Can a 140-mg Lisdexamfetamine Pilot Establish Cocaine Abstinence? What 17 Enrolled Adults Actually Showed

Answer first: An eight-week, open-label pilot of lisdexamfetamine for cocaine use disorder found that most of its 16 participants with follow-up data reached the intended 140-mg daily research dose, four required a dose reduction, and none discontinued the study medication because of an adverse effect. Self-reported cocaine-use days and daily spending declined significantly from baseline to the last three weeks of participation. But the trial had no placebo or comparison group, so the declines cannot be attributed causally to lisdexamfetamine. Although urine was collected for cocaine-metabolite testing, the article reported no randomized biochemical treatment contrast and no measured parent-stimulant concentration in forensic venous whole blood.

The original primary investigation was reported by Mariani and colleagues in 2021, Open-label pilot study of lisdexamfetamine for cocaine use disorder, The American Journal of Drug and Alcohol Abuse 47(3):402–409. Its original research is identified by DOI 10.1080/00952990.2021.1885677. Its useful contribution is a small, carefully monitored dose-tolerability feasibility observation—not a confirmed efficacy estimate, a forensic blood–drug calibration, or evidence that a particular person stopped using cocaine.

What is the central distinction between tolerating a dose and proving treatment efficacy?

The investigators primarily wanted to learn which doses of lisdexamfetamine participants could reach and continue while receiving clinical monitoring. Tolerability therefore occupied the foreground: daily amount achieved, medication adjustments for side effects, retention, and reasons for stopping. Reduced self-reported cocaine use was a secondary observation. That distinction changes how the numbers must be presented. A person remaining on medication despite an adverse effect provides one kind of information; a person using less cocaine after entering a treatment program provides another; neither alone establishes the effect that would have occurred in that person without the medication.

Open-label means clinicians and participants knew the assigned medication. Without a placebo group, improved outcomes after enrollment may also reflect participant motivation, structured appointments, treatment contact, the calendar-based recall interview, expectations, changes in availability of cocaine, or regression toward a less intense period of use. Statistical significance for change over time does not repair the missing counterfactual comparison. This was a feasibility study, not an experiment capable of separating the active molecule’s effect from those other explanations.

How many independent people were enrolled, and who contributed outcomes?

Seventeen treatment-seeking adults with DSM-IV cocaine dependence entered the study after recruitment at two academic sites. One supplied no post-enrollment information sufficient for the authors’ analyzed outcomes, leaving 16 adults in the reported follow-up calculations. Those numbers are different and should not be interchanged. The analyzed group is not sixteen medication visits or sixteen positive urine tests: it consists of sixteen distinct people. Their repeated visits, pill counts, calendar reports, vital-sign readings and urine collections are repeated observations within those individuals.

Among the seventeen enrolled, fifteen were men and two were women. The average age was 43.8 years, with standard deviation 8.4 years. The sample was selected rather than population-representative. Eligibility required reported recent cocaine use, willingness to participate in treatment, generally acceptable health, and the ability to attend monitoring. Multiple cardiovascular, psychiatric and other medical exclusions applied. That screening context substantially limits any inference about the cardiovascular safety of higher-dose stimulant treatment in the general population, emergency departments, unsupervised users or individuals with multiple active drug exposures.

What drug was administered, and what was not administered?

The research medication was oral lisdexamfetamine mesylate, a prodrug converted to dextroamphetamine. It is not methamphetamine, not cocaine and not a laboratory-standardized exposure to cocaine. Researchers did not administer cocaine at a known time, dose or route. They enrolled adults already experiencing cocaine dependence and then followed changes in those adults’ clinical and reported use patterns. Accordingly, the study cannot provide controlled cocaine pharmacokinetics or precise time-since-last-use estimates.

The investigators pursued a lisdexamfetamine target greater than the lower maximum used in the earlier randomized pilot that they discussed in their introduction. Those two investigations must be kept separate. The earlier 43-person blinded study examined lisdexamfetamine at 70 mg daily with a placebo comparator and reported a craving signal without a statistically significant full-sample urine-use effect. This 17-person open-label study was designed around upward dose titration toward 140 mg and tolerability. Combining their participants, treating their designs as equivalent, or asserting that the higher dose proved more effective would be scientifically indefensible.

How was the eight-week dosing schedule structured?

Participants were titrated during the first two weeks toward the intended study maximum, with the protocol describing a stepwise progression through 20, 40, 60, 70, 80, 100, 120 and 140 mg per day. The aim was to reach a tolerated maintenance amount, continue the highest tolerated amount for approximately four weeks, and then taper over two additional weeks. Instructions divided daily administration into two doses. For efficacy and tolerability summaries, the paper focused on the active treatment period rather than treating the subsequent taper as another efficacy period.

These amounts are a description of an investigator-supervised research protocol. They are not prescribing recommendations, safe-use advice or approved dosing for cocaine use disorder. The study was conducted in 2011–2012, published in 2021, and did not establish a general standard of care. Its clinical oversight included repeated pulse and blood-pressure observations, electrocardiograms at specified intervals, adverse-event assessments, and clinical authority to lower a dose when tolerability became problematic.

What did the actual dose and retention outcomes show?

Among sixteen people with post-enrollment data, the average maximum daily dose reached was 118.1 mg, standard deviation 40.4 mg. Inspection of the original participant-level dose table shows twelve at a reported highest dose of 140 mg, two at 60 mg, one at 70 mg and one at 20 mg. Thus, twelve of the sixteen analyzed individuals reached the target amount at least on the table’s highest-dose criterion. This does not prove that all twelve continuously took 140 mg throughout the maintenance period; “highest dose achieved” is not equivalent to measured adherence at every visit.

Average retention was 6.5 weeks, standard deviation 2.0 weeks, within an eight-week study. Four of sixteen participants experienced adverse effects requiring a downward dose adjustment while remaining in the study. The report did not describe anyone discontinuing study medication because of adverse effects, and it recorded no serious adverse event or safety-rule exit. These are useful observations about a carefully screened small cohort. They are not a statistical demonstration that high-dose lisdexamfetamine carries no risk, and they do not negate the four clinically consequential dose reductions.

Evidence table 1: actual participants, dose milestones and tolerability

A zero in the last two rows is not evidence that the underlying event probability is zero. With only sixteen analyzable people, uncommon outcomes could easily fail to occur by chance. The study also excluded participants with specified cardiovascular risks and provided monitoring that may not be available elsewhere. A safety conclusion confined to this particular research setting is more defensible than an unqualified statement about the molecule or any other dose.

Which adverse effects actually required the dose to be lowered?

The four reported adjustments were associated with recognizable stimulant-related clinical complaints, though the study design did not prove their precise mechanisms. One individual reported insomnia and anxiety; another decreased appetite, agitation, restlessness and dry mouth. A third had hypertension, while the fourth experienced tachycardia, hypertension and agitation. The clinicians responded by reducing doses, rather than automatically ending all medication. These observations help explain how the feasibility objective was achieved: flexibility and monitoring formed part of the intervention.

The absence of a serious event during the observation window must be read alongside those same symptom reports. There was no blinded reference group receiving an identical schedule of evaluations without lisdexamfetamine, and the sample was far too small to estimate rare cardiac outcomes. These data cannot establish that a particular person’s chest pain, arrhythmia, anxiety or blood-pressure change was or was not caused by a drug outside this trial.

Did participants report fewer days of cocaine use?

Yes. The reported mean percentage of days with cocaine use declined from 25% during the 28 days preceding enrollment to 12% during the last three weeks of participation. The authors’ within-person paired comparison yielded t(15) = 3.33 and p = .005. That is a genuine statistically significant decrease in the sample’s self-reported behavior between the selected time windows. It is not a placebo-adjusted estimate of the medication’s causal effect and does not establish sustained abstinence after treatment.

Calendar-based timeline follow-back interviews provided the use-day reports. Such methods can organize recall more rigorously than casual questioning, but reported use remains dependent on what participants knew, remembered and chose to disclose. Moreover, the initial window was the previous 28 days while the later window covered the last three weeks in a program that involved clinical contact, monitoring and behavioral support. Different windows and program participation complicate causal interpretation even when paired statistics are correctly calculated.

Did reported cocaine spending also decrease?

The mean reported amount spent on cocaine per day fell from $19.72 at baseline to $7.57 during the last three weeks of participation. The within-person paired test was t(15) = 3.60, p = .003. Spending and use-day frequency therefore both changed in the reported direction over time, but they are related self-report outcomes, not two independent biologic assays. Spending can also vary with local price, type or quantity acquired, credit, sharing and disclosure rather than reflecting a chemically measured amount consumed.

The paper includes individual-level spending and use-day values, showing substantial variation. Some people had a reported zero in the late period; others continued to report use. It is not scientifically sound to convert the group mean decline to a declaration that every participant stopped cocaine or that an unknown individual exposed to lisdexamfetamine would experience the same magnitude of benefit.

Evidence table 2: behavioral changes and the missing causal comparator

The same people generated multiple reports. A p value for a paired within-person change only addresses whether the observed differences are consistent with a null mean change under its model; it does not isolate treatment from expectancy, repeated assessment or concurrent support. In contrast, the urine row is about biological collections rather than reported days. It carries no invented p value because the source did not provide a controlled between-arm test of cocaine-metabolite positivity.

Were six people biochemically proven abstinent?

No such individual-level conclusion follows from the published summary. The report says six of the sixteen participants with outcome data were abstinent for their final three participation weeks; its use-history analyses relied on self-report. The article also tabulated the proportion of urine collections testing positive for cocaine metabolites across study visits, including a measure for the final three weeks. Several patient rows show that self-report and urine results need not coincide neatly. There is no basis to infer that each of the six had three weeks of uninterrupted, independently verified biochemical negativity with comprehensive specimen collection.

The participant-level table even reminds a careful reviewer why denominators matter. A reported zero-use interval may coexist with positive intermittent urine tests for some participants; an uncollected test is not a negative test. The study does not establish continuous observation of every use day. It cannot eliminate intermittent use between collections or determine exactly how long a specific participant had abstained before the last specimen. A forensic opinion about an actual individual would require the original laboratory reports, specimen dates, test targets and clinical chronology.

What specimen was actually collected for cocaine testing?

The study collected urine during screening, at entry and approximately twice weekly thereafter. It describes testing for cocaine metabolites; it does not provide a validated primary-method quantitation series for parent cocaine in simultaneously obtained human venous whole blood. Nor does it report a method for calculating cocaine dose, blood concentration or exact last-use time from the urine findings. A numerical percentage of cocaine-positive urine collections is a fraction of observed tests—not a concentration and not the percentage of participants who were continuously exposed.

The published Methods do not specify the full urine assay validation, an analytical decision cutoff for a particular cocaine metabolite, or a matched confirmatory chromatographic panel sufficient for a laboratory to reproduce the tests from this article alone. One should therefore avoid importing the 150-ng/mL urinary benzoylecgonine threshold described in a different lisdexamfetamine trial into this one. Two studies involving the same diagnosis and prodrug do not necessarily employ the same specimen procedures or definitions of a positive result.

The distinction between different biologic materials matters to every case. The independently published biological-matrices guide describes why results from urine and blood cannot be numerically interchanged. The separate published guide to confirmatory drug testing addresses the difference between a reported screen and a method specifically identifying or quantifying a compound. Neither general guide supplies missing method validation from the trial.

Were lisdexamfetamine or dextroamphetamine levels measured in blood?

Not as a quantitative research endpoint in this article. The assigned oral dose was recorded in milligrams per day, and the paper discussed how the prodrug is converted to dextroamphetamine. It did not provide serial, chemically measured parent-lisdexamfetamine or dextroamphetamine concentrations from participants’ legally collected venous whole-blood specimens. It also did not report a verified separation of d- and l-amphetamine in a forensic specimen. Discussion of red-cell conversion, prior pharmacokinetic publications, or a dose-equivalence estimate is not itself a blood measurement performed on these people.

For the same reason, the participant-specific 140-mg dose milestone cannot be converted into a universal nanogram-per-milliliter result, a legal impairment threshold, or a precise hour of administration. Dose, adherence, absorption, prodrug hydrolysis, co-medications, collection conditions and analytical methodology are independent considerations. A legally relevant whole-blood concentration must come from an appropriate sampled matrix and validated drug-specific method, not from a dosing table in a clinical feasibility report.

Did the study establish a safe maximum dose for all patients?

No. It supports a narrower descriptive observation: in this screened, monitored outpatient sample, a target of 140 mg was reached by a substantial fraction of participants, and four required downward adjustment rather than discontinuation for adverse effects. Selection criteria excluded clinically important risk groups. The small sample and relatively short observation period do not quantify rare cardiovascular, neuropsychiatric or longer-term outcomes. The report’s absence of serious adverse events is meaningful as an observation but is not a statistically tested claim of equivalent or universal safety.

Any clinical assessment would also need to distinguish the specific research protocol from contemporary prescribing rules and individualized medical decision-making. A study discussing doses above familiar commercial-label limits does not authorize those doses for an unrelated person. This article is evidence analysis, not medication advice. The clinical study’s monitoring procedures and risk exclusions are integral to its observed tolerability and cannot be separated from the numerical outcome.

How should the significant before-and-after changes be interpreted?

The two significant paired results answer a limited question: did participating individuals report different cocaine-use days and spending toward the end of treatment than before they entered it? They did. To answer whether lisdexamfetamine itself caused those changes, an adequately powered randomized comparison with an appropriate placebo and shared care is required. The authors explicitly recognized that nonspecific effects of trial participation could explain the apparent improvement. A causal conclusion would overstate the evidence even if every paired p value were much smaller than .05.

It is equally incorrect to say the study demonstrated that lisdexamfetamine was ineffective. An uncontrolled feasibility design cannot prove a drug effect either way. Its participants, dosing feasibility, clinician responses to adverse effects, retention and observed use reports can inform later trial planning. They cannot substitute for a prospective efficacy test, and no inferred comparison should be manufactured from published values belonging to a different cohort.

What genuine negative or null findings did the investigators report?

The clearest negatives are observed absence of particular events: no medication discontinuations because of reported adverse effects, no study exit for the stated safety criteria and no reported serious adverse events in this small monitored cohort. Those are descriptive counts, not null hypothesis tests. The article reported significant paired changes in two behavioral measures; it did not publish an inferential nonsignificant cocaine-use efficacy comparison against a placebo or a tested equivalence result for clinical safety. Calling any missing test a “negative efficacy result” would fabricate an analysis.

The study was also negative in the sense that it never produced certain kinds of forensic evidence: there was no drug-specific human whole-blood assay, no controlled cocaine dosing schedule, no validated urine-to-blood conversion, no quantified driving-performance outcome, and no event-causation experiment. “Not measured” is a statement about scope, not evidence of a zero concentration or normal function. The distinction protects both the research record and the subject of any unrelated forensic inquiry.

Why are comparisons with other lisdexamfetamine cocaine trials especially vulnerable to error?

Different studies may change several scientific variables simultaneously. This pilot enrolled seventeen people with no comparator and aimed for a 140-mg research target. A separate forty-three-person trial randomized participants to 70 mg or placebo and centered different cocaine-use and craving measures. Combining their participant counts would not create a newly randomized trial. Juxtaposing the lower-dose trial’s urine screening cutoff with this study’s dosing schedule would not establish that the same assay or decision rule was used. Likewise, treating the later study’s observed spending decline as a superior efficacy estimate would ignore differences in selection, contact, follow-up and analysis.

Evidence synthesis can compare these approaches in a carefully labeled review, but an original primary-research manuscript should preserve each paper’s identity and its own dominant question. The present analysis is about high-dose open-label tolerability and the inability of paired self-reports to prove causation, not a second retelling of the earlier blinded trial’s randomized urine-metabolite findings. That boundary also avoids presenting related but different studies as duplicates or independent replications of exactly the same endpoint.

Can these findings establish intoxication, driving impairment or a particular incident’s cause?

No. This research did not administer a standardized driving-simulator battery, perform on-road performance testing, construct a blood concentration–effect curve, or prospectively assess a particular accident. Urine testing for cocaine metabolites and a questionnaire about days of consumption do not directly measure moment-to-moment neurobehavioral capacity. A person’s individual tolerance, concurrent substances, illness, time of sampling and actual conduct cannot be recovered from the average responses of sixteen clinic patients.

For litigation, a careful reviewer would first identify whether the scientific proposition concerns presence of an analyte, amount in a validated specimen, timing, clinical effect, or specific causation. Evidence for one proposition must not be silently used as evidence for the others. If a case includes a positive urine screen, the original laboratory report and confirmation method control the identity and interpretive boundaries of that result. If it includes blood, the collection and assay must be evaluated independently rather than filled in with a clinical dosing table.

What is the appropriate bottom-line interpretation?

The study shows that a closely monitored, highly selected group of adults with cocaine use disorder could often reach an investigational lisdexamfetamine target of 140 mg, sometimes after a dose reduction, while reporting fewer cocaine-use days and less spending toward the end of participation. The two reported paired changes were statistically significant, but the unblinded, uncontrolled design does not establish medication efficacy. The source observed no serious safety event, but four of sixteen participants required a tolerability adjustment, and its size cannot establish universal safety.

Forensic toxicologists should preserve the distinction between clinical dose, patient report, intermittently collected urine testing and direct drug quantitation in the correct matrix. This trial established no validated human venous whole-blood concentration, individual last-use interval, driver impairment threshold, or causal explanation for a particular incident. Those limits are part of the result—not afterthoughts to an otherwise successful treatment headline.

Seven questions about the 140-mg lisdexamfetamine pilot

1. Did the study prove that 140 mg daily treats cocaine use disorder?

No. It had no randomized placebo group, so significant before-and-after self-reported use changes cannot demonstrate a medication-specific treatment effect. The dose target and tolerability were the main feasibility aims.

2. How many people received the medication, and how many contributed outcome data?

Seventeen adults enrolled; sixteen had post-enrollment information for the principal reported summaries. Repeated urine samples, pill counts and visits did not create additional independent participants.

3. Does reaching 140 mg show uninterrupted adherence at that dose?

No. Twelve of sixteen reached 140 mg as their tabulated highest daily dose. The table does not establish that each remained continuously at the target dose for every maintenance visit, much less the amount converted to dextroamphetamine in blood.

4. Were the paired reductions in cocaine-use days statistically significant?

Yes. Self-reported use-day percentages decreased from 25% to 12%, with t(15) = 3.33 and p = .005. The result is about change during participation and is not a placebo-adjusted causal efficacy estimate.

5. Did six participants have three weeks of laboratory-confirmed abstinence?

The study reported six participants abstinent in their final three weeks using its clinical follow-up information, but that summary does not independently validate uninterrupted biochemical absence for each person. Intermittent urine screens and self-reported calendars measure different things.

6. Did zero serious adverse events prove the dose was safe?

No. Zero reported serious events in this selected small sample is descriptive, not proof of a zero underlying risk. Four people required dose reductions for symptoms, and people with specified cardiovascular risks were excluded.

7. Can urine cocaine-metabolite results or the prodrug dose establish blood concentration or individual impairment?

No. This report provided no directly assayed matched human legal venous whole-blood stimulant or cocaine concentration, no validated urine-to-blood conversion, and no driving or specific-incident causal validation. Case-specific conclusions require specimen-appropriate testing and separate relevant evidence.

Original primary source and evidentiary limits

Mariani JJ, Choi CJ, Pavlicova M, Mahony AL, Brooks DJ, Grabowski J, Levin FR. Open-label pilot study of lisdexamfetamine for cocaine use disorder. American Journal of Drug and Alcohol Abuse. 2021;47(3):402–409. Read the original primary research by DOI.

This independent scientific interpretation describes a small uncontrolled clinical study and limits the claims to the outcomes it actually reported. It does not prescribe medication, assess a named person’s drug exposure or constitute an individualized forensic opinion.