Does Dexamphetamine Improve Brain-Tumor Fatigue? A 41-Patient Trial Found No Benefit but More Psychiatric Events

Answer: In a randomized, double-blind, placebo-controlled trial of adults with primary brain tumors and substantial fatigue, three months of oral dexamphetamine sulfate did not produce a statistically significant improvement over placebo on the prespecified Multidimensional Fatigue Inventory outcome. Among 41 patients evaluable at three months, the median within-person fatigue-score improvement was 14 points with dexamphetamine and 10 with placebo, but the between-group test yielded p = 0.17. In the larger, 46-patient treated safety population, psychiatric adverse events were reported by 15 of 23 patients assigned dexamphetamine and 7 of 23 assigned placebo (p = 0.018). The trial supports an important distinction between a clinical treatment that can cause recognizable effects and a demonstrated benefit on its intended endpoint. It did not chemically measure amphetamine in legally collected human venous whole blood or establish individual intoxication, driving impairment, or the cause of any particular incident.

The original empirical report is Laigle-Donadey and colleagues’ phase III ANOCEF trial in Neuro-Oncology Advances (original peer-reviewed study, DOI 10.1093/noajnl/vdz043). Interpreting it requires keeping efficacy, clinical adverse effects, nominal oral dosing, and forensic drug measurements separate rather than treating one as a surrogate for another.

What specific clinical question was the trial designed to answer?

The investigators were studying a difficult supportive-care problem: persistent severe fatigue in people living with a primary brain tumor whose disease was stable or responsive to treatment. Such fatigue can involve tumor location, prior surgery, radiotherapy, chemotherapy, concomitant medication, sleep disturbance, emotional symptoms, and ongoing illness. A stimulant might reasonably be considered as a candidate intervention because wakefulness and subjective energy are responsive to catecholaminergic signaling. But that mechanistic possibility is a hypothesis, not the result of a properly controlled comparison.

The experiment asked whether assignment to a three-month course of dexamphetamine sulfate improved patient-reported fatigue more than assignment to an identical-appearing placebo. The design also monitored safety, quality of life, mood, and formal cognitive testing. Those outcomes answer different questions. A patient may notice stimulation without a measurable between-group fatigue benefit, and an adverse psychiatric complaint does not become evidence of superior cognition or fitness for a particular task. The authors’ report is therefore useful precisely because its efficacy and safety results did not move together.

How many independent people participated?

Patient flow matters more than counting every assessment. The report describes 50 initially eligible people, 48 randomized, and 46 who actually received assigned treatment: 23 dexamphetamine and 23 placebo. Subsequent noncompletion or ineligibility for the primary analysis left 41 evaluable patients at three months: 22 receiving dexamphetamine and 19 receiving placebo. These are distinct denominators, not inconsistent estimates of one group of people. The 41-person number belongs to the principal efficacy evaluation; 46 belongs to the treated safety population.

Monthly follow-up questionnaires, repeated cognitive assessments, and multiple adverse events within a single patient do not create extra randomized individuals. For example, a patient reporting anxiety and a sleep problem can contribute more than one event while remaining one patient. Combining all event rows as though they represented separate participants would falsely increase the apparent sample size. Similarly, the 46 treated patients are not equivalent to the 41 with a complete primary fatigue comparison.

Which patients were represented, and which were not?

The enrolled population had clinically significant fatigue, defined at screening by an MFI-20 global score of at least 60 on the study’s 100-point scale. Tumors were stable or responding by the study’s assessment rules. Gliomas accounted for most evaluable cases, while a smaller number had primary central nervous system lymphoma or medulloblastoma. Previous treatments were not identical across patients. In the 41 evaluable participants, 34 had received radiotherapy and chemotherapy, which underscores that fatigue in this clinical setting has multiple plausible contributors.

Exclusions narrowed the interpretation. The study screened for conditions that could independently cause fatigue or make testing unreliable, and it excluded participants with customary contraindications to amphetamine therapy, including relevant psychiatric and cardiovascular histories. Adults with actively progressive disease, unstable medical conditions, or important contraindications should not be assumed to respond as this selected cohort did. Nor does this oncology cohort represent healthy adults, children receiving ADHD medication, drivers, workplace personnel, or patients being evaluated after a suspected intoxicating exposure.

Was the treatment randomized and blinded?

Yes. This was a multicenter randomized comparison with a one-to-one intended allocation and matched placebo tablets. Both participants and the trial team were blinded under the reported design. Randomization helps reduce systematic differences between treatment groups, and masking addresses expectation effects in a symptom questionnaire. These strengths do not remove every limitation: sample sizes remained small, several people were not in the three-month evaluable set, and some underlying clinical factors could still vary between arms by chance.

The proper estimand is the contrast in outcomes between randomized treatment strategies, not the within-person change in the treated arm alone. A therapy group improving after a few months does not, by itself, establish efficacy. Fatigue may change during oncology follow-up even without an active stimulant. That is why an actual placebo comparison, with the reported p value and the outcome’s uncertainty, is indispensable.

What dose and route of dexamphetamine were actually administered?

Patients were assigned oral dexamphetamine sulfate tablets, not methamphetamine, lisdexamfetamine, intravenous amphetamine, or an unspecified stimulant. Dosing began at 5 mg in the morning and 5 mg at noon, or 10 mg total daily, during the first ten days. With acceptable tolerability, the schedule increased to 10 mg twice daily for the next ten days, then to 15 mg twice daily, a maximum planned daily total of 30 mg. Clinical assessments informed whether that escalation was tolerated; a patient could remain at a lower dose or stop the trial agent when adverse effects required it.

After the three-month evaluation, the investigators progressively reduced assigned medication over approximately seven days. A maximum allowed nominal daily dose must not be restated as the amount every participant actually ingested every day. The paper’s efficacy and safety analyses concern assignment to a titratable oral regimen; they do not provide a serial individual tablet-adherence record suitable for reconstructing an unknown exposure in a legal case.

How was fatigue assessed?

The prespecified principal measure was the Multidimensional Fatigue Inventory, MFI-20, evaluated at enrollment and after three months. This is a structured questionnaire, not an actigraphy measure, an electroencephalographic recording, a blood drug assay, or an objective test of sustained driving ability. A decrease in its score can represent improved reported fatigue. The question in a randomized trial is whether the improvement is greater with the active regimen than with placebo by an amount supported by the actual comparison.

Additional questionnaires evaluated mood and quality of life. Formal neuropsychological tests included Trail Making tasks, verbal fluency, and other assessments. These instruments sample performance or experience in controlled clinical contexts. They do not necessarily measure the same construct, and their scores cannot legitimately be combined into an invented forensic impairment threshold. The investigators prespecified the MFI-20 comparison, so a secondary pattern cannot silently replace it as the answer to the original treatment hypothesis.

What did the primary fatigue outcome actually show?

At baseline, the median MFI-20 score was 70.5 in the evaluable dexamphetamine group and 70 in the evaluable placebo group. At three months, the medians were 61 and 65, respectively. The source reports a median individual score variation of 14 points in the dexamphetamine arm and 10 points in the placebo arm. The trial’s between-group comparison was p = 0.17, conventionally nonsignificant. The report therefore did not demonstrate superiority of dexamphetamine for the planned fatigue endpoint.

These statistics must be described carefully: subtracting the two displayed group medians is not, in general, equivalent to the median of within-patient changes. The investigators supplied both baseline/follow-up medians and the distribution of patient-level variations. Nor does the difference between medians prove that the true treatment effect is exactly zero. With only 41 evaluable patients, the study has finite statistical precision. The accurate conclusion is that the anticipated benefit was not established, not that the drug could never change fatigue in any possible patient.

Quantitative evidence table 1: efficacy comparisons in the evaluable patients

Values are reorganized from the study’s original Table 5. The two n values refer to independent evaluable patients. An interquartile range describes variation across patients, not confidence that one specified patient improved. Negative and positive numerical variations should be understood within each test’s scoring convention rather than labeled universally beneficial or harmful. The study supplies no original legal whole-blood amphetamine concentrations to place alongside these scores.

Why is improvement in both groups not evidence of a drug effect?

Patients assigned placebo also reported lower fatigue scores at follow-up. Several explanations are compatible with such a pattern: ordinary fluctuation in symptoms, clinical support, regression toward an extreme screening value, coexisting treatments, changes in sleep, expectations, and time since previous oncology interventions. The study was not designed to assign percentages of the change to each possible explanation. It did, however, include a control arm precisely so that spontaneous or nonspecific change would not automatically be attributed to dexamphetamine.

It would be incorrect to write that a 14-point median improvement proves a 14-point stimulant benefit. That value is a within-arm patient-level change. The between-arm inference comes from comparing changes under the randomized strategies. Here the reported comparison did not reject the null hypothesis at the conventional level. Conversely, p = 0.17 is not a probability that the treatment is ineffective, and failure to reach significance is not a formal equivalence or noninferiority finding. Those distinctions are relevant whenever a clinical drug trial is used to support broader claims about functional capacity.

Did formal attention or executive-function tests improve more with medication?

The investigators tested more than fatigue. Their neuropsychological battery included Trail Making Parts A and B, verbal fluency, Mattis and Marin ratings, and a memory measure. The reported between-treatment comparisons on the analyzable tests were nonsignificant. For example, the Trail Making A change comparison yielded p = 0.29 and Part B yielded p = 0.61. A small numerical difference between treatment-group medians cannot be promoted to a confirmed cognitive effect after those results.

The modified Wisconsin Card Sorting Test deserves a separate qualification. Approximately half its datapoints were missing, so that test could not be analyzed adequately. Calling it a definitive negative trial result would hide a missing-data problem. Nor should an absent test result be interpreted as proof of adequate executive functioning. The broader finding is that the analyzed measures did not demonstrate treatment superiority, while one potentially relevant measure had insufficient data for an evaluation.

What happened to quality of life, anxiety, and depression?

Clinical quality of life was assessed using EORTC instruments, including the QLQ-C30 and a brain-tumor module. The study did not find significant between-group quality-of-life improvement attributable to dexamphetamine. The reported group-by-time p values for the illustrated QOL measures were 0.10 and 0.06, respectively. These are nonsignificant comparisons, even when one is numerically close to 0.05. Inferences should follow the stated analysis rather than visual impressions from the box plots.

The Hospital Anxiety and Depression Scale did not show a significant between-group advantage on those summary clinical scores either. This is a different question from whether patients experienced particular psychiatric adverse effects during treatment. A global questionnaire may fail to show a sustained group trajectory even when specific individuals report short-lived anxiety, irritability, activation, or disrupted sleep. Combining those results into a single claim such as “no psychiatric effects occurred” would conflate two different measurement systems.

Were psychiatric adverse events more common with dexamphetamine?

Yes, in this trial’s safety reporting. Fifteen of the 23 patients assigned dexamphetamine had a psychiatric-category adverse event, compared with seven of the 23 assigned placebo. The original comparison was p = 0.018. The investigators describe complaints including hyperactivity, anxiety, sleep disturbance, and irritability. These are safety observations recorded during treatment. The report does not provide serial whole-blood amphetamine concentrations linked to each complaint, and it does not determine the mechanism of each participant’s event.

The psychiatric adverse-effect difference was more prominent than the fatigue outcome’s treatment contrast. That asymmetry is not a contradiction. A drug can have central nervous system effects while failing to improve a specified clinical endpoint. An individual may experience unwanted activation without a sustained improvement in the multidimensional experience of tumor-related fatigue. The data support discussion of a group-level tolerability concern in selected tumor patients, not a diagnosis that every participant developed stimulant-induced psychiatric illness.

Quantitative evidence table 2: adverse events in all 46 treated patients

These rows combine distinct patient-count endpoints from original Tables 2 and 4 and its accompanying text. Categories can overlap within the same patient: the numbers must not be added to estimate unique affected patients. The source separately inventories hundreds of individual event reports, which are not

What else does the safety record establish—and what does it not?

Recording adverse events prospectively and comparing blinded groups is stronger than presenting uncontrolled anecdotes. Yet it is not the same thing as assigning a chemically proven dose–response mechanism to every symptom. Tumor-related disease, accompanying medication, sleep disruption, and stress can contribute to adverse experiences in both arms. The authors’ patient-count comparisons establish which categories differed in this study’s selected population and which did not. They do not produce a universally applicable risk percentage for another population without further evidence.

Consider the grade-3 finding: three dexamphetamine patients and two placebo patients experienced at least one grade-3 event, and the reported p value was 1.000. Those numbers do not justify saying dexamphetamine caused no severe event or that the regimens are definitively equally safe. They indicate no statistically detected difference with very limited counts. The lack of grade-4 events and treatment-related deaths is reassuring as an observation, not a reliable estimate of the lifetime probability of an uncommon event.

Was there evidence of a concentration–response relationship?

No source-specific quantitative exposure–response relationship was established because the experiment did not perform the necessary chemical measurements. The investigators documented the planned oral dosing schedule and monitored clinical outcomes, but they did not report measured parent dexamphetamine or metabolites from serial human venous whole-blood, plasma, or serum specimens. The report therefore contains no calibrated patient-specific ng/mL values, analytical detection limits, chromatographic confirmation results, or validated concentration–fatigue and concentration–adverse-event models.

A 30-mg/day dose is a prescribed amount of drug, not a measured concentration in the circulation. Individual absorption, distribution, metabolism, renal elimination, co-medication, and timing could change exposure even when prescriptions match. Nothing in this study permits reverse-calculating a person’s blood concentration at a particular hour from assignment alone. The randomized trial can evaluate a treatment regimen; it is not a controlled validation of a forensic drug concentration cutoff or a pharmacokinetic conversion method.

Why do specimen definitions matter to forensic interpretation?

A legal toxicology assessment begins with a defined physical specimen, collection conditions, analytical method, and analyte. Human venous whole blood includes cells and plasma and is not interchangeable with centrifuged plasma or serum, oral fluid, urine, vitreous humor, hair, sweat, milk, tissue, or breath aerosol. A source’s result for one matrix cannot be silently relabeled as a result for another. Our guide to biological matrices in toxicology explains why those distinctions control what a laboratory result can support.

Here the contrast is more fundamental: the paper reported no quantitative parent-drug specimen concentrations at all. Clinical adverse-event forms and the MFI-20 do not become chemical analyses merely because a psychoactive medication was administered. A validated laboratory method would need to specify specimen preparation, reference standards, instrument, identity confirmation, detection and quantification limits, units, calibration, and quality controls. The overview of confirmatory drug testing provides context for that separate type of evidence.

Can the trial establish intoxication or driving impairment in one person?

No. The study did not enroll drivers for an on-road or validated driving-simulation assessment, did not collect contemporaneous legally quantified whole-blood dexamphetamine concentrations, and did not test a defined legal impairment cutoff. Its clinical endpoints were patient-reported fatigue, oncology quality of life, neuropsychological testing, and treatment-period safety. Even a statistically significant psychiatric event category cannot tell a fact-finder whether a specific person was acutely intoxicated while driving, whether a witness perceived accurately, or whether a collision occurred because of stimulant exposure.

Nor can the result be converted to methamphetamine impairment or to a different administered stimulant formulation without new evidence. Dexamphetamine is a specified amphetamine stereoisomer salt; lisdexamfetamine is a prodrug that yields dexamphetamine; methamphetamine is a different drug. Pharmacological relatedness does not mean the same dosing, toxicokinetics, biological measurements, or legal interpretations. For an individual case, the relevant clinical record, actual specimen identity, defensible analytical testing, time course, and validated impairment evidence must be evaluated independently.

What are the principal statistical limitations?

The evaluable efficacy group of 41 was modest and came from a selected medically complex population. Tumor type, prior treatment, fatigue severity, sleep, and co-medication were heterogeneous even though the disease was required to be stable or responding. Randomization controls allocation in expectation, but it does not guarantee balance on every characteristic in a small sample. Selection into an evaluable three-month set introduces an additional issue: the estimand for those completing assessment need not equal a perfect all-randomized intention-to-treat effect.

Numerous secondary cognitive, mood, quality-of-life, and adverse-event endpoints were reported. A single nominal p value near 0.05 should not be treated as a summary of all outcomes, especially without independently confirmed control of multiplicity for the full set. Conversely, a p value above .05 does not prove equality. The MFI finding is best reported as failure to demonstrate superiority at three months in this trial. The psychiatric-event finding is an observed randomized-group safety signal worthy of attention, not a validated individualized prediction equation.

How should researchers and clinicians use these findings?

The trial contributes controlled evidence against assuming that a stimulant’s activating pharmacology necessarily improves multidimensional fatigue in patients with stable primary brain tumors. Its primary outcome did not support superiority of this dexamphetamine regimen over placebo, and the psychiatric adverse-event category occurred more frequently with the active treatment. A clinician deciding how to manage fatigue must also consider contemporary guidelines, disease status, other causes of fatigue, existing treatments, and a patient’s specific safety factors; this individual trial does not supply a universal treatment prohibition or prescription recommendation.

For research, stronger follow-up work would require adequately sized groups, carefully specified endpoints and missing-data handling, measurement of important confounders, and transparent reporting of meaningful clinical effect sizes with confidence intervals. For a pharmacokinetic question, actual serial specimen analysis would be required. For a traffic-safety question, a validated performance design would be necessary. These are separate research objectives that cannot be retrospectively attributed to the ANOCEF trial.

What can an expert responsibly say about this study?

A defensible summary identifies the source, design, population, independent denominators, drug and route, primary outcome, adverse-event signal, and what was not measured. It can say that median fatigue scores improved within both arms, but the specified between-arm comparison was nonsignificant at p = .17. It can also say that psychiatric-category adverse events were reported by 15/23 versus 7/23 treated patients, p = .018. It should avoid statements that the trial determined a human whole-blood drug threshold, ruled out a rare event, or proves a particular person’s capacity, diagnosis, credibility, or legal responsibility.

This distinction between controlled group comparison and individual causal inference is especially important in legal and forensic settings. Neither treatment assignment nor a questionnaire score directly documents a toxicological specimen. An independent expert should disclose the small sample, heterogeneous underlying disease and therapies, absent chemical drug concentrations, and lack of validated task-specific impairment endpoints. The study supports a bounded scientific interpretation rather than a shortcut from medication use to personal blame or exoneration.

Frequently asked forensic and clinical questions

1. Did dexamphetamine significantly reduce primary brain-tumor fatigue?

Not compared with placebo in this three-month phase III trial. Although the evaluable active group had a median patient-level score change of 14 points and the placebo group a change of 10, the primary between-treatment p value was 0.17. That result does not establish superiority, but it also is not a formal proof that all possible dose regimens have identical effects. The conclusion applies to this small selected oncology sample and its actual treatment strategy.

2. Were the 41 efficacy patients the entire treated study?

No. Forty-six patients received study medication or placebo, split 23 and 23, and formed the treated safety denominator. Forty-one patients, 22 active and 19 placebo, were evaluable for the main three-month fatigue comparison. Repeated monthly questionnaires and adverse-event reports are observations within those people, not newly randomized subjects. Always attach the correct denominator to the outcome being discussed.

3. Was the oral dose the same as a measured blood level?

No. The investigators started an assigned oral regimen at 10 mg/day, with tolerated escalation to 20 and then 30 mg/day. These values describe tablet administration, not measured circulation. The study did not report validated amphetamine concentrations from legal whole blood or separated plasma/serum, and it did not publish an exposure–response curve. A nominal prescription cannot establish a person’s quantitative drug level or time since last dose.

4. What psychiatric safety difference was actually statistically significant?

Fifteen of 23 dexamphetamine patients versus seven of 23 placebo patients were recorded with psychiatric-category treatment-period events, p = .018. The authors discussed anxiety, hyperactivity, sleep-related complaints, and irritability. Those reports are relevant to group-level tolerability. They are not biochemical diagnoses or proof that any particular event was caused exclusively by a measured concentration of dexamphetamine.

5. Did the trial show dexamphetamine improves cognitive testing?

No consistent statistically established treatment advantage emerged from the reported analyzable neuropsychological comparisons. Trail Making A and B variation comparisons were p = .29 and p = .61, respectively. The modified Wisconsin Card Sorting Test was not adequately analyzable because roughly half its observations were missing; that missing endpoint cannot be counted as either a demonstrated benefit or a definitive null result.

6. Does no grade-4 toxicity mean the medication is proven safe for everyone?

No. No grade-4 treatment toxicity or treatment-related death was observed, but only 46 treated people contributed to the safety assessment. Three active and two placebo patients experienced grade-3 events, a nonsignificant between-arm comparison. Small event counts are compatible with considerable uncertainty about rare harms, and patients with important contraindications were excluded before randomization. A rare-event safety claim requires a much larger and more representative evidence base.

7. Can this study support a forensic opinion about an impaired driver or specific incident?

Not by itself. It measured three-month clinical symptoms and treatment-emergent events in primary brain-tumor patients, not validated on-road driving behavior or case-specific intoxication. There was no quantitative human venous whole-blood drug assay, no individual threshold, and no retrospective reconstruction of exposure at a specific event time. A forensic opinion must rely on independently suitable specimen analysis, timing, case evidence, and validated limits of inference rather than on this therapeutic trial’s group-level p values.

The primary data and reporting details are available in the original publication: Laigle-Donadey et al., phase III dexamphetamine trial, DOI 10.1093/noajnl/vdz043. This article is an independent evidence-based explanation; it does not reproduce the original journal’s tables or figures.