Spectrum of Movement Disorders in Hematological Malignancies: A Comprehensive Systematic Review of Clinical Phenotypes, Mechanisms, and Outcomes

Study Design

This systematic review was conducted according to the PRISMA 2020 guidelines and adhered strictly to a predefined methodological protocol. All methodological elements, including construction of the research question, eligibility criteria, search strategies, study selection procedures, data extraction framework, classification systems, and quality appraisal methods, had been finalized prior to initiation of the review. The review protocol was prospectively registered in PROSPERO with registration number CRD420251168610 [13].

Research Question and PICO Framework

The research question was formulated using the Population-Intervention-Comparator-Outcome paradigm. The population of interest consisted of human patients diagnosed with hematological malignancies as classified by the WHO. The exposure of interest was the onset of a movement disorder temporally or mechanistically attributable to the hematologic disease or its treatment. A formal comparator group was not required because the evidence base comprised descriptive case literature. The outcomes of interest included the movement disorder phenotype, the underlying pathophysiological mechanism, associated clinical and diagnostic features, therapeutic interventions, and neurological or hematological outcomes.

Eligibility Criteria

Reports were eligible for inclusion if they presented original clinical data involving one or more cases of movement disorders occurring in the context of a confirmed hematological malignancy. Eligible hematologic diagnoses included lymphoid neoplasms, myeloid neoplasms, plasma-cell neoplasms, and histiocytic or dendritic-cell neoplasms. Reports needed to provide a clearly described movement disorder and sufficient information to classify the hematologic condition. Case reports, case series, and observational cohort studies were included without restrictions on language, year of publication, or geographic region. Studies were excluded if they lacked original patient-level data, involved non-hematologic tumors, or described movement disorders fully attributable to unrelated degenerative, genetic, toxic, nutritional, or metabolic conditions. Editorials, reviews without extractable cases, and animal or in-vitro studies were also excluded.

Information Sources

Multiple electronic databases were searched systematically up to 13 October, 2025. These databases included PubMed, Embase, Scopus, and Google Scholar. Google Scholar was screened to the first 50 pages because of its broad coverage and the high prevalence of non-peer-reviewed material beyond that threshold. In addition to these electronic sources, reference lists of included articles and relevant narrative reviews were examined manually. English translations were obtained for non-English articles when needed.

Search Strategy

A comprehensive Boolean search strategy was developed and implemented exactly as described in the protocol.

Definition

Movement disorders were categorized based on the standard clinical criteria of the International Parkinson and Movement Disorder Society, encompassing parkinsonism, chorea, dystonia, tremor, myoclonus, ataxia, tics, stereotypies, and other hyperkinetic or hypokinetic extrapyramidal syndromes [14]. Each patient was counted only once. If more than one movement disorder was present, the case was classified according to the main or predominant movement disorder. Other associated movements were noted but not counted separately.

Study Selection Process

All references identified through database searches were imported into EndNote 21 (Clarivate Analytics, Philadelphia, United States) for duplicate removal. Two reviewers (AJ and VS) independently screened titles and abstracts, and full texts were obtained for all eligible studies. Full-text articles were then reviewed independently by two reviewers (RKG and SP) to determine final inclusion, with disagreements resolved by consensus. The study selection process followed PRISMA guidelines and is presented in a PRISMA flow diagram.

Data Extraction Procedures

Data were extracted by two reviewers (RK and VP), with uncertainties resolved through consensus with a third reviewer (SS). A standardized 30-variable template was used to capture detailed patient-level information. Demographics, hematologic diagnosis per WHO classification, disease stage, and available molecular or cytogenetic data were recorded. Neurological data included the specific movement-disorder phenotype, timing, progression, associated features, and antibody results. Magnetic resonance imaging (MRI), computed tomography (CT), positron emission tomography–computed tomography (PET-CT), and cerebrospinal fluid (CSF), electrophysiology, and laboratory findings were documented when reported. Hematologic treatments and neurological therapies were extracted, along with outcomes, relapse patterns, mortality, follow-up duration, and time to neurological improvement whenever available.

Classification as per Pathogenesis

Each case was classified into one of four mechanisms: paraneoplastic immune-mediated, direct CNS infiltration, treatment related, or secondary to systemic complications. Movement disorders were also categorized by phenotype, including parkinsonism, tremor, dystonia, chorea, myoclonus, opsoclonus–myoclonus, ocular flutter, cerebellar ataxia, gait disorders, or mixed presentations.

Quality Assessment

The methodological quality of all included case reports and case series was evaluated using the Murad et al. assessment framework, which examines four key domains: selection, ascertainment, causality, and reporting. Each case was classified as good, fair, or poor based on the number of domains satisfactorily addressed [15]. Quality Assessment was done by three reviewers (RKG, SP and SS). This task was performed by consensus of the three reviewers.

Data Synthesis

Because the available evidence consisted predominantly of case reports and case series with substantial clinical heterogeneity, meta-analytic pooling was not feasible. Extracted data were summarized using descriptive statistics. Categorical variables were reported as counts and percentages, and continuous variables as mean, median, range, and interquartile range when available. Cases were grouped by movement-disorder phenotype and mechanism, and neuroimaging patterns, associated features, treatments, and outcomes were analyzed within these categories for the current review.

Lymphoid Malignancy

Among the 152 lymphoid malignancy cases, movement disorders and CNS involvement formed the core neurological manifestations. Cerebellar ataxia was the predominant movement disorder and occurred in 84 cases, which represented 55.3 percent of the cohort. Parkinsonism was reported in 28 cases or 18.4 percent, while chorea and choreoathetosis were present in 18 cases or 11.8 percent. Myoclonus was documented in 14 cases or 9.2 percent, dystonia in 11 cases or 7.2 percent, and tremor or dyskinesias in 7 cases each, accounting for 4.6 percent. Rare movement disorders included stiff-person spectrum disorders in 6 cases or 3.9 percent and opsoclonus-myoclonus-ataxia in 4 cases or 2.6 percent. Many patients had more than one movement disorder phenotype, reflecting mixed cerebellar and extrapyramidal involvement. (Table 1 and Supplementary Table 1)

CNS involvement was also substantial. Direct CNS infiltration by lymphoma was documented in 51 cases or 33.6 percent, while paraneoplastic or autoimmune CNS injury accounted for 46 cases or 30.3 percent. Only 55 cases or 36.1 percent lacked malignancy related CNS involvement. Cognitive impairment occurred in 48 cases or 31.6 percent and cranial nerve abnormalities in 42 cases or 27.6 percent. CSF analysis showed abnormalities in approximately 70 percent of tested patients, including inflammatory pleocytosis in 54 cases or 35.5 percent, malignant cells in 11 cases or 7.2 percent, and intrathecal IgG synthesis in 18 cases or 11.8 percent. Neuroimaging revealed CNS pathology in the majority, with white matter changes seen in 32 cases or 21.1 percent and basal ganglia or brainstem lesions in 28 cases or 18.4 percent. Together, these findings demonstrated that lymphoid malignancies produced extensive cerebellar, extrapyramidal, and immunological involvement within the CNS. (Table 1 and Supplementary Table 1)

Neurological manifestations in lymphoid malignancies were primarily treated with definitive tumor-directed therapy, most commonly systemic chemotherapy (54.6 percent), rituximab-containing regimens (37.5 percent), high-dose methotrexate–based therapy (27.0 percent), and radiotherapy (25.0 percent), while stem-cell transplantation was required in a minority (19.1 percent). Adjunct symptomatic neurological therapy was selectively used and included corticosteroids (27 percent), antiepileptic drugs (20 percent), levodopa or other dopaminergic agents (19 percent), benzodiazepines (16 percent), antipsychotics or dopamine-blocking agents (11 percent), anticholinergics (6 percent), intravenous immunoglobulin (22 percent), and plasmapheresis (8 percent), whereas supportive care alone was sufficient in 32 percent and no neurological treatment was reported in 26 percent. Targeted therapies were uncommon overall, with rituximab used in 19 percent and other biologics each in about 1 percent, and most patients did not undergo transplantation (86.8 percent). Clinical outcomes were mechanism-dependent: marked or complete neurological improvement occurred in 41 percent, partial improvement in 28 percent, stabilization in 12 percent, and no improvement or worsening in 19 percent. Collectively, these findings indicate that neurological prognosis in lymphoid malignancy–associated movement disorders is largely determined by reversibility of the underlying pathophysiology and response to oncologic therapy rather than symptomatic neurologic treatment alone. (Table 1 and Supplementary Table 1)

Myeloid Malignancy

Across 78 reported cases of myeloid hematological malignancies, hyperkinetic movement disorders formed the dominant clinical spectrum, with generalized chorea documented in 32 patients (41 percent) and hemichorea or hemiballismus in 17.9 percent, establishing choreiform syndromes as the hallmark presentation. Cerebellar ataxia occurred in 19.2 percent, parkinsonism in 9 percent, and myoclonus in 7.7 percent, while rarer phenomena such as neuromyotonia, myorhythmia, and blepharospasm collectively accounted for 3.8 percent. (Table 2 and Supplementary Table 2)

Structural central nervous system leukemia was distinctly uncommon at 2.6 percent, and most patients, 71.8 percent, showed no demonstrable structural central nervous system involvement. Instead, pathophysiology was largely secondary, with drug induced neurotoxicity responsible for 17.9 percent of cases, vascular complications for 6.4 percent, and infectious etiologies for 3.8 percent. These data indicate that neurological manifestations in myeloid malignancies are typically mediated by systemic metabolic, vascular, immune, or treatment related mechanisms rather than direct leukemic infiltration. (Table 2 and Supplementary Table 2)

Neurological manifestations in myeloid hematological malignancies were most frequently managed by treating the underlying hematologic disorder, most commonly with phlebotomy ± aspirin (34.6 percent) or hydroxyurea-based therapy (25.6 percent). Specific neurological therapy was unnecessary in 42.3 percent of cases, indicating strong reversibility once systemic pathology was corrected. When symptomatic treatment for movement disorders was required, it most often consisted of anti-chorea or dopamine-blocking agents (12.8 percent), immunotherapy (11.5 percent), or targeted symptomatic modalities such as levodopa (6.4 percent) and botulinum toxin (2.6 percent). Overall, 78.2 percent of patients demonstrated neurological improvement, including complete resolution in 29.5 percent, marked improvement in 26.9 percent, and partial improvement in 21.8 percent. HSCT-related syndromes represented a minority (allogeneic 12.8 percent, autologous 1.3 percent), and targeted therapies were rarely required (2.6 percent), underscoring that most neurological complications are secondary phenomena responsive primarily to correction of the hematologic disorder. (Table 2)

Plasma-Cell Neoplasms

In the 22 reported cases of plasma-cell neoplasms associated with movement disorders. Parkinsonism was the leading movement-disorder phenotype and was documented in 10 cases, representing 45.5 percent of the cohort. Cerebellar syndromes or ataxia appeared in 6 cases or 27.3 percent, while tremor disorders were recorded in 4 cases or 18.2 percent. A further 6 cases, also 27.3 percent, showed other movement-disorder presentations including tardive-dyskinesia–like syndromes, multifocal myoclonus, dystonia, restless legs syndrome, stiff-person syndrome, and non-specific involuntary movements. Mixed movement disorders were common and often coexisted with peripheral or central neurological deficits. (Table 3 and Supplementary Table 3)

CNS involvement varied in mechanism and severity. Twelve cases, or 54.5 percent, had no detectable CNS infiltration. Immune or paraneoplastic processes were present in 5 cases or 22.7 percent, and infectious CNS involvement, including opportunistic pathogens, was observed in 2 cases or 9.1 percent. Three cases demonstrated other CNS pathology such as leukoencephalopathy or degenerative processes. CSF abnormalities were documented in 31.8 percent of patients, with infectious positivity in 9.1 percent. Neuroimaging was abnormal in nearly half of the cases, with cerebellar pathology in 22.7 percent, white-matter signal changes in 13.6 percent, and cortical or degenerative patterns in 9.1 percent. Overall, plasma-cell neoplasms produced a distinct movement-disorder profile dominated by parkinsonism, with CNS involvement arising from immune, infectious, or structural mechanisms rather than direct myeloma infiltration. (Table 3 and Supplementary Table 3)

Across 22 cases, systemic therapy included CAR-T and bortezomib regimens (each 22.7 percent), lenalidomide or melphalan regimens (each 18.2 percent), and surgery or radiotherapy (9.1 percent), with autologous transplantation in 36.4 percent. Neurologic treatment comprised immunotherapy in 31.8 percent, symptomatic drugs in 27.3 percent (levodopa, amantadine, propranolol, primidone, clonazepam, DBS/thalamotomy), and drug withdrawal in 18.2 percent (lenalidomide, thalidomide, pregabalin or co-trimoxazole). Outcomes showed complete resolution in 27.3 percent, marked improvement in 31.8 percent, partial improvement in 22.7 percent, and progression in 13.6 percent, with drug-induced (31.8 percent) and immune-mediated (27.3 percent) mechanisms predominating. (Table 3)

Cohort study

In a cohort of 232 CNS lymphoma patients, 20 (9%) had prodromal neuropsychiatric symptoms, and Parkinsonism appeared in 5 of these 20 cases (25%). All five had deep grey-matter involvement, predominantly the basal ganglia and corpus callosum, explaining the dopaminergic dysfunction. Parkinsonism developed over weeks to months and often mimicked rapidly progressive atypical parkinsonian syndromes. Importantly, four of the five patients (80%) improved after lymphoma-directed therapy, indicating reversibility once the infiltrative lesion was treated [16].

Underlying Mechanisms

The pathogenetic patterns observed across the three WHO-defined hematologic malignancy groups demonstrate that movement disorders arise through distinct and biologically coherent mechanisms. In lymphoid neoplasms, two pathways predominated: direct central nervous system infiltration and immune-mediated paraneoplastic processes. The prominence of direct infiltration aligns with contemporary evidence showing that diffuse large B-cell lymphoma and primary CNS lymphoma exhibit marked neurotropism, frequently targeting the cerebellum, deep gray nuclei, and brainstem. Recent neuropathological and imaging-based studies confirm that even small-volume lymphomatous deposits in these regions can produce early cerebellar or extrapyramidal syndromes. The substantial proportion of immune-mediated cases is equally notable. Modern paraneoplastic immunology literature describes that autoreactive B-cell clones in lymphoid cancers can generate onconeural antibodies, leading to paraneoplastic cerebellar degeneration, opsoclonus–myoclonus, or basal ganglia dysfunction through antibody-driven synaptic and neuronal injury. Together, these converging findings emphasize that both structural invasion and dysregulated tumor-driven immunity are central to the pathogenesis of movement disorders in lymphoid malignancies [17].

In contrast, the movement disorders seen in myeloid neoplasms predominantly reflected systemic and metabolic mechanisms. Hyperviscosity, leukostasis, and cytokine-mediated injury disrupted basal ganglia microcirculation and produced high rates of chorea. The rarity of direct CNS leukemia further supports the concept that myeloid malignancies exert their neurological effects primarily through distal vascular or metabolic pathways, not direct parenchymal invasion [18].

Plasma-cell neoplasms demonstrated yet another pathogenesis mechanism. Amyloid deposition, immune disturbances, and treatment-induced dopaminergic vulnerability were major contributors. The predominance of parkinsonism suggests preferential injury to basal ganglia circuits, potentially through amyloid infiltration, autoimmune inflammation, or medication-related dopaminergic suppression. Infectious mechanisms, particularly in immunocompromised patients receiving proteasome inhibitors or steroids, also contributed to CNS dysfunction [19].

Drug-induced movement disorders are an important diagnostic consideration in hematological malignancies. They typically show a clear temporal relationship to drug initiation or dose escalation, with acute or subacute onset and improvement after drug withdrawal, and often present as symmetric parkinsonism, dystonia, dyskinesias, or cerebellar toxicity. In contrast, disease-related movement disorders usually have an insidious course, asymmetric or mixed phenotypes, and are associated with active malignancy. Neuroimaging supports this distinction, with characteristic toxic patterns favoring drug effects and structural or infiltrative lesions indicating disease-related mechanisms.

Diagnostic Implications

The diagnostic implications of these pathophysiologic differences are substantial. The strong association between lymphoid malignancies and cerebellar syndromes highlights the need for early MRI with high-resolution cerebellar sequences, CSF analysis, and comprehensive paraneoplastic antibody testing. Because paraneoplastic cerebellar degeneration may precede the detection of lymphoma, neurologists should maintain a low threshold to investigate occult hematologic disease when encountering unexplained subacute ataxia [20].

In myeloid neoplasms, the predominance of choreiform syndromes indicates that diagnostic evaluation should prioritize vascular imaging, serum viscosity indices, and inflammatory markers, rather than relying solely on CSF or MRI for direct infiltration. Recognition of drug-induced cerebellar or basal ganglia toxicity also necessitates careful review of chemotherapy regimens, particularly cytarabine, methotrexate, and tyrosine kinase inhibitors [21, 22].

Plasma-cell neoplasms require an integrated approach, as neuroimaging may reveal white matter changes, cerebellar atrophy, or basal ganglia signal abnormalities. Because amyloid and immune-mediated processes often coexist, CSF biomarkers, serum free light chains, PET-CT, and electrophysiology may all contribute critical information. The review demonstrates that a single diagnostic modality is insufficient, and optimal evaluation requires multimodal correlation tailored to the hematologic subtype [19, 23].

Treatment Implications

Treatment strategies differed significantly across malignancy groups due to their distinct mechanisms. In lymphoid neoplasms, management often required simultaneous tumor-directed therapy and immunotherapy, especially when autoimmune cerebellar degeneration or anti-neuronal antibodies were present. Rituximab-based regimens, IVIG, corticosteroids, and plasmapheresis demonstrated favorable outcomes when initiated early. The marked neurological improvement underscores the importance of prompt recognition [24, 25].

For myeloid malignancies, addressing the systemic drivers of movement disorders was critical. Therapies that reduced blood viscosity, controlled leukocytosis, or modified cytokine loads provided meaningful neurological improvement. Drug-induced syndromes often resolved with discontinuation or dose adjustment of causative agents. These findings highlight the need for vigilance when monitoring patients receiving high-dose cytarabine or targeted therapies [26].

In plasma-cell neoplasms, management often required a combination of myeloma-directed therapy, treatment of immune dysregulation, and symptomatic management of parkinsonism or cerebellar dysfunction. Dopaminergic agents were variably effective, depending on the mechanism. Cases with amyloid-associated movement disorders required systemic management of AL amyloidosis in addition to neurological care [19].

Across hematological malignancies, movement disorders showed mechanism-specific clinical patterns, with onset and phenotype reflecting direct central nervous system involvement, immune-mediated processes, systemic effects, or treatment exposure; neurological outcomes were often favorable when the underlying hematologic disease and pathophysiology were promptly addressed. In our review, treatment of movement disorders was primarily directed at the underlying mechanism. Cases due to direct CNS involvement improved mainly with tumor-directed therapy. Paraneoplastic syndromes often had limited neurological recovery despite immunotherapy. In myeloproliferative neoplasms such as polycythemia vera, hyperkinetic movements frequently improved after correction of hyperviscosity with hematologic treatment. Drug-related cases improved after withdrawal of the offending agent. Parkinsonism was uncommon in myeloid neoplasms, and levodopa benefit was not consistently documented, suggesting a limited role compared with treating the primary hematologic cause.

This review has several limitations. Most available evidence comes from isolated case reports and small series, creating inherent publication and reporting biases. Clinical details, investigations, and outcomes were inconsistently documented, limiting uniform comparisons. Pathogenic attribution often relied on author interpretation rather than standardized criteria. The lack of controlled studies prevents estimation of true frequencies or causal relationships. Diagnostic workups, imaging protocols, and treatment approaches varied widely across reports and time periods, reducing generalizability. Follow-up information was frequently incomplete, making long-term outcomes difficult to assess. A further limitation relates to treatment-related movement disorders. Although cases associated with immunotherapies, including immune checkpoint inhibitors, monoclonal antibodies, and chimeric antigen receptor T-cell therapy, were classified under the treatment-related mechanism category, they were not analyzed as a distinct quantitative subgroup across malignancy types.