Can Aerobic Exercise Change Blood-Based Neural Injury Markers During Methamphetamine Recovery? A 56-Person Trial

Short answer: In a randomized rehabilitation-center trial, men assigned to twelve weeks of additional moderate-intensity aerobic exercise had a larger reported decrease in plasma neurofilament light chain (NfL) than men receiving conventional treatment alone. The between-group comparison was statistically significant. A change in serum neuron-specific enolase (NSE) was also statistically different, while serum S100B did not show a significant between-group difference. These findings concern laboratory biomarkers of possible neural or barrier-related processes—not measured methamphetamine concentrations, proof of a repaired blood–brain barrier, or a test of any person’s intoxication.

The underlying primary study by Zhicheng Zhu, Jisheng Xu, Yu Jin, Lu Wang, and Xue Li was published in Brain Sciences in 2022. The original randomized-trial report (DOI 10.3390/brainsci12111521) is the evidentiary basis for the numerical comparisons discussed here. Its strongest forensic lesson is that even when a study collects venous blood and produces quantitative laboratory results, one must still identify what chemical or protein was assayed, which fraction of blood was tested, and which clinical or legal question the assay could actually address.

What did the experiment establish?

The investigators randomized sixty men with a history of methamphetamine dependence into two groups of thirty. Both groups lived in a compulsory isolation drug-treatment setting and received conventional rehabilitation services. One group additionally undertook a structured aerobic-exercise program for twelve weeks. Fifty-six people completed the study and contributed the principal post-intervention comparisons: twenty-eight in the exercise arm and twenty-eight in the conventional-treatment arm. The main published laboratory differences therefore concern these men, not sixty complete pairs of measurements.

The exercise group exhibited a greater mean reduction in measured plasma NfL, a protein related to neuronal axons, than the comparison group. The original analysis reported a between-group p value of 0.001 for the change-score comparison. For NSE, an enzyme found in neurons and other tissues, the investigators reported a statistically significant difference between the distributions of signed serum change scores, with p = 0.010. For the third analyte, S100B, a protein often discussed in relation to glial processes, the between-group test was not statistically significant: p = 0.615.

It is tempting to condense the result into “exercise repairs methamphetamine-damaged brains.” That would exceed the measurements. There was no control group of never-exposed men, no directly imaged recovery of a blood–brain barrier, and no demonstrated before-and-after improvement on a neuropsychological or driving test. The defensible conclusion is narrower: in this specific residential population, assignment to extra exercise was associated with different changes in selected circulating protein biomarkers.

How many people were studied, and why do the denominators matter?

The trial flow begins with sixty-five people screened. Five were excluded before randomization, leaving sixty randomized in equal proportions. During follow-up, two exercise-group participants discontinued because of medical conditions; two conventional-treatment participants were transferred to prison. The analyzed post-treatment counts were therefore twenty-eight per group. A person who supplied an initial specimen but lacked a final specimen is not a new participant, and a pair of specimens from the same person does not constitute two independent volunteers.

The paper presents baseline characteristics for thirty participants per group, while its change-score and post-exercise tables report twenty-eight per group. That distinction affects denominators, confidence in precision, and treatment of missing data. The authors also supplied a secondary analysis in which missing measurements were statistically filled in. Such imputed values are modeled replacements, not new tubes of measured plasma or serum and not additional biological individuals.

All participants were men, aged within a specified adult enrollment range, recruited from one Chinese rehabilitation center. Their methamphetamine histories came from screening information and reported prior use. The investigation was not a controlled laboratory methamphetamine challenge: researchers did not give volunteers a precisely timed dose of methamphetamine before drawing a toxicology specimen. Generalization to women, people acutely intoxicated, persons outside residential treatment, or individual defendants requires evidence this experiment did not collect.

What exactly was the twelve-week intervention?

On top of conventional education and counseling, the exercise group had planned sessions Monday through Friday, approximately one hour each morning. The session design included ten minutes of warming up, thirty minutes of jogging or stationary cycling, and twenty minutes of stretching. Investigators targeted a monitored heart rate equal to 65–75% of an age-derived maximum. The planned period was twelve weeks, so a scheduled exercise session is an intervention exposure, not an independently randomized patient.

The control group continued conventional educational detoxification and psychological counseling without this added aerobic training. The comparison is therefore exercise added to usual residential treatment versus usual residential treatment. It does not independently isolate jogging from cycling, distinguish exercise from staff attention, or establish that a particular number of minutes is an optimum neuroprotective dose. Heart-rate targets describe cardiovascular exercise intensity; they are not a therapeutic methamphetamine dose or a forensic concentration threshold.

Randomization improves the case for a treatment-group comparison, but the participants could know whether they were exercising. The researchers described blinding investigators and outcome assessors, not concealing a daily hour of exercise from the people performing it. The setting, adherence, concurrent rehabilitation practices, and four losses to follow-up matter when assessing how confidently the observed laboratory separation can be attributed to the added program.

Which physical specimens and laboratory assays were used?

The authors collected fasting venous blood in the morning before the program and after its completion. However, the assays were not performed on intact forensic venous whole blood. One collection was allowed to clot and was centrifuged to obtain a separated fraction for the serum tests. A second collection used an EDTA anticoagulant and was centrifuged to prepare plasma. These supernatants were frozen before analysis. Both fractions arise from venous sampling, but neither is interchangeable with unseparated whole blood.

Plasma NfL was measured using the Quanterix Simoa NF-Light assay on an HD-1 analyzer, with results reported in picograms per milliliter (pg/mL). The investigators measured serum NSE and S100B with immunoassay kits, expressing those results in nanograms per milliliter (ng/mL). One nanogram is one thousand picograms, but converting the prefixes does not make the underlying proteins equivalent or establish that their concentrations should change together.

None of the three laboratory targets was methamphetamine, amphetamine, or a chiral stimulant metabolite. The paper did not report a quantitative chromatography or mass-spectrometry measurement of methamphetamine in the human legal venous whole-blood fraction, separated serum or plasma, urine, oral fluid, hair, sweat, vitreous, or exhaled breath. A positive urine screen within the preceding year was a recruitment criterion, not a documented contemporaneous stimulant concentration-time series.

Numerical evidence: the original NfL result

Before the intervention, the NfL mean was numerically higher in the future exercise group than in the comparison group, but the baseline contrast was not statistically significant (p = 0.109). After twelve weeks, the means were closer. The original paper then compared each person’s baseline-minus-follow-up value between the groups. A positive change score under this definition denotes a decline in circulating NfL over time, not direct visualization of repaired neural tissue.

The two change-score averages differ by 1.15 pg/mL (1.75 minus 0.60) as an arithmetic description of the published means. That is not a validated clinical minimum important difference, a calculated change in blood–brain barrier permeability, or a measure of how much damage any person suffered. The study does not supply a reference group of never-exposed healthy adults from which to derive an individualized definition of normality. Nor does it show that every exercise participant’s NfL fell, as opposed to a group mean moving downward despite variation across individuals.

What happened to serum NSE and S100B?

The other two proteins behaved differently. For NSE, the original nonparametric between-group comparison of signed change values was significant. The paper reports a more negative signed change summary for the exercise group than for the comparison group; the chosen change convention was described as baseline-minus-follow-up. However, the published NSE table combines arithmetic post-treatment means with median-like change summaries and contains tensions between the surrounding prose, plotted distributions, and change-sign description. The result should therefore be reproduced as the reported group comparison without silently recalculating a physiologic direction from mismatched summary statistics.

For S100B, the between-group difference in the reported signed change distributions was not statistically significant. The conventional group’s value and the exercise group’s value were numerically close; a chart’s individual dots and a modest numerical difference should not be substituted for the actual inferential result. In the authors’ missing-data sensitivity procedure, the S100B result approached but did not cross a conventional 0.05 significance threshold (p = 0.051). Treating 0.051 as proof of a demonstrated effect would reverse the study’s own classification.

These data support a three-part outcome statement rather than a generalized claim that every neurological marker improved: NfL change was significant; NSE’s reported signed distribution differed; S100B did not reach the study’s stated criterion. The investigators did not directly validate whether NSE changes arose from central nervous system damage rather than other potential sources of biomarker variation. A specific marker’s statistical movement must be interpreted alongside assay validation, biological specificity, sample handling, and the clinical question—not by relying solely on a suggestive biomarker name.

Why is a protein biomarker not a methamphetamine blood level?

A quantitative laboratory result is not automatically a forensic drug test. A drug test requires a method directed at the relevant chemical entity: methamphetamine, amphetamine, potentially the separate stereoisomers when that distinction matters, or a specified metabolite. Laboratory documentation ordinarily needs the specimen matrix, analytical technique, calibration, identification criteria, validated reporting range, collection timing, quality controls, and uncertainty before the number is used to address a forensic question.

NfL, NSE, and S100B are proteins measured for possible pathophysiologic meaning. Their concentrations do not quantify stimulant mass, reveal whether the detected stimulant would be the d- or l-isomer, or identify a route of intake. Even a clinically plausible biomarker difference cannot be converted into milligrams of methamphetamine previously used, the clock time of last ingestion, an individual’s current impairment, or a certainty about an unrelated incident.

The useful analogy is a distinction between a response marker and an exposure measurement. A hormone, cytokine, enzyme, or axonal protein can respond to disease, exercise, medication, infection, trauma, or stress; a confirmed chemical drug concentration directly addresses a different analytical target. To explore the matrix distinction further, see the site’s separately published guide to biological matrices and its discussion of what does and does not follow from a reported concentration.

Can the findings prove that the blood–brain barrier healed?

No. Circulating NSE and S100B are sometimes investigated as indirect markers relating to nervous tissue and barrier injury. Plasma NfL is widely discussed as an axonal-associated biomarker. But this trial did not independently visualize endothelial tight junctions, measure permeability across a subject’s blood–brain barrier, image repair of a specific cerebral lesion, or correlate each participant’s marker change with a validated direct barrier measure. The conclusion that the barrier itself was restored is therefore a hypothesis suggested by surrogate measurements, not a directly demonstrated anatomical finding.

Substantial biological variation, different tissue origins, physiological response to exercise, assay characteristics, and missing reference ranges can complicate interpretation. The paper did not construct a validated diagnostic classifier distinguishing methamphetamine-related neural injury from its alternatives. It did not show that reducing a protein concentration always means improving cognition. The intervention may be a promising subject for further controlled study without creating a laboratory shortcut to individual neurologic causation.

Why do significant and nonsignificant endpoints belong in the same explanation?

Selective reporting can make a small trial appear more consistent than it was. In this study, NfL and NSE passed the investigators’ threshold on their principal between-group change comparisons, but S100B did not. The baseline NfL comparison was also nonsignificant, as were baseline NSE and S100B. These null findings are positive information about the pattern of evidence: not every source analyte separated the treatment groups, and initial randomization did not produce statistically detectable baseline differences on the reported markers.

A p value above 0.05 does not prove exact equivalence, especially in a modest trial. Conversely, a small p value for a biomarker does not establish that the result is sufficiently specific for a personal diagnosis, let alone for the much more demanding interpretation of a defendant’s condition at an event. The comparisons were performed for three related biochemical endpoints in a relatively small trial; a careful reader should examine the family of questions and the methods used rather than pretending one p value summarizes the entire experiment.

The laboratory evidence is particularly vulnerable to being overstated if post-treatment group means are mixed with signed within-person changes or if an imputed-data analysis is treated as a second independent cohort. The authors used a missing-data procedure as a robustness exercise, not as a new randomized experiment. The apparent precision of an additional decimal place cannot compensate for source inconsistencies or replace direct measurements.

What other outcomes were not tested at all?

The investigators discussed inhibitory control, memory, and blood–brain barrier biology as possible mechanisms and future questions. They did not administer a validated battery demonstrating treatment-induced recovery of attention, executive function, neuropsychological impairment, or simulated or real driving. Nor did they assay methamphetamine concentrations after a known controlled oral, smoked, injected, or intranasal dose. There is no contemporaneous legal venous whole-blood drug concentration paired with an impairment rating to build an individual probability model.

Likewise, the trial did not ascertain whether any participant committed a particular act, caused a traffic collision, reported an accurate memory, or was impaired during a police encounter. It did not measure on-road near misses, validated crash risk, or differential legal responsibility. Statements about those outcomes would be inventions, not reasonable extrapolations from the blood protein measurements. Absence of a measurement must be described as not assessed, rather than as a demonstrated significant result or a demonstrated nonsignificant result.

How should an expert approach a specimen-based claim?

Start by separating the question into three levels: exposure, biological response, and functional consequence. This trial addresses selected biochemical responses to added exercise in previously methamphetamine-dependent men. It does not directly quantify an individual’s drug exposure at a known time or establish a contemporaneous functional deficit. Even an excellent treatment-effect study in its own domain might be irrelevant to whether a specific specimen establishes intoxication under a particular statute.

If an actual toxicology report is involved, request the original specimen documentation, which blood component was collected and processed, the targeted chemical analytes, calibration and validation, quality controls, chain of custody, and relevant collection or incident time. Confirmatory methods answer a different question from enzyme immunoassays used for these proteins. The site’s published overview of confirmatory drug testing illustrates why identity and concentration require method-specific evidence.

If the dispute instead concerns the medical significance of NSE, S100B, or NfL, identify the specimen handling, preanalytical conditions, assay platform, clinical history, competing causes, and established interpretation framework for that exact protein. An expert should not automatically translate a plasma NfL value into a serum value or a whole-blood stimulant concentration. Differences in the biological fraction and in what the instrument detects are central to the validity of the answer.

What are the main limitations of this 56-person result?

The sample was small, male-only, and drawn from one constrained rehabilitation environment. The intervention could not be blinded to participants. Four people did not complete follow-up, and source tables switch denominators between baseline and post-treatment analyses. The researchers relied on three biomarker assays and short-term measurements before and after a twelve-week program, with no extended follow-up demonstrating durability. Without a never-exposed comparison group, the study cannot isolate all effects of historical methamphetamine use from age, health, detention setting, ordinary abstinence, or other features of rehabilitation.

The publication also has reporting limitations. One original trial-flow graphic contains an inconsistent allocation label for the control group. The NSE change-score presentation and some sensitivity-analysis statistics do not reconcile cleanly with surrounding descriptive prose. The prudent course is to retain the original reported statistics, distinguish mean from median-style summaries, and refrain from inventing corrected values. The paper’s broad statements about blood–brain barrier repair and inhibitory-control restoration outrun the set of actual tests performed.

Finally, a study of this type should not be confused with validated therapeutic drug monitoring. In drug monitoring, the target compound, dose history, sampling time, assay range, and physiological purpose govern concentration interpretation. Here the proteins were measured to explore neural or barrier-related biology. For the separate conceptual framework, the site’s already published therapeutic drug monitoring guide explains why timed concentrations and physical matrix identity matter.

What is the legally and scientifically defensible conclusion?

Among the fifty-six men who completed this randomized rehabilitation trial, twelve weeks of added aerobic training were associated with a larger decline in measured plasma NfL, a statistically different reported serum NSE change distribution, and no statistically significant between-group change in serum S100B. The result is potentially relevant to hypotheses about exercise and neurological recovery, but the actual tests were circulating protein immunoassays. They were not direct measures of brain repair and were not assays of methamphetamine or its metabolites.

For the full original Methods, numerical tables, limitations, and author interpretations, consult the primary Zhu and colleagues trial report, DOI 10.3390/brainsci12111521. Forensic interpretation remains disciplined when it preserves independent biological sample size, separates plasma and serum from whole blood, carries genuine nulls alongside favorable results, and explicitly declines to answer unmeasured person-specific questions.

Frequently asked forensic-science questions

Did the researchers find methamphetamine in the participants’ blood?

No quantified parent methamphetamine or amphetamine in whole venous blood or in separated plasma or serum was reported. The measured circulating targets were NfL, NSE, and S100B proteins. An eligibility rule involving a prior positive urine screen did not provide a timed quantitative drug result for any incident or follow-up specimen.

Were all sixty randomized men included in the published after-treatment analyses?

No. Sixty were randomized, but four discontinued or were transferred, leaving twenty-eight analyzed completers per arm. Baseline tables included thirty per arm. The paper’s missing-data procedure offered a sensitivity comparison; it did not create additional independent humans or measured laboratory specimens.

Which result had the clearest reported between-group separation?

For plasma NfL, the original baseline-minus-follow-up mean decline was 1.75 ± 1.40 pg/mL with exercise versus 0.60 ± 1.21 pg/mL with conventional treatment, with a reported between-group p = 0.001. This concerns group distributions, not a validated therapeutic or forensic cutoff for one participant.

Was there a significant improvement in all three blood proteins?

No. The reported serum S100B change comparison was nonsignificant, p = 0.615, and the missing-data sensitivity procedure reported p = 0.051, also above 0.05. Serum NSE showed a statistically significant difference but its signed-change summary requires careful source-level reading; it must not be silently rearranged into an unsupported personal recovery score.

Does a lower plasma NfL mean the blood–brain barrier has been proved intact?

No. NfL can inform hypotheses about axonal injury and disease processes, but this trial did not directly measure each person’s barrier permeability, tight-junction structure, or anatomically verified repair. A protein concentration cannot by itself establish those other facts.

Could the exercise regimen be used to calculate time since methamphetamine use?

No. Exercise was the randomized intervention, not a controlled drug dose. The original study did not measure methamphetamine pharmacokinetic elimination curves, serial urine or blood drug concentrations, or a validated last-use model. A clinical biomarker trend cannot provide an ingestion clock.

Can these group findings establish intoxication, driving impairment, or causation in a particular case?

No. The experiment did not evaluate a particular driver or event, did not quantify drugs in legally collected whole blood, and did not validate a personal impairment classifier. Any individual legal opinion would need its own facts, analytical record, functional evidence, and a method fit for the exact question.