# Etiological Heterogeneity in Sporadic ALS: A Survey of Proposed Subgroups and Trial-Tractable Populations

**Date**: 2026-05-04
**Scope**: Sporadic ALS (sALS), which accounts for ~90% of all cases and lacks an identified monogenic cause. This document reviews the strongest current hypotheses about internal heterogeneity within sALS, organized by biomarker-defined, clinically-defined, and exposure-defined subgroups. It concludes with an assessment of which sub-populations appear most tractable for focused clinical trials.

---

## 1. Proposed Subgroup Frameworks

### 1.1 Molecular / Transcriptomic Subtypes

The most rigorous evidence for biological heterogeneity within sALS comes from unsupervised clustering of gene expression data from motor cortex and peripheral blood.

**Three motor-cortex molecular phenotypes (Marriott et al., 2023, *Acta Neuropathol Commun*; PMID: 38129934)**
Using hierarchical clustering on 112 KCL BrainBank sporadic/motor cortex samples followed by validation in TargetALS US motor cortex (N=93) and Italian/Dutch blood datasets (N=15, N=397), the authors identified three reproducible molecular subtypes:
- **Synaptic and neuropeptide signalling cluster** – enriched for transcripts involved in synaptic transmission and neuropeptide pathways.
- **Oxidative stress and apoptosis cluster** – characterised by redox-stress and programmed cell-death signatures.
- **Neuroinflammation cluster** – distinguished by immune cell-type proportions and inflammatory transcript enrichment.

These signatures were ALS-specific (AUC 0.88 ± 0.10 vs controls) and motor-cortex-specific (perfect discrimination from occipital cortex and cerebellum). Importantly, each cluster tracked with distinct onset and progression-related clinical outcomes, supporting the hypothesis that different mechanisms drive pathogenesis in different patient subsets.

**Blood-based molecular subtypes (Grima et al., 2023, *Neurobiol Dis*; PMID: 39904421)**
In 96 Australian sALS cases analysed by whole-blood RNA-seq, clustering identified four peripheral subgroups with distinct immune-cell proportions and gene-expression profiles, reflecting divergent peripheral immune involvement. A 20-gene classifier distinguished sALS from controls with 78% accuracy (sensitivity 79%, specificity 75%). This suggests that peripheral molecular stratification is feasible, though prognostic value requires further validation.

**Phenotype-specific PBMC transcriptomics (Dragoni et al., 2025, *Neurobiol Dis*; PMID: 39904421)**
Analysis of PBMC RNA-seq from 48 unmutated sALS patients subdivided by clinical phenotype (Classic, Bulbar, Flail Arm, Flail Leg, Pyramidal) found largely phenotype-specific gene-expression changes, with only one shared upregulated gene across all phenotypes (Y3_RNA, an interferon-response component). This supports the view that sALS is not a single transcriptomic entity and that clinical phenotypes may map onto distinct molecular programmes.

---

### 1.2 Biomarker-Based Stratification: Neurofilaments and Progression Rate

**Neurofilament light chain (NfL) as a progression-stratification tool (Huang et al., 2020, *Ann Clin Transl Neurol*; PMID: 32515902)**
In a longitudinal NEALS Biofluid Repository study (108 PALS, 85 with ≥2 visits), patients were stratified into fast- and slow-progression groups by ALSFRS-R decline rate. Key findings:
- CSF and plasma NfL were significantly higher in fast progressors and in C9orf72 carriers than in slow progressors.
- Cytokines (MCP-1, IL-18) followed the same pattern.
- NfL levels were stable longitudinally, making them suitable for enriching trial cohorts and monitoring target engagement.

*NfL-based stratification is already clinically actionable*: it can identify a more homogeneous fast-progression cohort in whom a neuroprotective signal can be detected in a shorter timeframe.

**Cryptic exon-derived peptides (Anjum et al., 2025, *Front Mol Biosci*; PMID: 40661315)**
UNC13A-derived cryptic peptides permit TDP-43 pathology-linked molecular stratification, distinguishing sporadic ALS from C9orf72-linked disease. This represents a blood-accessible molecular classifier tied to pathobiology rather than symptomatology.

---

### 1.3 Neurophysiological Subtypes

**EEG-derived subphenotypes (Dukic et al., 2021, *Brain*; PMID: 34791079)**
Resting-state high-density EEG and data-driven clustering (similarity network fusion + spectral clustering) identified four reproducible subphenotypes in 95 ALS patients, characterised by distinct patterns of network disruption:
- Somatomotor α-band synchrony impairment
- Frontotemporal β-band activity and γL-band synchrony disruption
- Frontoparietal γL-band comodulation changes

These clusters correlated with distinct clinical profiles and predicted different trajectories/outcomes. Stability was confirmed on repeat EEG sessions. The approach offers a non-invasive, functionally grounded stratification method that maps onto underlying neurobiology rather than symptom topography.

---

### 1.4 Clinical Phenotype Subgroups

**Bulbar-onset vs limb-onset (Kim et al., 2017, *PLOS ONE*; PMID: 28095425)**
Voxel-based morphometry in 62 sALS patients showed markedly different atrophy patterns:
- **Limb-onset**: atrophy largely confined to motor cortex and adjacent sensorimotor regions.
- **Bulbar-onset**: more widespread grey-matter loss involving bilateral frontotemporal, left superior temporal and supramarginal gyri.

Bulbar-onset patients had shorter survival and greater functional impairment; ALSFRS-R correlated with more extensive cortical atrophy. This is consistent with the long-standing observation that bulbar-onset ALS carries a worse prognosis and may represent a partially distinct biological process or more aggressive form of the same process.

**Other recognised clinical phenotypes**
The eight standard ALS phenotypes (classic, bulbar, pyramidal, pure LMN, flail arm, pure UMN, flail leg, respiratory) show variable survival and may have different underlying biology. Dragoni et al. (2025) showed that these phenotypes carry distinct PBMC transcriptomic signatures in unmutated sALS, suggesting that clinical stratification is not merely descriptive but captures molecular divergence.

---

### 1.5 Pathological Subtypes (Even in Sporadic Disease)

Although ~90% of sALS cases clinically lack a family history, neuropathological classification reveals that the vast majority (~97%) exhibit TDP-43 proteinopathy. A minority are TDP-43-negative and may harbour currently unidentified FUS-like or SOD1-like pathology. While SOD1 and FUS mutations are typically familial, C9orf72 repeat expansions—the most common genetic cause—are found in 6–10% of apparently sporadic cases (van Rheenen et al., 2012; Perrone & Conforti, 2020; PMIDs: 22843265, 32497448), blurring the familial/sporadic distinction. This means a meaningful fraction of "sporadic" disease is actually monogenic.

---

### 1.6 Inflammatory Endotype

**Post-hoc trial evidence for an inflammatory subgroup (Miller et al., 2022, *Muscle Nerve*; PMID: 35098554)**
NP001 (a macrophage-activation regulator) failed in its primary Phase 2B endpoint when analysed across the whole cohort (138 participants). However, a post-hoc age-stratified analysis identified a **40–65-year-old subset** in which NP001-treated patients showed a **36% slower ALSFRS-R decline** and **51% less vital-capacity loss** versus placebo. A significantly greater proportion of non-progressors were in the NP001 arm (p = 0.004). Combined with earlier Phase 2A data showing benefit in participants with higher baseline hs-CRP, this supports the existence of an **inflammation-driven sALS subset** that responds to immunomodulation.

**Microglial dysfunction as a stratification platform (Quek et al., 2022, *J Neuroinflammation*; PMID: 35227277)**
Monocyte-derived microglia-like cells from 30 sALS patients recapitulated TDP-43 pathology, impaired phagocytosis, altered cytokine profiles, and abnormal morphology. Critically, rapid-progressor microglia-like cells showed specific phagocytic impairment, elevated DNA damage, and **NLRP3 inflammasome activation**, suggesting a mechanistic pathway that could define an inflammatory endotype responsive to anti-inflammatory agents.

---

### 1.7 Environmental Exposure-Defined Subgroups

Meta-analytic evidence supports several environmental/occupational associations with ALS risk, raising the possibility of exposure-specific etiological pathways:

| Exposure | Summary Effect | Key Reference |
|----------|---------------|---------------|
| Solvents (env/occupational) | OR 1.31 (95% CI 1.11–1.54) | Zhang et al., 2023 (*Neurol Sci*; PMID: 36897461) |
| Lead | OR 1.46 (95% CI 1.16–1.83) | Meng et al., 2020 (*Neurol Sci*; PMID: 31578652) |
| Pesticides | RR 1.35 (95% CI 1.02–1.79) | Gunnarsson & Bodin, 2019 (*IJERPH*; PMID: 30691095) |
| Electromagnetic fields | RR 1.26 (95% CI 1.07–1.50) | Gunnarsson & Bodin, 2019 |
| Air pollution (PM, NO2, NOx) | No association (HR ~1.0) | Chalitsios et al., 2026 (*Neurology*; PMID: 41931746) |
| Persistent organic pollutants (PCBs, OCPs) | Mostly inverse/null; suggestive HCB/HCH association | Tang et al., 2025 (*Environ Health Perspect*; PMID: 40488711) |

**Assessment**: Solvent, lead, and pesticide exposures constitute the strongest environmental hypotheses. However, exposure-defined subgroups have been explored mainly in epidemiological studies, not as trial-enrichment strategies. The biological mechanisms linking specific exposures to ALS remain poorly characterised, and there is no current biomarker that tags an "exposure-positive" sporadic subgroup.

---

## 2. Genetic "Dark Matter" Within Sporadic ALS

A critical source of heterogeneity in sALS is the **cryptic genetic liability** hidden within apparently sporadic cases:

- **C9orf72 expansions**: Found in 6.1% of Dutch sALS (van Rheenen et al., 2012); 8.8% of Norwegian ALS, with half the carriers being clinically sporadic (Olsen et al., 2025; PMID: 39316038); up to ~10% across many cohorts.
- **Common mutations in ALS genes collectively**: Present in up to 10% of apparently sporadic cases (Perrone & Conforti, 2020).
- **Multiple risk-variant co-occurrence**: In ~4.1% of sporadic cases vs 1.3% of controls, with significant enrichment of C9orf72 carriers co-harbouring ATXN2, NIPA1, or SMN1 variants (Dekker et al., 2016; PMID: 26777436).

**Implication**: "Sporadic" is partly an artefact of incomplete family-history assessment and variable gene-penetrance. Any trial in sALS that does not systematically genotype participants risks diluting a treatment signal across monogenic and truly non-monogenic subgroups.

---

## 3. Assessment of Evidence and Contradictions

| Subgroup Framework | Strength of Evidence | Key Limitations / Contradictions |
|-------------------|---------------------|----------------------------------|
| **Motor-cortex molecular subtypes (3 clusters)** | Strong – validated across 3 independent cohorts and 2 tissue types (brain + blood) | Requires post-mortem or research-level RNA-seq; clinical utility depends on blood-proxy validation and standardisation |
| **Blood RNA-seq subtypes (4 clusters)** | Moderate – single discovery cohort, independent blood validation needed | Prognostic value unclear; technical complexity limits deployment |
| **NfL fast/slow progressor stratification** | Strong – replicated across multiple studies; commercially measurable | Describes rate, not mechanism; enrichment shortens trial duration but does not address pathophysiology-specific therapy |
| **EEG subphenotypes (4 clusters)** | Moderate-high – stable on repeat assessment, predictive of outcome | Requires specialised equipment and analysis pipelines; not yet externally validated in a prospective trial context |
| **Bulbar vs limb onset** | Strong prognostic discrimination | Likely captures continuum rather than discrete biology; overlap in atrophy patterns exists |
| **Inflammatory endotype (NP001/CRP)** | Moderate – supported by Phase 2A CRP enrichment and post-hoc Phase 2B age-stratified analysis | Post-hoc nature; needs prospective validation; no consensus inflammatory biomarker panel for trial enrichment |
| **Environmental exposure subgroups** | Moderate (solvents, lead, pesticides); weak/null (air pollution) | Exposure assessment is retrospective and subject to recall bias; no biological mechanistic validation in human tissue |
| **Hidden monogenic disease (C9orf72, etc.)** | Strong | Challenges the sporadic/familial dichotomy itself; supports mandatory broad-panel genetic testing in all ALS trials |

---

## 4. Trial-Tractable Sub-Populations

Based on the evidence reviewed, four subgroups stand out as **currently tractable for focused trials**:

### 4.1 NfL-Enriched Fast Progressors
- **Rationale**: NfL is an accessible, validated prognostic biomarker. Enrolling participants with elevated baseline NfL (or faster predicted progression) shortens trial duration, reduces sample size requirements, and increases power to detect neuroprotective effects.
- **Status**: Already used in trial design (e.g., tofersen, several anti-inflammatory programmes). Does not require new technology.
- **Limitation**: Stratifies by prognosis, not mechanism, so therapy must have broad applicability.

### 4.2 UNC13A CC Genotype Subgroup (Pharmacogenomic)
- **Rationale**: Lithium carbonate showed survival benefit in exploratory meta-analysis of patients homozygous for the C-allele at SNP rs12608932 in UNC13A (Willemse et al., 2022; PMID: 36471413). The ongoing confirmatory European/Australian RCT targets exactly this genotype-restricted population.
- **Status**: First pharmacogenomics-driven ALS trial; if positive, it establishes a template for genotype-directed therapy in sALS.
- **Limitation**: Benefit proven only in post-hoc/exploratory analysis; confirmatory data pending.

### 4.3 Inflammatory Endotype (CRP-Elevated / NP001-Responsive Age Range)
- **Rationale**: NP001 data suggest an inflammatory subset exists within whom macrophage-directed therapy is beneficial. A prospective trial enriched for participants with baseline hs-CRP >1.13 mg/L and aged 40–65 years would directly test this hypothesis.
- **Status**: Phase 2B signals exist; requires a dedicated prospectively stratified trial.
- **Tractability**: High – CRP is routinely measured, and an age filter is trivial to apply.

### 4.4 Molecular Subtype–Matched Therapy (Research-Stage but Promising)
- **Rationale**: The three validated motor-cortex molecular subtypes (synaptic/neuropeptide, oxidative stress/apoptosis, neuroinflammation) offer a mechanism-based taxonomy. Patients in the oxidative-stress cluster, for example, might preferentially respond to edaravone or other antioxidant strategies; those in the neuroinflammation cluster to immune-modulating agents.
- **Status**: Requires retrospective drug-response analyses using existing trial biobanks, followed by prospective enrichment trials if signals emerge.
- **Tractability**: Medium-term – depends on implementing gene-expression or surrogate-biomarker profiling in trial workflows.

---

## 5. Summary and Recommendations

Sporadic ALS is not a single disease. The evidence supports the existence of **at least four non-mutually-exclusive dimensions of heterogeneity**:

1. **Mechanistic** (molecular transcriptomic clusters)
2. **Prognostic** (fast vs slow, NfL-defined)
3. **Pathophysiological** (inflammatory endotype, neurofilament signature)
4. **Cryptogenetic** (hidden C9orf72 and other monogenic cases within the sporadic label)

For clinical trial design, the most pragmatic immediate steps are:
- **Mandate broad genetic testing** (C9orf72, SOD1, FUS, TARDBP, and other major genes) in all sALS trials to exclude monogenic cases that may respond to mutation-specific therapies (e.g., tofersen for SOD1, antisense approaches for C9orf72).
- **Use NfL** to enrich for fast progressors and thus shorten trial duration.
- **Pursue prospective validation** of the inflammatory endotype using hs-CRP and/or age-enrichment strategies.
- **Build molecular profiling** (RNA-seq or targeted transcriptomic/blood biomarker panels) into next-generation trial platforms to enable mechanism-matched therapy assignment.

The shift from treating "sporadic ALS" as a uniform entity to recognising and exploiting its internal heterogeneity represents the most promising path forward for improving trial success rates and developing precision therapies in this devastating disease.

---

## References (Selected)

1. Marriott H, et al. "Unsupervised machine learning identifies distinct ALS molecular subtypes in post-mortem motor cortex and blood expression data." *Acta Neuropathol Commun* 2023; 11:190. PMID: 38129934
2. Grima N, et al. "RNA sequencing of peripheral blood in amyotrophic lateral sclerosis reveals distinct molecular subtypes." *Neuropathol Appl Neurobiol* 2023. PMID: 39904421
3. Dragoni F, et al. "Whole transcriptome analysis of unmutated sporadic ALS patients' peripheral blood reveals phenotype-specific gene expression signature." *Neurobiol Dis* 2025. PMID: 39904421
4. Huang F, et al. "Longitudinal biomarkers in amyotrophic lateral sclerosis." *Ann Clin Transl Neurol* 2020; 7:2013. PMID: 32515902
5. Dukic S, et al. "Resting-state EEG reveals four subphenotypes of amyotrophic lateral sclerosis." *Brain* 2022; 145:451. PMID: 34791079
6. Kim H-J, et al. "Relationship between Clinical Parameters and Brain Structure in Sporadic ALS Patients According to Onset Type." *PLOS ONE* 2017; 12:e0168424. PMID: 28095425
7. Miller RG, et al. "Phase 2B randomized controlled trial of NP001 in amyotrophic lateral sclerosis: Pre-specified and post hoc analyses." *Muscle Nerve* 2022; 66:52. PMID: 35098554
8. Quek H, et al. "ALS monocyte-derived microglia-like cells reveal cytoplasmic TDP-43 accumulation, DNA damage, and cell-specific impairment of phagocytosis." *J Neuroinflammation* 2022; 19:78. PMID: 35227277
9. van Rheenen W, et al. "Hexanucleotide repeat expansions in C9ORF72 in the spectrum of motor neuron diseases." *Neurology* 2012; 79:506. PMID: 22843265
10. Perrone B, Conforti FL. "Common mutations of interest in the diagnosis of amyotrophic lateral sclerosis." *Expert Rev Mol Diagn* 2020; 20:723. PMID: 32497448
11. Willemse SW, et al. "Lithium carbonate in ALS patients homozygous for the C-allele at SNP rs12608932 in UNC13A: protocol for a confirmatory trial." *Trials* 2022; 23:1011. PMID: 36471413
12. Zhang G, et al. "Environmental and Occupational solvents exposure and ALS: a systematic review and meta-analysis." *Neurol Sci* 2023; 44:1733. PMID: 36897461
13. Meng E, et al. "Population-based study of environmental/occupational lead exposure and ALS." *Neurol Sci* 2020; 41:1035. PMID: 31578652
14. Gunnarsson LG, Bodin L. "Occupational Exposures and Neurodegenerative Diseases—A Systematic Literature Review and Meta-Analyses." *IJERPH* 2019; 16:337. PMID: 30691095
15. Anjum F, et al. "Emerging biomarkers in amyotrophic lateral sclerosis." *Front Mol Biosci* 2025; 12:1608853. PMID: 40661315