How chronic neuropathic pain rewires the dorsal horn — and how gabapentinoids quiet it
Nociception is the nervous system's protective early-warning system: specialized peripheral sensory neurons detect actual or threatened tissue damage and relay that information to the spinal cord and brain. In its healthy, unsensitized state, this system is proportionate — pain intensity tracks stimulus intensity, and the response resolves once the stimulus ends. Understanding this baseline circuit is essential before examining how it can be pathologically amplified into chronic pain.
Two principal classes of primary afferent fiber carry nociceptive information from the periphery:
• Aδ fibers: thinly myelinated, 1–5 μm diameter, conducting at 5–30 m/s. They mediate the fast, sharp, well-localized "first pain" felt immediately after injury (e.g., a pinprick), and respond mainly to mechanical and thermal stimuli. • C fibers: unmyelinated, <1.5 μm diameter, conducting at only 0.5–2 m/s. They are polymodal — responding to mechanical, thermal, and chemical stimuli alike — and mediate the dull, diffuse, poorly localized "second pain" that follows and persists.
Transduction occurs through specialized membrane receptors: TRPV1 channels open above ~43°C (capsaicin receptor), TRPM8 responds to cold, ASIC channels detect tissue acidosis, and Piezo2 transduces mechanical force. Once threshold is reached, voltage-gated sodium channels — particularly Nav1.7, Nav1.8, and Nav1.9, expressed densely in nociceptors — generate and propagate the action potential. Rare loss-of-function mutations in Nav1.7 cause congenital insensitivity to pain, underscoring its gatekeeper role.
Primary afferent fibers enter the spinal cord and synapse in the dorsal horn, organized into ten cytoarchitectural layers first mapped by Bror Rexed. Nociceptive C fibers terminate predominantly in lamina I (the marginal zone) and lamina II (substantia gelatinosa), while Aδ fibers terminate in laminae I and V. Lamina V contains wide-dynamic-range (WDR) neurons that integrate both innocuous (Aβ, touch) and noxious (Aδ/C) input — these neurons are central players in the amplification processes covered in later stages.
Projection neurons in laminae I and V send axons across the midline and ascend via the anterolateral system, chiefly the spinothalamic tract, to the thalamus, and onward to the primary/secondary somatosensory cortex (sensory-discriminative component: location, intensity, quality) as well as the insula and anterior cingulate cortex (affective-motivational component: unpleasantness, threat).
In the healthy state, dorsal horn output is proportionate to peripheral input and is actively shaped by inhibitory circuitry. The gate control theory proposed by Melzack and Wall in 1965 established that large-diameter, low-threshold Aβ fiber input can activate inhibitory interneurons that dampen nociceptive transmission — the physiological basis for why rubbing an injured area reduces pain. Descending pathways from the periaqueductal grey (PAG) and rostral ventromedial medulla (RVM), acting through serotonergic and noradrenergic projections, provide additional top-down modulation.
Under normal conditions this proportionate, self-limiting system serves an unambiguously protective function: it produces rapid withdrawal reflexes and motivates protective behavior, then falls silent once the threat resolves.
Gate control theory reframed pain not as a fixed hard-wired alarm, but as the net output of an actively modulated spinal gate — a concept that became the conceptual foundation for understanding how that gate can later become pathologically "stuck open" in chronic pain.
When tissue is injured or inflamed, the local chemical environment around nociceptor terminals changes dramatically. A cocktail of mediators released by damaged cells, platelets, mast cells, and infiltrating immune cells — often called the "inflammatory soup" — acts directly on nociceptor membranes, lowering their activation threshold. The result is peripheral sensitization: nociceptors that once required strong stimuli now fire in response to mild ones, producing primary hyperalgesia confined to the injured tissue itself.
Tissue damage triggers release of a diverse mix of algogenic and sensitizing chemicals into the local extracellular environment:
• Bradykinin — released from plasma kininogen, acts on B2 receptors, strongly sensitizes and directly excites nociceptors • Prostaglandin E2 (PGE2) — produced by cyclooxygenase (COX) from arachidonic acid, acts on EP receptors to sensitize (the target of NSAIDs) • Nerve growth factor (NGF) — released by fibroblasts, mast cells, and keratinocytes; binds TrkA receptors on nociceptor terminals • Histamine, serotonin, ATP, and protons (local acidosis) — each acting on distinct receptors (H1, 5-HT receptors, P2X, ASIC)
These mediators do not act in isolation — they interact synergistically, and their combined effect is substantially greater than any single mediator alone, which is why multimodal inflammation produces robust sensitization.
Sensitizing mediators act through G-protein-coupled receptors that activate intracellular kinases — chiefly protein kinase A (PKA) and protein kinase C (PKC) — which phosphorylate ion channels on the nociceptor membrane within minutes:
• TRPV1 phosphorylation shifts its thermal activation threshold downward, in severe inflammation below normal body temperature (37°C), producing spontaneous firing and heat hyperalgesia • Nav1.8 phosphorylation increases current density and lowers the voltage threshold for action potential firing • Reduced outward K⁺ currents further increase membrane excitability
NGF acts on a slower but more durable timescale: acute TrkA signaling causes rapid trafficking of TRPV1 to the terminal membrane, while retrograde transport of the NGF-TrkA complex to the dorsal root ganglion cell body drives transcriptional upregulation of substance P, brain-derived neurotrophic factor (BDNF), and Nav1.8 — sensitization that can persist for days.
NGF is such a central sensitizing signal that anti-NGF monoclonal antibodies (e.g., tanezumab) have been developed as analgesics for osteoarthritis pain — directly targeting this peripheral sensitization pathway rather than downstream inflammation.
Primary hyperalgesia is the clinical correlate of peripheral sensitization: increased pain sensitivity confined to the site of injury itself — the exaggerated tenderness of sunburned skin is a familiar example. Two components are typically described: heightened sensitivity to heat (largely peripheral, TRPV1-mediated) and heightened sensitivity to mechanical stimuli (which involves both peripheral and, as later stages show, central mechanisms).
A distinct population of "silent" or mechanically-insensitive nociceptors — comprising an estimated 30% or more of cutaneous and visceral C fibers — are normally unresponsive to mechanical stimulation even at high intensity. Following sensitization, these afferents acquire mechanical sensitivity, expanding the effective nociceptor pool and receptive field driving input into the spinal cord.
Peripheral sensitization alone cannot explain the full amplification seen in chronic pain — the spinal cord itself changes with repeated input. When C fibers fire repetitively above a critical frequency, dorsal horn neurons begin firing progressively more action potentials per identical stimulus, a short-term plasticity phenomenon called wind-up. Wind-up is the mechanistic bridge between transient peripheral signaling and the lasting synaptic changes of true central sensitization.
Wind-up, first described electrophysiologically by Mendell and Wall in 1965, is a frequency-dependent increase in the number of action potentials fired by dorsal horn wide-dynamic-range (WDR) neurons in response to a train of identical C-fiber stimuli. Critically, the peripheral stimulus itself does not change in intensity — only the spinal neuron's output escalates, stimulus after stimulus, as long as the inter-stimulus interval stays below roughly 3 seconds (a repetition rate above ~0.33 Hz).
Wind-up is short-lived — it decays within seconds to a couple of minutes after the stimulus train stops — and is fully reversible. It is not, by itself, central sensitization, but it is the physiological gateway through which sustained nociceptive drive can trigger the lasting synaptic changes described in Stage 4.
Fast glutamatergic transmission via AMPA receptors produces excitatory postsynaptic potentials (EPSPs) lasting only milliseconds — far too brief, on their own, to summate meaningfully. Wind-up depends on a second, slower signal: repetitively firing C fibers co-release the neuropeptides substance P and calcitonin gene-related peptide (CGRP), which act on NK1 and CGRP receptors respectively to generate a much slower depolarizing potential lasting many seconds.
When stimuli arrive faster than this slow potential decays, successive depolarizations summate temporally, producing a rising "staircase" of baseline membrane potential in the postsynaptic dorsal horn neuron — the electrical signature of wind-up.
At the normal resting membrane potential (around −70 mV), the NMDA receptor channel pore is physically occluded by an extracellular Mg²⁺ ion in a strongly voltage-dependent manner. This makes NMDA receptors "coincidence detectors": they require both glutamate binding AND sufficient postsynaptic depolarization to conduct current — glutamate alone is not enough.
As the staircase of summating slow EPSPs from Stage 2 raises baseline depolarization toward roughly −40 mV, the electrostatic force holding Mg²⁺ in the pore weakens and the ion is progressively expelled. Each successive stimulus in the train encounters a less-blocked channel, allowing progressively more Ca²⁺ conductance — the direct mechanistic explanation for the amplifying "staircase" of wind-up, and the on-ramp to central sensitization.
Quantitative sensory testing (QST) uses repeated identical pin-prick or thermal stimuli to measure an individual's temporal summation ("wind-up ratio"). An exaggerated wind-up ratio is used clinically as a biomarker of central pain amplification risk in conditions such as fibromyalgia and chronic low back pain.
Once the Mg²⁺ block is displaced, glutamate binding to NMDA receptors triggers a large influx of Ca²⁺ into the dorsal horn neuron — a signal potent enough to activate intracellular kinase cascades that durably strengthen the synapse, much like long-term potentiation (LTP) in the hippocampus. The dorsal horn neuron becomes hyperexcitable and hyperresponsive independent of ongoing peripheral input: this is central sensitization, the defining neurophysiological event of chronic pain, and it produces allodynia and secondary hyperalgesia that spread beyond the original injury.
Sustained NMDA-receptor-mediated Ca²⁺ influx, reinforced by Ca²⁺ entering through voltage-gated Ca²⁺ channels (VGCCs), activates two key intracellular kinases: Ca²⁺/calmodulin-dependent protein kinase II (CaMKII) and protein kinase C (PKC). CaMKII undergoes autophosphorylation at threonine-286, converting it into a persistently active form that no longer requires ongoing Ca²⁺ signaling — a molecular "switch" that stays flipped on.
Active CaMKII and PKC phosphorylate existing AMPA receptors (increasing their conductance) and drive trafficking of additional GluA1-containing AMPA receptor subunits into the postsynaptic membrane over the following minutes. The synapse is now structurally and functionally stronger — an LTP-like change occurring at a spinal pain synapse rather than a hippocampal memory synapse, sometimes described as a spinal "pain memory trace."
Clifford Woolf's foundational 1983 description, refined into the current IASP definition, characterizes central sensitization as increased responsiveness of nociceptive neurons in the central nervous system to normal or subthreshold afferent input — occurring independent of continued peripheral nociceptor activity. Downstream of the initial Ca²⁺ signal, extracellular signal-regulated kinase (ERK) phosphorylation and altered gene transcription (e.g., upregulation of COX-2 and dynorphin in the dorsal horn) can prolong this hyperexcitable state for hours to days beyond the initiating stimulus.
The receptive fields of sensitized dorsal horn neurons also expand, and critically, previously non-nociceptive, low-threshold Aβ mechanoreceptor input — ordinary touch — can now activate neurons in the pain-signaling pathway.
Woolf's core insight was that pain hypersensitivity is not always a symptom of ongoing tissue damage — it can be a property of the nervous system itself, generated and maintained centrally even after peripheral injury has healed.
Central sensitization produces two hallmark clinical signs distinct from the primary hyperalgesia of Stage 2:
• Allodynia — pain evoked by stimuli that are not normally painful, such as light touch or gentle warmth — because low-threshold Aβ fibers now drive the sensitized WDR circuit • Secondary hyperalgesia — increased pain sensitivity to noxious stimuli in tissue surrounding, and sometimes far beyond, the original injury site — mediated centrally rather than by any local peripheral sensitization in that spared tissue
This state is compounded by a loss of descending inhibitory control: impaired diffuse noxious inhibitory control (DNIC), measured clinically as reduced conditioned pain modulation (CPM), further disinhibits the already-amplified dorsal horn circuit.
The IASP recognizes three mechanistic descriptors of pain, and central sensitization is the thread linking two of them to the third:
• Nociceptive pain arises from actual or threatened tissue damage with a normally functioning somatosensory nervous system (Stage 1 of this model) • Neuropathic pain arises from a lesion or disease directly affecting the somatosensory nervous system itself — diabetic peripheral neuropathy, post-herpetic neuralgia, and traumatic nerve injury are classic examples • Nociplastic pain arises from altered nociceptive processing despite no clear evidence of tissue damage or a somatosensory lesion — it is driven by central sensitization itself, and fibromyalgia is its prototypical example, alongside subsets of chronic low back pain and irritable bowel syndrome
Central sensitization is the shared final common pathway: it can develop after either a neuropathic nerve lesion or an inflammatory/nociceptive injury, and once established, it can persist and generate pain (a nociplastic state) long after the original tissue or nerve pathology has resolved.
Pregabalin and gabapentin do not block NMDA receptors, and they do not simply sedate the nervous system generically. Instead, they act at a specific structural node in the sensitized circuit: the α2δ-1 auxiliary subunit of voltage-gated Ca²⁺ channels on the presynaptic terminals of dorsal horn neurons. By reducing the trafficking of these channels to the membrane, gabapentinoids curb excitatory neurotransmitter release preferentially at the synapses that have become sensitized — quieting the amplified circuit while largely sparing normal nociceptive signaling.
Gabapentin and pregabalin bind with high affinity to α2δ-1 (and, less potently, α2δ-2), auxiliary subunits of N-type and other voltage-gated Ca²⁺ channels (VGCCs) in the presynaptic terminal membrane. Contrary to early models, gabapentinoids do not directly plug the Ca²⁺ channel pore. Instead, chronic binding disrupts forward trafficking of newly synthesized Ca²⁺ channel complexes to the presynaptic active zone — including disruption of α2δ-1's interaction with trafficking partners such as thrombospondin — reducing the density of functional Ca²⁺ channels at the terminal over a timescale of days rather than minutes.
Critically, α2δ-1 expression is markedly upregulated in dorsal horn neurons and dorsal root ganglia after nerve injury and inflammation. This upregulation confers a degree of use- and state-dependent selectivity: gabapentinoids preferentially quiet synapses that have become sensitized, with comparatively little effect on channels at normally-functioning, non-upregulated synapses.
Fewer functional presynaptic Ca²⁺ channels mean less depolarization-evoked Ca²⁺ entry at the nerve terminal, which proportionally reduces vesicular release of glutamate, substance P, and CGRP onto the postsynaptic dorsal horn neuron. This dampens both the fast AMPA-mediated EPSPs and the slow NK1-mediated depolarization that together drive wind-up (Stage 3), indirectly reducing the depolarization available to sustain NMDA receptor Mg²⁺ unblock and thereby de-amplifying — though not instantly reversing — the sensitized circuit.
Gabapentinoids do not erase the synaptic "memory trace" of central sensitization — they quiet the amplifier from the presynaptic side. Because the underlying potentiation can persist, abrupt discontinuation often allows the sensitized circuit's excitability to re-emerge, which is why gradual tapering and multimodal therapy are typically recommended.
Gabapentin is absorbed via the saturable L-amino acid transporter (LAT1) in the gut, producing nonlinear, dose-limited bioavailability — roughly 60% at low doses, falling to about 33% at high doses — which flattens the dose-response curve at higher doses and motivated development of the prodrug gabapentin enacarbil for more predictable absorption. Pregabalin, in contrast, is absorbed passively and its pharmacokinetics are linear across the clinical dose range, giving roughly 90% bioavailability and a faster, more predictable onset of action (around 1 hour, versus gabapentin's slower multi-day titration).
Gabapentinoids have established efficacy in pain conditions with a clear central-sensitization component: diabetic peripheral neuropathy, post-herpetic neuralgia, and — for pregabalin specifically — fibromyalgia. The typical number needed to treat (NNT) for ≥50% pain relief is roughly 6–8, meaning most treated patients do not achieve major relief, and effect sizes in fibromyalgia trials in particular are modest.
Common dose-limiting adverse effects include somnolence, dizziness, peripheral edema, and weight gain. Gabapentinoids are largely ineffective for acute nociceptive pain, since they act preferentially on the upregulated channels of sensitized circuits rather than on normally-functioning nociceptive synapses. They also carry recognized misuse and dependence potential, prompting reclassification as controlled substances in a number of jurisdictions since 2019.
| Product | Indication | Trial Design | Key Result |
|---|---|---|---|
| Gabapentin | α2δ-1 / α2δ-2 ligand | Nonlinear, LAT1-saturable absorption; bioavailability 33–60%; t½ 5–7 h | Requires slow up-titration, TID dosing |
| Pregabalin | α2δ-1 / α2δ-2 ligand | Linear absorption, ~90% bioavailable; t½ ≈6 h | Faster onset, BID/TID; Schedule V (US) |
| Gabapentin enacarbil | Prodrug of gabapentin | Transporter-mediated (non-saturable) absorption, sustained release | More predictable exposure; used in PHN, RLS |
| Mirogabalin | α2δ-1-selective ligand | Higher selectivity for α2δ-1 over α2δ-2 subunit | Theorized lower sedation/dizziness burden |