patients often present with different paraneoplastic associations and may have less neuroplastic reserve for recovery.
2. Treatment Delay
The time between symptom onset and the initiation of immunotherapy remains a critical modifiable factor. A longer delay (OR 0.49) significantly decreased the likelihood of a favorable outcome, reinforcing the “time is brain” mantra in neuroimmunology.
3. Movement Disorders
The presence of orofacial dyskinesias, choreoathetosis, or dystonia at presentation (OR 0.32) serves as a clinical marker for more extensive neuronal dysfunction and potentially higher antibody burdens.
4. ICU Requirement
Patients requiring intensive care for ventilation or autonomic instability (OR 0.34) represent a more severe phenotype, which logically correlates with a more difficult recovery path.
5. CSF Leucocyte Count
An increased white cell count in the CSF at diagnosis (OR 0.65) indicates a more robust inflammatory response within the central nervous system, which was independently associated with poorer long-term results.
Key Findings: Predictive Accuracy Across the Spectrum
The study’s findings are robust, with the NEOS2 score demonstrating an accuracy (AUC) of approximately 80% for both one-year functional outcomes and three-year return-to-work rates.
Short-term Response: The NEOS2-T Model
Interestingly, the researchers found that the same variables (excluding age) were highly effective in predicting which patients would improve within just two weeks of first-line immunotherapy. The NEOS2-T model achieved an AUC of 81-84%. This is perhaps the most clinically actionable finding, as it allows clinicians to anticipate treatment failure early and consider rapid escalation to second-line therapies.
Long-term Success: NEOS2-W
At the three-year mark, 73% of the cohort had resumed work or school. The NEOS2-W model predicted this outcome with an AUC of 80%. The ability to provide families with a data-driven probability of returning to normal life is a significant advancement in clinical counseling.
Probability Nuance
Applied as an ordinal measure, the NEOS2 score allows for granular predictions. For instance, patients with the lowest scores had a 100% likelihood of a good one-year outcome and an 80% chance of early improvement. Conversely, those at the high end of the score spectrum had a 97% risk of first-line treatment failure and a 94% chance of failing to return to school or work within three years.
Clinical Implications: Tailoring Immunotherapy
The most significant impact of the NEOS2 score lies in its potential to guide aggressive early treatment. Currently, many protocols favor a “wait and see” approach for 10-14 days after first-line therapy. The NEOS2 score suggests that for patients with high scores at diagnosis, this delay may be counterproductive. These individuals might benefit from “front-loading” treatment—starting rituximab or cyclophosphamide concurrently with steroids and IVIg.
Furthermore, the score provides a standardized tool for stratifying patients in future clinical trials. By ensuring that treatment arms are balanced according to NEOS2 risk profiles, researchers can more accurately assess the efficacy of new therapeutic interventions.
Expert Commentary and Limitations
While the NEOS2 score is a powerful tool, it must be used as an adjunct to, not a replacement for, clinical judgment. The study is largely retrospective, covering a period from 2007 to 2022. During this time, standard of care evolved significantly, which may influence the weight of certain variables.
Additionally, while the score predicts “return to work/school,” it does not capture the quality of that return. Many patients who return to work still suffer from subtle cognitive deficits, executive dysfunction, or emotional lability that an mRS score or employment status might miss. Future iterations of prognostic tools should ideally incorporate neuropsychological testing and quality-of-life metrics.
Conclusion: Moving Toward Personalized Neuroimmunology
The NEOS2 score represents a major step forward in the personalized management of anti-NMDAR encephalitis. By utilizing five readily available clinical markers, it provides a clear window into a patient’s future—from their immediate response to steroids to their likelihood of returning to a career three years later. As we move toward a more proactive treatment paradigm, the NEOS2 score will likely become a staple in the neurology ward, helping clinicians optimize therapy and provide clear, evidence-based hope to patients and their families.
Funding and References
This study was supported by the Dioraphte charity (project 2001 0403).
Reference: Brenner J, Bastiaansen AEM, Guasp M, et al. Development and validation of the NEOS2 score for prediction of long-term outcomes and improvement after first-line immunotherapy in patients with anti-NMDAR encephalitis: an international cohort study. Lancet Reg Health Eur. 2025 Dec 11;62:101562. doi: 10.1016/j.lanepe.2025.101562.
