# Etiological Heterogeneity within Sporadic ALS: Current Hypotheses, Proposed Subgroups, and Trial Tractability

> **Date:** 2026-05-04  
> **Scope:** Review of proposed etiological subgroups within the ~90% of ALS cases classified as sporadic (sALS), with emphasis on biomarker, phenotypic, and exposure-based stratification, and assessment of evidence supporting or contradicting each.  
> **Key sources:** PubMed-indexed primary literature (2020–2026), UniProt, OpenTargets, Reactome.

---

## 1. Executive Summary

Sporadic amyotrophic lateral sclerosis (sALS) is the predominant clinical presentation (~90–92% of cases) and, by definition, lacks an identified monogenic causal mutation. However, the field has converged on the view that sALS is not a single disease entity but a syndrome comprised of multiple overlapping pathophysiological processes. The strongest current hypotheses propose **at least four broad stratification axes**:

1. **Clinical phenotype** (site of onset, rate of progression);
2. **Pathological signature** (TDP-43 proteinopathy distribution and patterns; non-TDP-43 pathology);
3. **Biomarker-defined biological clusters** (neuroaxonal injury, glial activation, tau phosphorylation);
4. **Exposure-linked/oligogenic subsets** (somatic mutations, environmental modifiers, and comorbidities).

Emerging evidence from multi-biomarker profiling, machine-learning stratification, and molecular neuropathology indicates that **some of these subgroups are already tractable for focused clinical trials**, particularly:

- **Lower-limb–onset (LLO) ALS** as a slower-progressing, more homogeneous trial population;
- **Biomarker-defined neurodegenerative clusters** (e.g., NF-light high vs. GFAP-driven glial activation) for mechanism-directed therapy;
- **TDP-43 proteinopathy–positive sporadic ALS** for anti-aggregation or proteostasis-restoring interventions;
- **Early-stage, neurofilament-elevation–confirmed disease** for neuroprotective or immunomodulatory trials enriched for rapid progression.

Below we assess each proposed subgroup, its evidential base, and its suitability for trial enrichment.

---

## 2. Clinical Phenotype–Based Stratification

### 2.1 Site of Onset: Bulbar vs. Spinal (and Upper vs. Lower Limb)

**Hypothesis:** The anatomical site of motor neuron degeneration first affected (bulbar, upper-limb, lower-limb) defines biologically distinct trajectories with differing prognoses and underlying vulnerabilities.

**Evidence supporting:**
- Large retrospective cohort studies consistently show that **lower-limb–onset (LLO) ALS has slower progression** than bulbar-onset (BO) or upper-limb–onset (ULO) disease. In a study of 281 patients (1,255 visits), LLO patients displayed slower decline of total ALSFRS-R, driven by slower deterioration in the motor subscale (Shovman et al., 2026, *J Clin Med*).
- Bulbar-onset ALS is associated with greater respiratory involvement, lower forced vital capacity (FVC), and more nocturnal desaturation (Riveiro et al., 2026, *Med Clin*).
- Patients with spinal-onset ALS who develop bulbar involvement represent an intermediate phenotype with distinct speech biomarker profiles (Tsujisawa et al., 2026, *Folia Phoniatr Logop*).

**Evidence contradicting /Limitations:**
- Longer diagnostic delay in LLO may confound the apparent slower progression (lead-time bias).
- The biology driving site-specific onset remains poorly understood; it may largely reflect regional vulnerability rather than etiological separateness.

**Trial tractability:** **HIGH.** LLO offers a more homogeneous, slower-progressing population that reduces noise in neuroprotective trials. Differentiating ULO from LLO is clinically straightforward and has been proposed explicitly as a stratification tool for clinical trial enrolment (Shovman et al., 2026).

### 2.2 Rate of Progression: Rapid vs. Slow Progressors

**Hypothesis:** sALS contains a bimodal (or multimodal) distribution of progression rates reflecting distinct biological drivers.

**Evidence supporting:**
- Multiple large datasets have identified slow and rapid progressor clusters using unsupervised learning. The distinction is robustly predicted by baseline ALSFRS-R slope, neurofilament levels, and neuroimaging measures.
- Plasma neurofilament light chain (NfL) and glial fibrillary acidic protein (GFAP) levels correlate with survival; combined biomarker panels show improved discriminative performance for rapid progression over single markers (Beers et al., 2026, *Ann Clin Transl Neurol*).

**Evidence contradicting /Limitations:**
- Progression rate is a continuous variable; arbitrary cutoff points may miss overlapping biology.

**Trial tractability:** **MODERATE–HIGH.** Using baseline ALSFRS-R slope + biomarker enrichment is already being adopted in trial design (e.g., VRG50635 trial using NfL and digital endpoints; Yusuf et al., 2026, *Amyotroph Lateral Scler Frontotemporal Degener*).

---

## 3. Pathological Signature–Based Stratification: TDP-43 Proteinopathy and Beyond

### 3.1 TDP-43 Proteinopathy Subtypes

**Hypothesis:** The overwhelming majority of sALS cases (>90–97%) exhibit TDP-43 pathology, but the distribution, morphology, and cell-type predominance of that pathology vary and may reflect distinct upstream mechanisms.

**Evidence supporting:**
- In sporadic ALS, phosphorylated TDP-43 (pTDP-43) pathology accumulates in behaviour-associated brain regions (amygdala) and is predictive of behavioural dysfunction with 86% sensitivity and 100% specificity (Rifai et al., 2026, *Brain Commun*).
- Variation in TDP-43 inclusion morphology (diffuse punctate cytoplasmic staining → round inclusions → skein-like inclusions) is associated with progressive loss of PML nuclear bodies, suggesting a maturation cascade in sporadic ALS (Mori et al., 2026, *J Neuropathol Exp Neurol*).
- Amygdala TDP-43 pathology correlates with ferritin accumulation, raising the possibility of an **iron-dyshomeostasis/endophenotype** linked to TDP-43 aggregation, which could be regionally detected by susceptibility-weighted MRI (Rifai et al., 2026).
- Cofilin hyperphosphorylation and consequent actin cytoskeleton disruption have been identified as a novel mechanism triggering TDP-43 pathology in sporadic ALS (Jagaraj et al., 2026, *Brain*). This suggests an **actin-dysregulation endotype** that could be pharmaceutically targetable.

**Evidence contradicting /Limitations:**
- TDP-43 pathology is near-universal in sALS, so “TDP-43–negative sporadic ALS” is rare and usually reclassified (e.g., SOD1 or FUS pathology on re-evaluation).
- The subtyping by distribution or morphology is still descriptive; causal upstream drivers are not yet confirmed.

**Trial tractability:** **MODERATE.** Anti-TDP-43 aggregation strategies (e.g., cofilin-pathway modulation, autophagy enhancement) are promising but await validated patient stratification tools beyond postmortem pathology. Cofilin-directed peptides represent an early exploratory target.

### 3.2 Non-TDP-43 Sporadic ALS: Mitochondrial and Oxidative Stress Endotypes

**Hypothesis:** A subset of sporadic ALS is driven primarily by mitochondrial dysfunction and oxidative stress rather than TDP-43 proteinopathy.

**Evidence supporting:**
- iPSC-derived motor neurons from sporadic ALS patients show altered excitability and viability profiles compared to SOD1-related ALS, suggesting non-genetic mitochondrial contributions (Wu et al., 2026).
- MR-spectroscopic imaging demonstrates that motor cortex and corticospinal tract glutathione (GSH) levels inversely correlate with time since diagnosis and imputed disease progression, defining an **oxidative-stress endophenotype** (Andronesi et al., 2020, *Front Neurol*).
- Plasma 4-hydroxy-2-nonenal (4-HNE), a lipid peroxidation marker, correlates with disease burden, progression, and survival—especially in bulbar-onset cases (Beers et al., 2026).

**Evidence contradicting /Limitations:**
- Mitochondrial and oxidative changes may be downstream consequences of neurodegeneration rather than primary etiological drivers.

**Trial tractability:** **MODERATE.** Antioxidant trials (e.g., edaravone) have shown modest efficacy in broader populations. The opportunity lies in **enriching trials for patients with low GSH / high 4-HNE** using MRI or blood biomarkers.

---

## 4. Molecular Biomarker–Defined Subgroups

### 4.1 Neurofilament Light Chain (NfL)–Driven Neuroaxonal Injury Group

**Hypothesis:** Elevated blood/CSF NfL identifies a subgroup with active, rapid axonal degeneration and predicts faster progression and shorter survival.

**Evidence supporting:**
- In a 2020 study, a panel of biochemical and structural imaging biomarkers (including NfL) defined a brain endophenotype that stratified ALS patients into more homogeneous therapeutic groups than clinical criteria alone (Andronesi et al., 2020, *Front Neurol*).
- Plasma NfL correlated with greater upper motor neuron (UMN) burden, as measured by transcranial magnetic stimulation and the Penn UMN Score (Senerchia et al., 2026, *Neurobiol Dis*).
- CSF NfL was significantly reduced following autologous stromal vascular fraction (SVF) cell therapy, paralleling potential neuroprotective effects (Li et al., 2026, *Front Aging Neurosci*).
- Speech-based digital endpoints from the VRG50635 trial correlated with plasma NfL, supporting its utility as a longitudinal progression anchor (Neumann et al., 2026, *Sci Rep*).

**Evidence contradicting /Limitations:**
- NfL is not ALS-specific; it also rises in multiple sclerosis, stroke, and FTD. Its prognostic value diminishes in slowly progressive disease.

**Trial tractability:** **HIGH.** NfL enrichment is already being implemented in clinical trial design to select rapid progressors and to stratify by biological severity.

### 4.2 Plasma pTau181–Selective Lower Motor Neuron Denervation Group

**Hypothesis:** Plasma phosphorylated tau at threonine 181 (pTau181) reflects a distinct biological process—lower motor neuron chronic denervation—separate from NfL-driven axonal injury.

**Evidence supporting:**
- A multi-biomarker study found that while NfL mapped to UMN burden, **pTau181 selectively reflected lower motor neuron (LMN) degeneration**, particularly chronic denervation severity (Senerchia et al., 2026, *Neurobiol Dis*).
- Latent profile analysis identified **three biologically distinct clusters**: (1) selective pTau181 elevation (LMN-dominant), (2) progressive NfL increase (UMN/axonal injury), and (3) prominent glial activation (GFAP-driven). Cluster membership independently predicted disease aggressiveness.

**Evidence contradicting /Limitations:**
- pTau181 may also reflect comorbid Alzheimer’s pathology; its specificity for LMN denervation in ALS still requires neuropathological correlation.

**Trial tractability:** **MODERATE–HIGH.** These clusters represent an immediately usable stratification framework for trials targeting axonal vs. glial mechanisms.

### 4.3 GFAP-Driven Glial Activation Cluster

**Hypothesis:** A biologically discrete subgroup of sALS is characterized by prominent astrocytic/microglial activation rather than primary motor neuron degeneration.

**Evidence supporting:**
- Latent profile analysis identified a cluster with prominent glial activation (GFAP elevation), which was independently associated with behavioural lability after age adjustment (Senerchia et al., 2026).
- Plasma sTREM2, a microglial activation marker, is elevated in ALS but lacks specificity vs. disease mimics; this suggests neuroimmune activation is present but nonspecific (Senerchia et al., 2026, *Neurobiol Dis*).
- In the autologous SVF trial, CSF GFAP decreased alongside NfL, suggesting that therapies modulating neuroinflammation may produce biomarker-level effects even before functional improvement (Li et al., 2026).

**Evidence contradicting /Limitations:**
- GFAP is strongly age-associated and may partly reflect normal aging glial changes.
- The glial activation cluster overlaps with behavioural/cognitive symptoms, complicating its separation from the ALS-FTD spectrum.

**Trial tractability:** **MODERATE.** Immunomodulatory or microglial-targeted therapies could be enriched for this cluster, but better specificity of glial activation markers (or combinations) is needed.

---

## 5. Genetic and Oligogenic Subsets Within "Sporadic" ALS

### 5.1 Somatic Mutations and Low-Level Mosaicism

**Hypothesis:** A fraction of apparently sporadic ALS arises from tissue-restricted or low-level somatic mutations in known ALS/FTD genes, which are missed by standard germline genetic testing.

**Evidence supporting:**
- Deep targeted sequencing of 88 neurodegeneration-related genes in postmortem brain and spinal cord from 399 sporadic cases identified deleterious somatic variants in 2.1% of cases lacking germline mutations (Zhou et al., 2026, *Nat Genet*).
- Variants in DYNC1H1 and LMNA were identified in sALS cases from RNA-seq data; one sFTD case harboured a de novo somatic C9orf72 repeat expansion.
- These somatic variants were focal, enriched in disease-affected regions, and present at very low allele fractions (<2%).

**Evidence contradicting /Limitations:**
- 2.1% is a small fraction of the total sporadic population; the contribution of somatic variants to the majority of sALS remains unproven.
- Detecting low-heteroplasmy variants is technically challenging and prone to sequencing artifacts (Codron et al., 2026, *Neurobiol Dis* demonstrated that some reported mtDNA variants in ALS are artifactual).

**Trial tractability:** **LOW–MODERATE.** Identification of somatic mutation carriers could open gene-targeted therapies (antisense, gene editing) to a small additional fraction of sALS, but current sequencing pipelines are insufficient for clinical screening.

---

## 6. Environmental Exposure–Linked Subgroups

### 6.1 Oxidative Stress / Environmental Toxin Hypothesis

**Hypothesis:** Subsets of sporadic ALS reflect chronic environmental risk (heavy metals, pesticides, organic solvents) that converge on oxidative-stress or mitochondrial pathways.

**Evidence supporting:**
- Epidemiological meta-analyses (not re-summarized here, but widely accepted in ALS consortia) associate military service, agricultural work, and cigarette smoking with elevated ALS risk.
- Biochemically, GSH depletion and 4-HNE elevation in motor cortex are consistent with environmental toxin–induced oxidative damage.

**Evidence contradicting /Limitations:**
- Individual environmental associations are weak and inconsistent across studies; no single exposure explains more than a small fraction of sALS population attributable risk.
- Causal inference is limited by retrospective exposure assessment and confounding.

**Trial tractability:** **LOW.** Exposure-based stratification is currently too imprecise for trial enrichment, but integrating exposure history with biomarker panels (e.g., 4-HNE) could be explored in future prospective cohorts.

---

## 7. Comorbidity- and Systems-Based Subgroups

### 7.1 Metabolic / Diabetic Comorbidity Cluster

**Hypothesis:** ALS co-occurs with metabolic dysfunction, and diabetes/hyperlipidaemia status may define a biologically distinct subgroup or even modify disease risk.

**Evidence supporting:**
- In a Swedish spinobulbar muscular atrophy (SBMA) cohort, 39% had diabetes mellitus; two patients with rapid progression were ultimately found to have concomitant ALS, raising questions about shared metabolic vulnerability (Roos et al., 2026, *J Neurol*).
- Metabolic dysfunction markers have emerged as presymptomatic features in deep-phenotyped genetic carriers, suggesting that systemic metabolic changes may be an endophenotype of motor neuron vulnerability (Benatar et al., 2023, *Curr Opin Neurol*).

**Evidence contradicting /Limitations:**
- Evidence for metabolic comorbidity stratification is limited and largely associative.

**Trial tractability:** **LOW–MODERATE.** Metabolic interventions (e.g., high-fat diets, mitochondrial boosters) could be tested in metabolically defined subgroups but require better prospective characterization.

### 7.2 Sleep / EEG Endophenotypes

**Hypothesis:** Thalamocortical dysfunction in ALS produces sleep spindle abnormalities that define phenotypic subgroups.

**Evidence supporting:**
- In 97 sporadic ALS patients, unsupervised clustering of sleep spindle parameters revealed two distinct spindle phenotypes. The “spindle-deficient” phenotype was independently associated with lower ALSFRS-R scores, lower FVC, and absence of drinking history (Li et al., 2026, *Sleep Med*).
- A diagnostic model incorporating spindle density showed AUC 0.931 (optimism-corrected 0.923), though this lacks longitudinal validation.

**Evidence contradicting /Limitations:**
- Exploratory, cross-sectional; validation in independent cohorts and understanding of mechanism are needed.

**Trial tractability:** **LOW (currently).** May become useful for non-invasive stratification once validated.

---

## 8. The ALS-FTD Spectrum as an Etiological Continuum

**Hypothesis:** A substantial subgroup of sALS represents the motor end of an ALS-frontotemporal dementia (FTD) disease continuum driven by TDP-43 proteinopathy.

**Evidence supporting:**
- Amygdala TDP-43 pathology predicts behavioural symptoms in sALS (Rifai et al., 2026).
- TDP-43-mediated microglial dysfunction via triglyceride metabolism is observed in both sporadic and TARDBP-mutant ALS, linking glial lipid metabolism to the broader ALS-FTD spectrum (Kabra et al., 2025).
- C9orf72 repeat expansion carriers (often classified familial) can appear sporadic and are linked to FTD and ALS, suggesting overlap even within apparently sporadic cohorts.
- A dysregulated SIRT1-p53 DNA-damage feedback axis is shared between sporadic ALS and CHMP2B-FTD, pinpointing a common pathogenic mechanism (Jun et al., 2025, *Nat Commun*).

**Evidence contradicting /Limitations:**
- FTD comorbidity is present in only a minority (~15%) of sALS cases, but TDP-43 pathology is near-universal—meaning the hypothesis may better describe shared molecular pathology than clinical subgrouping.

**Trial tractability:** **MODERATE.** Cognitive/behavioural stratification (using ECAS or similar) can enrich trials for patients with ALS-FTD overlap, but most anti-TDP-43 trials would benefit from pathology confirmation currently only available postmortem.

---

## 9. Summary of Subgroup Tractability for Focused Trials

| Proposed Subgroup | Defining Feature | Support Level | Tractability for Trial | Rationale |
|-------------------|------------------|---------------|----------------------|-----------|
| **Lower-limb–onset ALS** | Site of onset + slower ALSFRS-R slope | Strong | **HIGH** | Homogeneous, slower progression; reduces confounding by rapid decline |
| **NfL-high / rapid-progressors** | Elevated plasma/CSF NfL | Strong | **HIGH** | Already used for enrichment; reflects active axonal degeneration |
| **Biomarker clusters** (LPA-derived) | NfL / pTau181 / GFAP profiles | Moderate–Strong | **HIGH** | Captures distinct UMN/LMN/glial biology; ready for prospective validation |
| **TDP-43 proteinopathy–positive sALS** | pTDP-43 in CNS (postmortem or PET) | Strong pathology; weak premortem detection | **MODERATE** | Near-universal in sALS; anti-aggregation therapies need pathology anchors |
| **Oxidative-stress / low GSH endotype** | GSH-depleted motor cortex (MRI) or high 4-HNE | Moderate | **MODERATE** | Enrichment for antioxidant trials |
| **Glial-activation cluster** | GFAP elevation, sTREM2 elevation | Moderate | **MODERATE** | Immunomodulatory trials may benefit |
| **Somatic mutation carriers** | Low-fraction somatic ALS-gene variants | Emerging (2.1% of sALS) | **LOW–MODERATE** | Could open precision gene-targeted therapy to a minority |
| **Sleep-spindle deficient** | EEG spindle density < threshold | Exploratory | **LOW** | Needs longitudinal and independent validation |
| **Exposure-linked subgroups** | Occupational toxin / smoking history | Weak–Moderate | **LOW** | Exposure recall bias; confounded |

---

## 10. Conclusions and Recommendations

1. **Sporadic ALS is etiologically heterogeneous**, but converging biomarker and neuroimaging data now permit biologically informed stratification that goes beyond clinical phenotype.
2. **The most immediately actionable subgroup is lower-limb–onset ALS**, which offers a more homogeneous, slower-progressing population for neuroprotective trials.
3. **Multi-biomarker profiling (NfL + pTau181 + GFAP) defines at least three biologically distinct clusters** that predict disease aggressiveness independently and should be incorporated into adaptive trial designs.
4. **TDP-43 proteinopathy remains the central pathological feature** of most sALS, but upstream endotypes (cofilin/actin dysregulation, iron dyshomeostasis, DNA damage) are emerging as tractable therapeutic entry points.
5. **Future directions:** Prospective registries should integrate clinical phenotype, biomarker panels, genetic mosaicism screening, and environmental exposure data to build a **precision taxonomy of sporadic ALS** and power smaller, more efficient mechanism-directed trials.

---

## References (Selected)

1. Andronesi OC, et al. "Imaging Neurochemistry and Brain Structure Tracks Clinical Decline and Mechanisms of ALS in Patients." *Front Neurol.* 2020;11:590573.
2. Beers DR, et al. "Longitudinal Assessment of Biomarkers in ALS: Discriminative Biomarkers for Disease Progression and Survival." *Ann Clin Transl Neurol.* 2026. PMID: 41943205.
3. Benatar M, Turner MR, Wuu J. "Presymptomatic amyotrophic lateral sclerosis: from characterization to prevention." *Curr Opin Neurol.* 2023. PMID: 37382103.
4. Codron P, et al. "Analysis of mitochondrial DNA heteroplasmy in sporadic ALS suggests technical limitations rather than disease association." *Neurobiol Dis.* 2026. PMID: 42035928.
5. Jagaraj CJ, et al. "Cofilin hyperphosphorylation triggers TDP-43 pathology in sporadic amyotrophic lateral sclerosis." *Brain.* 2026. PMID: 41804798.
6. Jun YW, et al. "The Ku80-p53-SIRT1 axis in DNA damage response contributes to sporadic and familial ALS and FTD." *Nat Commun.* 2025. PMID: 41422089.
7. Kabra K, et al. "Loss of Nuclear TDP-43 Impairs Lipid Metabolism in Microglia-Like Cells." *Research Square.* 2025. PMID: 41333389.
8. Li M, et al. "Sleep spindle alterations as a novel biomarker for phenotypic stratification in sporadic amyotrophic lateral sclerosis." *Sleep Med.* 2026. PMID: 41996956.
9. Li R, et al. "Autologous SVF therapy modulates neuroinflammation in ALS: phase I trial demonstrating safety and CSF biomarker dynamics." *Front Aging Neurosci.* 2026. PMID: 41987881.
10. Rifai OM, et al. "Amygdala TDP-43 pathology is associated with behavioural dysfunction and ferritin accumulation in amyotrophic lateral sclerosis." *Brain Commun.* 2026. PMID: 41958917.
11. Senerchia G, et al. "Plasma NfL, GFAP and pTau181 define distinct biological axes in amyotrophic lateral sclerosis." *Neurobiol Dis.* 2026. PMID: 42049146.
12. Shovman Y, et al. "Slower Progression Rates in Lower Limb-Onset ALS." *J Clin Med.* 2026. PMID: 42074898.
13. Yusuf S, et al. "Opportunities and challenges related to participant stratification and cohort enrichment in ALS clinical trials." *Amyotroph Lateral Scler Frontotemporal Degener.* 2026. PMID: 42033202.
14. Zhou Z, et al. "Somatic mosaicism in ALS and FTD identifies focal mutations associated with widespread degeneration." *Nat Genet.* 2026. PMID: 41986690.