Ten Studies, Three Brain Networks, and the Slow Case That Spinal Manipulation Is Doing Something Central
Cao and colleagues pulled together every neuroimaging study that has scanned the brains of patients with lumbar disc herniation and chronic low back pain before and after spinal manipulation therapy. Ten studies cleared the screen. The brain functional changes spinal manipulation produced clustered in the prefrontal cortex, the visual network, and the default mode network. The signal is consistent. The mechanism is not yet settled.
Key Points
- Cao and colleagues conducted a scoping review of brain functional changes spinal manipulation produces in adults with lumbar disc herniation and chronic low back pain. Eight databases searched through August 2025. Of 169 articles screened, 10 met inclusion.
- The included neuroimaging studies converged on three patterns: modulation of the prefrontal cortex, the visual network, and the default mode network. These changes correlated with reductions in pain intensity (VAS), functional disability (ODI, JOA), and emotional symptom load (SAS, SDS).
- Acupoint-targeted manipulation elicited distinct brain activation patterns compared with non-acupoint stimulation. The review treats this as hypothesis-generating, not as confirmation of an acupoint-specific mechanism.
- This is a scoping review, not a systematic review with meta-analysis. The job is to map the field and identify gaps, not to estimate effect size. The authors are explicit that standardized clinical and imaging protocols are still missing.
- The clinical signal supports a central-nervous-system component to how spinal manipulation reduces chronic low back pain. It does not establish causation, and the included sample sizes are small by neuroimaging standards.
- This commentary connects the Cao 2026 finding to the broader manual-therapy evidence base, including the Cochrane de Zoete 2026 spinal manipulation review and the JAMA Bronfort 2026 PACBACK trial.
People with chronic low back pain from a lumbar disc herniation often feel pain that does not match what shows up on imaging. The pain is real, but the explanation is not only mechanical. The brain is part of the picture.
Researchers in China pulled together every published study that put patients with lumbar disc herniation into an MRI scanner before and after spinal manipulation therapy. Ten studies fit. Across those studies, the same parts of the brain kept showing up as different after treatment: regions tied to attention and decision-making (the prefrontal cortex), regions tied to sensory processing (the visual network), and regions tied to mind-wandering and self-referential thought (the default mode network). When pain dropped, those brain regions changed. When mood improved, the same regions changed.
That does not prove spinal manipulation is rewiring the brain. The studies are small, the methods vary, and the brain changes that show up on imaging may follow pain relief rather than cause it. What it does suggest is that the pain-relief story is not only a story about joints and muscles. The brain is participating, and it can be measured.
Why This Paper Matters
Chronic low back pain related to lumbar disc herniation is one of the most common pain presentations in adult musculoskeletal care. The classical explanation, mechanical compression of a nerve root by displaced disc material, accounts for some of the symptom picture but not all of it. Imaging-positive disc herniations are common in asymptomatic adults. Pain-positive patients often have imaging that looks no worse than the asymptomatic group. The mismatch between tissue findings and the lived pain experience has driven a generation of research into central pain processing.
Spinal manipulation has been studied for decades as a mechanical intervention. The Cochrane de Zoete 2026 review pooled 76 randomized trials and concluded that spinal manipulation produces small-to-medium pain reductions in chronic low back pain, at low certainty, with a benign safety profile. That body of work tells us the intervention works at the population level. It does not tell us why.
The Cao review steps into the why question. It asks what the brain looks like, before and after, in patients who receive spinal manipulation for chronic low back pain related to lumbar disc herniation. That is a different layer of evidence from a clinical trial. The point is not to estimate how much pain comes down. The point is to map which neural systems track with that pain reduction.
What the Study Did
The review was led by Cao and colleagues at the Hubei University of Chinese Medicine and the Hubei Provincial Hospital of Traditional Chinese Medicine in Wuhan, China. The review was registered with PROSPERO (CRD420251109911) before data extraction. The methodological framework followed established scoping review guidance.
The team searched eight databases from inception to August 1, 2025. Inclusion required original human studies of patients with lumbar disc herniation and chronic low back pain, an intervention involving spinal manipulation or closely related manual therapy, and pre-and-post neuroimaging using functional MRI or a comparable technique. Of 169 records screened, 10 studies met all inclusion criteria.
Outcomes captured across the included studies fell into three categories. First, brain activity and connectivity measures, including regional activation, functional connectivity strength, and network-level changes. Second, clinical pain and disability outcomes including the Visual Analogue Scale (VAS), the Oswestry Disability Index (ODI), and the Japanese Orthopaedic Association (JOA) score. Third, emotional symptom measures including the Self-Rating Anxiety Scale (SAS) and the Self-Rating Depression Scale (SDS).
Because this is a scoping review, no pooled effect size was calculated. The synthesis is descriptive and pattern-based. The job is to map what has been studied, what the convergent findings are, and where the gaps sit.
What the Study Found
Three brain regions or networks recurred across the ten included studies as showing reliable change after spinal manipulation therapy.
The prefrontal cortex showed altered activation and connectivity in the majority of studies that imaged it. The prefrontal cortex carries weight in pain processing as the region that integrates sensory pain signals with attention, expectation, and emotional context. Changes here are consistent with a top-down modulation story: treatment reduces the centrally-amplified component of chronic pain, not only the peripheral nociceptive input.
The visual network showed connectivity changes that, on first read, are surprising. Chronic pain has been shown in other neuroimaging work to recruit visual processing regions in ways that healthy controls do not. The visual network functions as part of the salience and attention system in chronic pain states. A normalization of visual-network connectivity is consistent with reduced pain-related attentional capture.
The default mode network, the set of regions active during self-referential and mind-wandering states, showed altered connectivity across multiple studies. Chronic pain is associated with default mode network disruption in a body of independent imaging work. Treatment-associated normalization of default mode network connectivity in the Cao pool is consistent with that broader literature.
Clinical correlations followed the imaging findings. Reductions in VAS pain scores, ODI disability, and JOA functional limitation tracked with the brain changes. SAS anxiety and SDS depression scores improved alongside the imaging signal. The consistency between brain and clinical measures across the included studies is the strongest part of the synthesis.
One subset of studies compared acupoint-targeted manipulation with non-acupoint stimulation. The acupoint-specific protocols produced distinct brain activation patterns relative to non-acupoint controls. The review treats this as a hypothesis worth testing in larger controlled designs, not as confirmation of an acupoint-specific mechanism.
Mechanism and Clinical Context
The mechanistic story this review supports is not a replacement for the mechanical story. It is an addition to it.
For a patient with lumbar disc herniation and chronic low back pain, the peripheral nociceptive picture is real. Inflamed nerve roots, restricted spinal segments, altered loading patterns, and protective muscle guarding are all part of why the patient hurts. Spinal manipulation acts on those mechanical contributors directly. That is the layer most chiropractic education emphasizes, and it remains correct.
What the Cao synthesis adds is the central layer. Chronic pain is not only a peripheral signal arriving at a passive brain. The brain actively shapes how that signal is experienced. Prefrontal cortex involvement reflects the attention and expectation channel. Visual network involvement reflects sensory salience. Default mode network involvement reflects the way chronic pain hijacks self-referential processing. When those regions normalize during treatment, the patient's experience of pain shifts in a way that is not fully accounted for by what is happening at the disc.
This is not new in pain neuroscience as a general principle. The contribution of this review is to apply it specifically to spinal manipulation for lumbar disc herniation, and to show that the same three networks recur across an independent set of imaging studies. That convergence across small studies, with different teams, different MRI scanners, and different analysis pipelines, is more interesting than any one study alone.
None of this proves that spinal manipulation causes the brain changes in the way the lay reading might suggest. The brain changes could follow pain relief rather than cause it. They could be partially driven by the therapeutic encounter itself, including expectation effects and clinician contact, which are known to modulate pain-related brain activity. The review does not separate intervention-specific effects from context effects. That is a design limit of the included studies, not a reading error.
Limitations and Caveats
The scoping-review format and the small underlying evidence base set a clear ceiling on what can be claimed.
Sample sizes in the included neuroimaging studies are small. Most fMRI investigations of spinal manipulation enroll fewer than 50 patients. That is enough to generate a signal but not enough to settle effect-size questions or to produce stable connectivity estimates. Replication across larger imaging cohorts is the next required step.
Methodological heterogeneity is meaningful. The included studies used different scanner field strengths, different task and resting-state paradigms, different preprocessing pipelines, and different statistical thresholds. Pattern-level convergence across that heterogeneity is encouraging. It is not the same as a pooled estimate from harmonized methods.
Most of the included studies were single-arm pre-post designs without an active comparator. A pre-post imaging change after any therapeutic encounter cannot be attributed cleanly to the specific intervention. Sham-controlled or active-comparator imaging designs are still rare in this literature.
The geographical distribution of the included studies is narrow. The work clusters heavily in the Chinese imaging literature on tuina and chiropractic-adjacent manual therapies. Generalization to spinal manipulation as practiced in North American and European chiropractic settings should be made with caution. The mechanical interventions are similar but not identical, and the patient selection criteria differ across systems.
Causal direction is unresolved. The included studies report correlation between brain changes and clinical improvement, not causation in either direction. Brain changes could drive pain relief, follow pain relief, or both could share an unmeasured upstream cause. The data does not adjudicate.
Publication bias is plausible. Imaging studies with null findings on functional connectivity are systematically harder to publish than studies with significant findings. The 10 included studies all show some imaging change. That is partially the literature speaking and partially the publication filter speaking.
None of these limits invalidate the convergent finding. They define the size of the claim that can honestly be made: brain functional changes spinal manipulation produces are consistent across small imaging studies, cluster in three plausible networks, and track with clinical and emotional improvement. That is a real signal at hypothesis-generating strength, not at confirmatory strength.
What I Take From This for Practice
I see a steady stream of patients in my Tulsa walk-in clinic with chronic low back pain related to lumbar disc herniation. The typical case is an adult who has had imaging, has been told the disc is the problem, and has cycled through medications, physical therapy, or both with partial relief. The clinical question I am answering is rarely whether spinal manipulation will fix the disc. It is whether spinal manipulation will move the pain enough to let the patient function and progress through care.
The Cao review does not change my hands. It changes my language.
For clinical practice, I take three concrete things from this paper.
One: The pain experience in chronic low back pain has a central component that is measurable.
When a patient with a confirmed lumbar disc herniation reports pain that does not seem to match the imaging, the central-processing layer is part of why. That is not a weakness of the patient or a sign of malingering. It is how chronic pain works at the level of the prefrontal cortex, the visual network, and the default mode network. Naming this for patients shifts the conversation.
Two: Conservative care can act on the central layer without requiring central-acting medications.
The brain changes in the Cao pool happened during spinal manipulation therapy, an intervention that is mechanical at the point of contact. The brain does not require a pharmacological key to shift. Reducing nociceptive input, restoring movement, and changing how the patient experiences their own spine in space all feed back into the central system. That is consistent with how I see most chronic low back pain patients respond clinically.
Three: The honest framing for patients is mechanism-plausible, not mechanism-proven.
I am not telling patients that spinal manipulation rewires their brain. I am telling them that the pain they are experiencing has a central component, that conservative care can act on that component, and that imaging research is starting to show what that looks like in the brain. The strongest version of the claim that survives the GRADE-style scrutiny is the cautious version. That is the one that holds up over years of patient relationships.
What I tell patients in my Tulsa walk-in clinic when they ask why spinal manipulation seems to help has not fundamentally changed. The intervention reduces local mechanical contributors to pain. It restores movement. It changes the input the brain is receiving. The Cao review supports the layer of that explanation that involves the brain and gives me a defensible citation for it.
What the Review Does Not Resolve
The review does not establish that spinal manipulation causes the observed brain changes. The included designs are not built to support a causal claim. A sham-controlled imaging trial with adequate power would be required for that, and the field does not yet have one.
The review does not tell us which patients are most likely to show brain changes with spinal manipulation. Subgrouping by pain duration, herniation severity, baseline pain catastrophizing, or prior treatment exposure would clarify which patient profiles drive the effect. The included studies are too small individually to support that subgrouping.
The review does not adjudicate between spinal manipulation as the active ingredient and the broader therapeutic encounter as the active ingredient. Clinician contact, expectation, and the manual therapy ritual itself all modulate central pain processing. Whether the brain changes are specific to the manipulation or driven partly by the surrounding context is not separable in this dataset.
The review does not bridge to long-term outcomes. The included imaging studies measure brain change over short windows, typically a single course of care. Whether the brain changes persist, fade, or deepen over months and years is unstudied at scale.
None of these unresolved questions undermine the headline. They define the next research agenda for pain neuroimaging in this population.
Ten neuroimaging studies, three brain networks, and a consistent signal that spinal manipulation therapy is doing something measurable at the central level in patients with lumbar disc herniation and chronic low back pain. The brain functional changes spinal manipulation produces cluster in the prefrontal cortex, the visual network, and the default mode network, and track with reductions in pain, disability, and mood symptoms. The certainty of evidence is hypothesis-generating, not confirmatory. The direction of the field is consistent with what a generation of pain neuroscience research has predicted. The size of the causal effect is the next question the field has to earn the right to answer.
Editorial Disclosure
The Forward Review is editorial commentary, not peer reviewed. Written by Jason Forward, DC, a licensed chiropractor at Vanguard Chiropractic in Tulsa, Oklahoma. Commentary reflects clinical interpretation and does not represent the views of the original study authors, the journal Frontiers in Neurology, or any institution.
Conflict of interest: Dr. Forward is a practicing chiropractor whose clinical work includes care for adults with lumbar disc herniation and chronic low back pain. No financial relationships with the cited authors, the publisher, or any party to this work.
Scope of practice note: Imaging interpretation and neurological diagnosis are outside chiropractic scope of practice. Discussion of brain imaging findings in this commentary refers to published research and is not a representation that chiropractic care includes neuroimaging assessment.
Corrections policy: Errors are corrected with a dated correction notice appended to the article. Original text remains accessible via revision history.
Level of evidence: Scoping review (10 studies). No pooled effect size estimate. Hypothesis-generating evidence on brain functional changes spinal manipulation produces in lumbar disc herniation with chronic low back pain.
- What Actually Helps a Child With ADHD, Ranked by Evidence Instead of PopularityForward Rev. 2026;1(2)
- Hands-On Care for Muscle and Joint Pain: A 2026 Review of ReviewsForward Rev. 2026;1(2)
- Which Neck Exercise Actually Works? A 2026 Network Meta-AnalysisForward Rev. 2026;1(2)
- Exercise or Hands-On Therapy for Low Back Pain? A 2026 Network Meta-AnalysisForward Rev. 2026;1(2)
View Vol. I, No. i and the full archive