MED01-P2 — ChatGPT Stage 1 independent pass — RAW, verbatim

Received 2026-10-06 via Danny’s paste into the Claude session (ChatGPT chat reply). Archived unedited below the line; nothing added, removed or reworded (the \& sequences inside some URLs are as pasted). Filed to git only under the §M1 independence firewall — NOT mirrored to Drive and NOT shown to Gemini until both researchers’ Stage 1 and Stage 2 are filed. First Stage 1 pass to arrive for P2. Verification has not begun; nothing in this file is verified.

Filing notes (Claude):

• An earlier ChatGPT attempt (same day) stopped before any research: the two attachments were not accessible to its file tools, and it reported prior MED01 context visible. Danny restarted in a new chat; per Danny, account memory was still on.
• This pass self-reports prior MED01 context visible in the account (disregarded by its own account), attachments received as pasted markdown, no Drive or project file opened, and no Probe Appendix seen.
• It reports the four canonical ESearch URLs inaccessible in its browsing environment and gives no counts (not estimated). The verifier will run the canonical counts.
• Self-identified model string: “GPT-5.6 Sol”.
• Danny’s same-day paste labelled “Gemini” was a duplicate of this text (same model line, same utm_source=chatgpt.com links) and is NOT filed as Gemini’s pass.

MED01-P2 — Stage 1 independent research pass

Brief: MED01-P2 core brief v1.0, FROZEN 2026-10-03. I followed the Stage 1 instruction to answer §§1–5 only and did not use or attempt to anticipate the Probe Appendix. Pasted markdown
Governing protocol: PROTOCOL-MEDICAL v0.5 BASELINE. Pasted markdown (2)
Model/product: GPT-5.6 Sol
Run date: 2026-10-03
Live browsing: Yes, including live PubMed pages, ClinicalTrials.gov, Health Canada, EMA, GOV.UK, NCBI Bookshelf/NASEM, and other web-indexed literature. The specified NCBI ESearch API URLs themselves were not retrievable in this browsing environment, so the four canonical counts are explicitly reported as unreproducible in this pass rather than estimated.

Context-isolation disclosure: prior MED01 context was visible to me in the account, including earlier P1/P2-related conversations. I disregarded that material when selecting findings and sources. I did not open any shared project Drive or any project file beyond the two attachments supplied here. The frozen brief requires the evidence map to describe the literature rather than render present-day efficacy or safety judgments; I have kept that boundary throughout. Pasted markdown

1. Narrative answer to A–I

A. Volume and shape of the literature

Canonical PubMed counts

I attempted the exact canonical expression specified in §4a, changing only the dates. The ESearch host returned inaccessible-tool errors for all four calls; therefore I do not report counts or reconstruct them from other searches. Under the brief, a count without a working URL is unreproducible, and estimation is forbidden. Pasted markdown

The broader shape is nevertheless clear from bibliometric work. A 2022 Scopus analysis, used here retrospectively, found a steep rise in cannabis/cannabinoid publications over the preceding two decades; across its full 1829–2021 dataset, medicine was the largest subject area, followed by pharmacology/toxicology/pharmaceutics and biochemistry/genetics/molecular biology. The largest publication-producing countries were the United States, United Kingdom, and Canada; major funding sponsors were heavily U.S.-based, particularly NIH/HHS/NIDA. This is not a P2-period source for what was believed clinically, but it is useful for retrospective mapping of output and geography. PubMed Central (PMC)

A second bibliometric analysis covering publications through 2017 similarly found medicine prominent and identified substantial European and North American output. It was published in 2018, so it is also retrospective, not period evidence. MDPI

The literature was heterogeneous. It encompassed clinical neurology and pain, addiction/psychiatry and epidemiology, receptor and endocannabinoid pharmacology, toxicology, driving studies, pregnancy cohorts, regulatory medicine, and pharmaceutical development. Bibliographic counts cannot capture non-indexed reports, conference literature, regulatory dossiers, non-English sources missing from the database, or studies that remained unpublished.

Natural shape across time

My literature-derived periodization is:

2000–04: consolidation of endocannabinoid pharmacology plus larger clinical trials of oral cannabinoid preparations, especially in MS; systematic reviews were also reassessing older antiemetic trials. The 2003 CAMS trial enrolled hundreds rather than dozens of participants. PubMed

2005–09: a standardized botanical THC:CBD oromucosal product (Sativex/nabiximols) accumulated trials and regulatory decisions; controlled smoked-cannabis neuropathic-pain studies appeared in North America; longitudinal psychiatric-harms synthesis matured; rimonabant demonstrated both the translational reach and hazards of pharmacologically manipulating the endocannabinoid system. Canada

2010–14: controlled cannabinoid studies extended across neuropathic pain and MS, registration became more visible, and professional/systematic assessments became more explicit about evidence stratification. The 2014 AAN review included 34 studies, eight rated Class I. PubMed

2015–17: large cross-condition evidence syntheses appeared; purified CBD development in severe childhood epilepsies moved from expanded-access pharmacology into phase-3-style randomized testing; drug–drug interaction work became clinically relevant; the 2017 National Academies assessment integrated both therapeutic and harms literatures. PubMed

These are analytic phases for this map, not fixed historical categories.

B. Science that redirected the field

By 2000, CB1, CB2, anandamide and 2-arachidonoylglycerol were already identified; period reviews describe an increasingly explicit endocannabinoid system comprising endogenous ligands, receptors and metabolic pathways and frame these components as pharmacological targets. This mattered because research no longer had to ask only what plant cannabis or THC did; investigators could target receptor signaling and endocannabinoid synthesis/degradation. PubMed

What this did not show was that manipulating any component would necessarily improve human disease. The most striking counterexample was rimonabant, a CB1 antagonist/inverse agonist. It progressed through human obesity development and received EU marketing authorization in 2006. By October 2008, EMA’s CHMP concluded its benefits no longer outweighed its risks, citing roughly doubled psychiatric-disorder risk versus placebo, and recommended suspension. The manufacturer subsequently discontinued development; EU authorization was formally withdrawn in January 2009. European Medicines Agency (EMA)

The other major endocannabinoid-targeting caution was FAAH inhibition. BIA 10-2474 was intended to inhibit the enzyme that degrades fatty-acid amides/endocannabinoids. In a phase I multiple-dose cohort, severe neurological injury occurred after 50 mg/day: among exposed volunteers described in the 2016 report, one became brain dead and others developed cerebellar, memory, consciousness and MRI abnormalities. The report did not establish the mechanism. A 2017 activity-based proteomics study subsequently showed BIA 10-2474 inhibited several additional lipases in human cells/tissues and altered lipid networks in human cortical neurons, but that was mechanistic evidence, not proof that those off-targets caused the clinical syndrome. PubMed

On the cannabinoid side, purified CBD became a distinct clinical-development object rather than a proxy for whole-plant cannabis. Expanded-access epilepsy work generated interaction signals—particularly with clobazam—and a 2017 randomized trial of a pharmaceutical CBD oral solution in Dravet syndrome supplied controlled human evidence. That development simultaneously sharpened the product-specificity problem: data on a purified oral pharmaceutical preparation cannot be generalized to dispensary cannabis, mixed extracts, inhaled products, or CBD products of different purity/dose. PubMed

C. What was studied in humans

The human clinical literature identified in this pass clusters most densely around multiple-sclerosis symptoms, neuropathic/chronic pain, chemotherapy-related nausea/vomiting, HIV/AIDS-related symptoms, severe epilepsy, sleep outcomes, and smaller bodies in rheumatic pain, movement disorders and several other neurologic indications. Period syntheses also searched glaucoma, Tourette syndrome, anxiety, depression, psychosis and other conditions, but evidence volume and design varied considerably. PubMed

Multiple sclerosis: oral THC/extracts → nabiximols

Keystone full extraction — CAMS, Zajicek et al., 2003

• Population: stable multiple sclerosis with muscle spasticity; 667 enrolled, 630 treated; 33 UK centers.
• N randomized/analyzed: 630 treated; 611 followed for primary endpoint.
• Interventions: oral cannabis extract, oral Δ9-THC, or placebo.
• Dose/titration: detailed titration not accessible to me from the source display consulted.
• Route: oral.
• Duration: 15 weeks.
• Comparator: placebo.
• Primary outcome/timepoint: change in overall spasticity on Ashworth scale at end of trial.
• Comparative primary result: no statistically detected treatment effect; cannabis extract vs placebo estimated mean-difference reduction 0.32, 95% CI −1.04 to 1.67; THC vs placebo 0.94, 95% CI −0.44 to 2.31; overall p=0.40.
• Other findings: patient-reported spasticity and pain favored active treatment; authors noted some unmasking.
• AEs/SAEs/withdrawals: not accessible to me from the abstract display used.
• Registration: no registration identified in my documented ClinicalTrials.gov/WHO-oriented web searches; the trial predates compulsory/near-universal prospective registration norms.
• Primary-outcome match: not assessable without a verified registry/protocol.
• Funder: Multiple Sclerosis Society grant shown by PubMed.
• Applicability: does not establish effects of smoked cannabis, nabiximols, retail cannabis, or other conditions. PubMed

Later MS development increasingly used nabiximols/Sativex, an oromucosal standardized cannabis extract containing THC and CBD. In the 2011 enriched-design trial, 572 entered a four-week single-blind nabiximols phase; 272 met a ≥20% improvement threshold and 241 responders were randomized to continue nabiximols or switch to placebo. The enrichment means the randomized estimate applies to initial responders rather than an unselected MS-spasticity population. ClinicalTrials.gov identifies the study as NCT00681538. PubMed

Handoff: condition-specific review of MS spasticity, pain, bladder symptoms, tremor, mobility, sleep and quality of life.

Neuropathic pain and smoked cannabis

Keystone full extraction — Abrams et al., 2007

• Population: adults with painful HIV-associated sensory neuropathy; stable ART or no ART; prior cannabis exposure required by registry.
• N: 50 completed the trial; randomized enrollment number is not shown in the abstract excerpt I could verify.
• Product: cannabis cigarette containing 3.56% THC versus cannabinoid-extracted placebo cigarette.
• Dose: smoked three times daily.
• Route: smoked/inhaled.
• Duration: 5 treatment days during inpatient study.
• Comparator: placebo cannabis cigarette.
• Primary outcomes: daily chronic-pain ratings and proportion with >30% reduction; experimental hyperalgesia testing also performed.
• Comparative result: median daily-pain reduction 34% with active cannabis versus 17% placebo, p=0.03; >30% pain reduction in 52% versus 24%, p=0.04.
• AEs/SAEs: abstract says no serious adverse events; detailed AE frequencies/withdrawals not accessible to me.
• Registration: NCT00046722, opened live; registry title “Effects of Marijuana on Neuropathic Pain in HIV-Related Peripheral Neuropathy: A Randomized, Double-Blind, Placebo-Controlled Study.”
• Primary-outcome match: registry material retrieved shows experimentally induced pain as an outcome, but I did not obtain a sufficiently complete historical registry version to adjudicate prospective primary-outcome matching.
• Funder: NIH extramural support is indexed; registry sponsor is Center for Medicinal Cannabis Research.
• Applicability: 5-day exposure, selected cannabis-experienced adults, one potency and smoked route; no long-term claim. PubMed

Keystone full extraction — Ware et al., 2010

• Population: adults with post-traumatic or postsurgical neuropathic pain.
• N: 23 recruited; 21 completed.
• Product: herbal cannabis at 0%, 2.5%, 6.0% and 9.4% THC.
• Dose: 25 mg plant material per inhalation, three times daily during exposure days.
• Route: smoked through a pipe.
• Design: randomized crossover.
• Duration: four 14-day periods; drug administered on first five days, followed by nine-day washout.
• Comparator: 0% THC cannabis.
• Primary outcome: daily average pain intensity, 11-point scale.
• Prespecified comparative contrast: 9.4% vs 0% THC: 5.4 vs 6.1; difference 0.7 points, 95% CI 0.02–1.4.
• AEs: common drug-related AEs at 9.4% included headache, dry eyes, burning in painful areas, dizziness, numbness and cough.
• SAEs/withdrawals: detailed counts not accessible to me from abstract.
• Registration: ISRCTN68314063, identifier stated by the publication; I was not able to independently open the ISRCTN record through the available search interface, so the registration is publication-reported rather than live-registry-verified here.
• Funder: not accessible to me from the abstract display.
• Applicability: very small trial and short exposure; route/product-specific. PubMed

Handoff: HIV neuropathy; traumatic/postsurgical neuropathy; other chronic neuropathic pain; route-specific comparisons.

Rheumatoid arthritis

A 2005 e-publication/2006 print RCT randomized 58 adults with rheumatoid arthritis to Sativex/nabiximols or placebo over five weeks, administered as evening oromucosal spray. It measured movement/rest pain, morning stiffness, sleep and DAS28. Because I could verify only the abstract and not a complete trial report or registration, I treat it as a small period RCT that expanded the indication map, not a foundation for a broad pain conclusion. PubMed

Handoff: inflammatory/rheumatic pain.

Severe epilepsy and purified CBD

The field changed sharply near the P2 boundary. The 2014 Cochrane review found only four low-quality reports of CBD in epilepsy and concluded that reliable efficacy conclusions could not be drawn; those reports involved small numbers and short exposure. PubMed

By 2017, a multicenter double-blind randomized trial had tested pharmaceutical CBD oral solution as adjunctive therapy in Dravet syndrome.

Keystone full extraction — Devinsky et al., 2017

• Population: 120 children and young adults with Dravet syndrome and drug-resistant seizures, on standard antiseizure therapy.
• N randomized: 120.
• Product: pharmaceutical formulation of cannabidiol oral solution.
• Dose: 20 mg/kg/day.
• Route: oral.
• Duration: 14-week treatment, compared with 4-week baseline.
• Comparator: placebo.
• Primary outcome: change in convulsive-seizure frequency.
• Comparative result: median convulsive seizures/month changed 12.4→5.9 CBD vs 14.9→14.1 placebo; adjusted median difference in change −22.8 percentage points, 95% CI −41.1 to −5.4; p=0.01.
• ≥50% responder: 43% vs 27%; OR 2.00, 95% CI 0.93–4.30; p=0.08.
• Seizure-free: 5% vs 0%; p=0.08.
• AEs: diarrhea, vomiting, fatigue, pyrexia, somnolence and abnormal liver-function tests occurred more frequently with CBD.
• Withdrawals: more withdrawals in CBD arm; exact count not accessible in the abstract retrieved.
• Registration: NCT02091375, identifier reported in the publication. I verified the identifier through the live PubMed record; a standalone ClinicalTrials.gov result for that exact record was not surfaced by my search, so I do not claim historical registry-field verification beyond identity.
• Primary outcome match: not adjudicated because historical registered fields were not retrieved.
• Funder: GW Pharmaceuticals.
• Applicability: specific syndrome, pharmaceutical oral CBD, adjunctive therapy, studied dose; not evidence for mixed cannabis or general epilepsy. PubMed

Handoff: Dravet syndrome; Lennox–Gastaut and other treatment-resistant epilepsies; pediatric CBD pharmacology and hepatic/DDI safety.

Chemotherapy-induced nausea/vomiting

The important P2 literature here was partly synthetic rather than new primary trial production. A 2001 systematic review pooled 30 randomized comparisons / 1,366 patients, overwhelmingly older trials, testing oral nabilone, oral dronabinol/THC and intramuscular levonantradol; none involved smoked cannabis. Thus it is a P2 source for the review’s 2001 conclusion, while much underlying primary evidence remains pre-2000/P1. PubMed

Handoff: chemotherapy-induced nausea/vomiting under modern antiemetic standards.

Cancer pain

Nabiximols was studied in opioid-treated cancer pain; a 2012 randomized placebo-controlled graded-dose trial is identifiable as a P2 RCT. I did not designate it keystone because this Stage 1 pass did not retrieve enough complete trial details to support a clinical candidate without over-extracting from a partial record. PubMed

Handoff: cancer pain and opioid-adjunctive cannabinoid therapy.

D. Products and regulatory decisions

The P2 period did not contain one undifferentiated “medical cannabis” intervention. Its major clinical objects included:

• smoked herbal cannabis with measured THC potency;
• oral THC/dronabinol;
• oral cannabis extracts containing multiple cannabinoids;
• nabiximols/Sativex, a standardized THC:CBD oromucosal spray;
• purified pharmaceutical CBD oral solution;
• synthetic cannabinoids such as nabilone;
• endocannabinoid-system drugs not themselves plant cannabinoids, notably rimonabant and FAAH inhibitors.

Sativex/nabiximols

Canada, 15 April 2005. Health Canada issued Sativex a Notice of Compliance with conditions (control no. 091289) for adjunctive symptomatic relief of neuropathic pain in adults with multiple sclerosis. The official record identifies the dosage form as spray, buccal route, cannabidiol 25 mg/mL and Δ9-THC 27 mg/mL. The later Health Canada page records that this indication was eventually removed in 2019; that later event is P3/outside the P2 window and is not used here to rewrite the 2005 status. Canada

Canada, 1 August 2007. Health Canada issued a second NOC/c, control no. 109461, for adjunctive analgesic treatment in adults with advanced cancer pain; this indication was likewise removed after the P2 window and is noted only as retrospective record context. Canada

United Kingdom, June 2010. MHRA granted marketing authorization to Sativex as add-on treatment for symptom improvement in patients with spasticity due to MS. The UK Home Office subsequently moved Sativex into the normal controlled-medicine scheduling framework; official GOV.UK material states this authorization and the scheduling process. GOV.UK

Rimonabant/Acomplia

EU, 19 June 2006. The European Commission granted centralized marketing authorization for Acomplia (rimonabant), adjunct to diet and exercise in obesity or overweight with specified risk factors. EMA product no. EMEA/H/C/000666. European Medicines Agency (EMA)

EU, 23 October–13 November 2008. CHMP recommended suspension after concluding benefits no longer outweighed risks; EMA reported an approximate doubling of psychiatric disorders compared with placebo. Marketing was suspended in member states on 13 November 2008. Manufacturer withdrawal followed, with formal EC withdrawal in January 2009. The 2009 withdrawal is a P2 milestone because it occurred inside 2000–2017 even though the presently hosted public statement was issued/reposted later. European Medicines Agency (EMA)

Nabilone/Cesamet

An FDA medical review for NDA 18-677 documents oral synthetic cannabinoid nabilone, 1-mg capsules, for severe nausea and vomiting associated with cancer chemotherapy and describes pivotal randomized crossover comparisons with prochlorperazine and placebo. The underlying approval history substantially predates P2; I therefore do not treat the existence of Cesamet as a new P2 approval event merely because the product re-entered U.S. commerce in the 2000s. The historical regulatory chain needs verifier-level docket review before assigning a specific P2 “reapproval” milestone. FDA Access Data

Purified CBD

The 2017 Dravet RCT was a pivotal clinical-development event, but U.S. FDA approval of Epidiolex occurred in 2018, outside P2. Therefore there is no P2 FDA approval event for purified CBD in this Stage 1 regulatory table; the process is handed to P3. The 2017 state was: phase-3 clinical evidence existed, but the U.S. approval process remained open as of 2017-12-31. PubMed

E. Study quality and reporting

The design profile changed materially across the era. Early P2 syntheses often relied on small, older crossover trials, especially in chemotherapy-related nausea/vomiting. By the mid/late P2 period, the literature included multicenter randomized parallel trials, crossover inhaled-cannabis experiments, enriched randomized-withdrawal designs, and eventually large pediatric pharmaceutical trials. PubMed

Important recurring limitations:

Small samples/short duration. Ware’s neuropathic-pain study had only 23 recruits and five active-treatment days per crossover cycle. Abrams’ trial lasted five treatment days. Such designs can assess short-term symptom and acute AE outcomes but cannot establish chronic safety. PubMed

Blinding fragility. Psychoactive THC can reveal treatment assignment. CAMS explicitly reported a degree of unmasking. This is a recurring interpretive issue for THC-rich interventions, particularly in crossover designs. PubMed

Objective versus patient-reported outcomes. CAMS is a particularly useful example: its primary Ashworth-scale analysis did not show a treatment effect, whereas patient-reported spasticity/pain measures favored active treatment. The difference should not be collapsed into either “worked” or “failed.” PubMed

Enrichment. The 2011 nabiximols study randomized only participants who first improved by ≥20% during an initial nabiximols phase. That improves power to assess maintenance among responders but narrows generalizability. PubMed

Registration changed over time. Modern keystones such as NCT00681538 and NCT02091375 are identifiable; older trials frequently predated prospective-registration norms. The Abrams study is linked to NCT00046722. ClinicalTrials.gov

Product heterogeneity. The JAMA 2015 review pooled a literature containing multiple cannabinoid preparations and indications. Even where meta-analysis was possible, the result cannot be treated as evidence about every cannabis product. PubMed

F. Conditions of research

F1. Natural phases

The strongest literature-derived turning points I identify are:

1. Large oral-cannabinoid clinical trials emerge — exemplified by CAMS in 2003.
2. Standardized botanical pharmaceuticals and regulator-facing development become prominent — Sativex decisions beginning in Canada in 2005 and UK in 2010.
3. Controlled inhaled-cannabis trials reappear in North America — HIV neuropathy and other neuropathic pain, 2007–2010.
4. Endocannabinoid drug-development failures become scientifically consequential — rimonabant 2008 and BIA 10-2474 in 2016.
5. Purified CBD enters modern randomized pediatric drug development — Dravet syndrome, 2017. PubMed

F2. Law, access and supply

The most clearly documented research-conditions story in this search is U.S.-specific and should not be universalized.

The 2017 National Academies report documented multiple federal and state approvals for U.S. clinical cannabis research—FDA IND review, DEA registration/security requirements, NIDA supply processes and sometimes state-level approvals. It concluded that Schedule I classification and product access created research barriers. The report also documented that, as of December 2016, University of Mississippi remained NIDA’s sole supplier of research cannabis and that available research material did not reproduce the full product diversity of state markets. NCBI

Two procedural changes occurred within P2. HHS discontinued a duplicative Public Health Service review in June 2015, and in August 2016 DEA announced a policy intended to increase the number of registered marijuana manufacturers supplying researchers. DEA’s policy statement explicitly said its purpose was to facilitate legitimate research. NCBI

The crucial distinction is between documented administrative conditions and claims about how many studies they prevented. NASEM concluded the barriers impeded research; I did not identify a period dataset capable of quantifying how many trials were never initiated because of any one regulation. NCBI

Canada and the UK provide a different research environment in the record: standardized nabiximols proceeded through conventional pharmaceutical trials and medicinal-product authorization while remaining a controlled substance. Health Products Canada

F3. Geography and funding

The P2 literature was international: CAMS was UK-based; nabiximols trials were multinational; smoked-cannabis neuropathy work was prominent in California/Canada; the CBD Dravet trial involved centers in the U.S., UK/Europe and Australia. PubMed

Retrospective bibliometrics place the U.S., UK and Canada among the most productive countries and NIH/HHS/NIDA among the largest funders across the broader corpus. PubMed Central (PMC)

Funding sources differ by evidence strand: CAMS lists Multiple Sclerosis Society support; the Abrams study indexes NIH extramural support and CMCR sponsorship; the Dravet RCT was funded by GW Pharmaceuticals. These differences matter because the literature was generated through a mix of public, charitable and commercial pharmaceutical funding. PubMed

G. Harms research

The harms literature was not one body. It contained experimental exposure studies, prospective cohorts, case-crossover designs, population epidemiology, addiction laboratory studies, pregnancy meta-analysis, therapeutic-trial adverse-event reporting and pharmacokinetic interaction studies.

I did not find a defensible quantitative basis in this pass for saying the era devoted “more” attention to benefits or harms overall.

Dependence and withdrawal

Full extraction — Budney et al., 2003

• Design: controlled outpatient observational/withdrawal study.
• Population: 18 marijuana users; parallel assessment in 12 ex-users.
• Exposure: usual cannabis smoking for 5 days followed by 45 days abstinence.
• Comparator: ex-users plus within-person smoking-to-abstinence course.
• Outcome: withdrawal symptom timing/severity.
• Result: aggression, anger, anxiety, decreased appetite/body weight, irritability, restlessness, shakiness, sleep problems and stomach pain followed a reproducible time course; onset generally days 1–3, peak days 2–6, most lasting 4–14 days.
• Adjustment: not applicable — design.
• Funder: not extracted from accessible display.
• Applicability: heavy/current users; does not estimate withdrawal prevalence in all cannabis users. PubMed

A 2004 period review judged converging laboratory and clinical evidence sufficient to characterize cannabis withdrawal as reliable after cessation of chronic heavy use, while acknowledging questions of magnitude and clinical importance. DSM-5 subsequently incorporated cannabis withdrawal during the P2 period. PubMed

Psychosis and psychiatric outcomes

The 2007 Moore et al. systematic review included 35 longitudinal population-based studies. For psychotic outcomes, ever-use was associated with pooled adjusted OR 1.41 (95% CI 1.20–1.65). As observational evidence, this supports association and temporal-risk mapping, not a simple individual causal claim. PubMed

By 2017, NASEM rated the statistical association between cannabis use and development of schizophrenia/other psychoses as substantial evidence, with greater risk among more frequent users. That is NASEM’s dated evidence rating, not a new rating assigned by this pass. NCBI

Cognition and adolescent exposure

Full extraction — Meier et al., 2012

• Population: Dunedin birth cohort, N=1,037, born 1972/73 and followed to age 38.
• Exposure: repeated cannabis-use/dependence assessments at ages 18, 21, 26, 32 and 38; adolescent-onset examined.
• Comparison: differing persistence/onset patterns within cohort.
• Outcome: neuropsychological performance comparing age-13 pre-exposure testing with age-38 testing; informant-reported cognitive problems.
• Time window: childhood to midlife.
• Principal finding: persistent cannabis use was associated with broader neuropsychological decline; greater impairment appeared among adolescent-onset persistent users.
• Adjustment: article reports analyses addressing education and several alternative explanations; full covariate specification would require extraction from the manuscript.
• Funder: PubMed lists NIH/NICHD/NIMH/NIA/NIDA and UK MRC grants.
• Applicability: observational association; exposure is not randomized and should not be phrased as definitive proof of neurotoxicity. The authors used stronger causal language in their interpretation, but this map retains the design limitation. PubMed

NASEM in 2017 rated evidence for acute impairment in learning/memory/attention as moderate, while evidence for persistent impairment after sustained abstinence was limited. NCBI

Driving and psychomotor impairment

Hartman et al. conducted a controlled driving-simulator study examining cannabis with and without alcohol. Its product/exposure was controlled cannabis administration and the study measured lane-position/lateral-control outcomes. This experimental literature complements epidemiological crash-risk studies rather than replacing them. PubMed

NASEM’s 2017 synthesis rated the statistical association between cannabis use and increased motor-vehicle crash risk as substantial evidence. NCBI

Pregnancy and lactation

A 2016 systematic review/meta-analysis evaluated prenatal cannabis exposure and maternal/child outcomes. Exposure ascertainment and tobacco/other confounding were recurring issues in the underlying observational literature. PubMed

NASEM in 2017 concluded there was substantial evidence of a statistical association between maternal cannabis smoking and lower birth weight, limited evidence for maternal pregnancy complications and NICU admission, and insufficient evidence for several later offspring outcomes. This conclusion largely concerned smoked prenatal cannabis exposure, not every cannabinoid product/route. NCBI

I did not identify a comparably developed direct lactation outcome literature during this search. That is “none identified in my searches,” not evidence that no studies existed.

Respiratory outcomes

Tashkin’s 2013 period review summarized evidence that regular marijuana smoking was associated with large-airway injury and chronic-bronchitis symptoms, with symptoms tending to diminish after cessation. PubMed

A 2012 longitudinal cohort of 299 habitual users followed for a mean 9.8 years examined changes in marijuana/tobacco smoking and chronic bronchitis. Another Dunedin analysis found frequent cannabis use associated with cough, sputum and wheeze and reductions after quitting. These are observational associations involving smoked cannabis. PubMed

Cardiovascular outcomes

Full extraction — Mittleman et al., 2001

• Design: case-crossover observational study among patients with acute MI.
• Population: people presenting with myocardial infarction; exact analyzed N not extracted from the PubMed display available to me.
• Exposure: self-reported marijuana use immediately preceding MI.
• Comparison: each patient’s usual exposure pattern/control periods in case-crossover design.
• Outcome: onset of myocardial infarction.
• Principal estimate: not extracted from the accessible result display, so I do not reproduce a remembered estimate.
• Adjustments: not accessible to me.
• Funder: not extracted.
• Interpretation: period evidence investigating acute cardiovascular triggering; association/design-specific, not a general chronic cardiovascular-effect claim. DOI 10.1161/01.CIR.103.23.2805, PMID 11401936. PubMed

Acute toxicity and poisonings

By 2017, NASEM found moderate evidence associating cannabis with overdose injuries, including respiratory distress, in pediatric populations in U.S. states where cannabis was legal. It simultaneously found insufficient evidence concerning death from cannabis overdose. These are distinct propositions. NCBI

Drug–drug interactions

Full extraction — Geffrey et al., 2015

• Population: 13 children with refractory epilepsy taking clobazam while receiving CBD under expanded-access IND 119876.
• Design: observational pharmacokinetic analysis within expanded access.
• Product: CBD, formulation/dose details not fully accessible from abstract.
• Route: presumably oral within the IND but route is not explicit in the accessible abstract, so not inferred.
• Comparator: baseline concentrations before/early in CBD exposure.
• Outcome: clobazam and N-desmethylclobazam concentrations, seizure frequency, adverse effects.
• Result: mean clobazam level +60% ±80% at four weeks; N-desmethylclobazam +500% ±300%; 10/13 had clobazam dose reductions; side effects reported in 10/13 and improved after dose reduction.
• Funder: not accessible to me from abstract.
• Applicability: very small uncontrolled interaction study; seizure changes cannot isolate a CBD treatment effect from clobazam/metabolite changes and other factors. PubMed

A larger 2017 expanded-access analysis identified dose-associated changes in serum levels of several antiseizure medications and more frequent sedation at higher N-desmethylclobazam levels, plus higher AST/ALT values among participants on valproate. PubMed

Therapeutic-trial harms

The 2017 CBD Dravet trial reported higher frequencies of diarrhea, vomiting, fatigue, fever, somnolence and abnormal liver-function tests in the CBD group, plus more withdrawals. PubMed

The short-term smoked-cannabis neuropathic-pain trials reported psychoactive and respiratory/irritant-type adverse events; absence of serious events in trials lasting only days cannot establish long-term safety. PubMed

H. What period syntheses said

Tramèr et al., 2001 — chemotherapy-induced nausea/vomiting

This quantitative systematic review rested on 30 randomized comparisons involving 1,366 patients, largely pre-2000 primary studies. It concluded that oral nabilone, oral dronabinol/THC and intramuscular levonantradol had greater antiemetic efficacy than several then-used comparators, while adverse effects were also important. As required by the boundary rule, the review itself is P2; its older trials remain P1 evidence. PubMed

Koppel/AAN, 2014

The American Academy of Neurology systematic review included 34 studies, eight rated Class I. For MS it judged oral cannabis extract effective and nabiximols/THC probably effective for some patient-centered spasticity measures; evidence varied by outcome and condition, and evidence for several movement disorders and epilepsy was inadequate. These are AAN’s contemporaneous classifications, not this program’s certainty judgments. PubMed

Gloss & Vickrey/Cochrane, 2014 — epilepsy

The Cochrane review found the available CBD epilepsy reports too small/low-quality to support reliable efficacy conclusions and stated that long-term safety could not be assessed. This is especially important historically because it shows that the state of evidence in 2014 differed substantially from the 2017 Dravet RCT. PubMed

Whiting et al., JAMA 2015

This systematic review/meta-analysis searched 28 databases through April 2015 and covered randomized clinical trials in nausea/vomiting, appetite/HIV, chronic pain, MS/paraplegia spasticity, depression, anxiety, sleep, psychosis, glaucoma and Tourette syndrome. Its role in this map is as a period synthesis demonstrating both the breadth of indications and the heterogeneity of products and evidence. PubMed

National Academies, 2017

NASEM applied its own evidence categories across therapeutic and harms domains. It concluded there was conclusive/substantial evidence for treatment of chronic pain in adults, oral cannabinoids for chemotherapy-induced nausea/vomiting, and oral cannabinoids for patient-reported MS spasticity; moderate evidence for some short-term sleep outcomes; and limited or insufficient evidence for numerous other proposed therapeutic uses. NCBI

On harms, NASEM found substantial evidence for associations including schizophrenia/psychoses, motor-vehicle crashes and lower birth weight after maternal cannabis smoking; other endpoints received moderate, limited or insufficient ratings. NCBI

It also formally concluded that U.S. regulatory barriers, product-access constraints, funding structure and methodological heterogeneity impeded research. NCBI

I. Synthesis targets

I-1. Era table

“Approximate volume” here is qualitative because the canonical PubMed counts could not be reproduced and I did not run a condition-level systematic census.

I-2. Areas for which no human study was identified in my documented searches

This means only that I did not identify one in this Stage 1 search.

• direct human trials establishing disease modification, rather than symptom relief, for most major chronic neurologic indications;
• controlled human therapeutic evidence for lactation-specific cannabinoid exposure outcomes;
• robust human therapeutic trial evidence for many proposed psychiatric indications such as depression or PTSD by the end of 2017;
• convincing controlled evidence for cannabinoids as anticancer/antitumor therapy in humans; preclinical anticancer literature does not answer that question;
• human therapeutic trials sufficient to establish broad effects of minor cannabinoids beyond THC/CBD in the major clinical indications mapped here;
• robust long-term randomized safety studies of smoked cannabis;
• evidence allowing trial results with pharmaceutical CBD or standardized nabiximols to be generalized to heterogeneous non-pharmaceutical retail products.

I-3. Open threads handed to P3

• FDA/EMA and other approvals of pharmaceutical purified CBD after the 2017 randomized epilepsy program.
• Results/publications from trials completed or ongoing at the P2 boundary.
• Later reinterpretations of P2 observational harms, including cognition/adolescent-exposure analyses.
• Later confirmatory or failed nabiximols studies and indication changes.
• Post-2017 evolution of U.S. research-supply regulations.
• Post-2017 studies of cannabis formulations that better resemble legal-market edibles, concentrates, vaporized products and commercial CBD preparations.
• Any post-2017 re-analysis/integrity correction affecting P2 keystone studies.

I-4. End-of-era evidence-state statement

Replicated or controlled human evidence identified by end-2017: symptom-focused randomized evidence existed for multiple-sclerosis spasticity, several neuropathic/chronic-pain settings, older cannabinoid antiemetic indications, and by 2017 purified oral CBD for Dravet syndrome. The underlying interventions were heterogeneous—oral THC/extracts, nabiximols, smoked cannabis, synthetic cannabinoids and purified CBD—and should remain separate. Period syntheses assigned their own certainty levels, most comprehensively NASEM in 2017. PubMed

Preliminary human evidence or non-human evidence only: many proposed indications remained dependent on small pilots, uncontrolled studies, mechanistic rationale or animal evidence. Epilepsy illustrates how quickly that category could change: the 2014 Cochrane review could not draw reliable efficacy conclusions, while the 2017 Dravet RCT supplied controlled syndrome-specific evidence for pharmaceutical CBD. PubMed

Claims appearing in the period for which this search did not identify traceable controlled human support: broad disease-modification claims, generalized anticancer treatment claims, and generalizations from one cannabinoid/formulation/route to “cannabis” as a class. These should remain unsupported or condition-specific until later rounds examine them.

2. Candidate claims

CAND-01

Claim: The cannabis/cannabinoid literature expanded substantially during the P2 era and was distributed across medicine, pharmacology/toxicology, molecular biology, neuroscience and addiction-related journals rather than a single clinical field.
Type: literature-shape.
Support: “A bibliometric analysis of the cannabis and cannabinoid research literature,” PMID 35610633, DOI 10.1186/s42238-022-00133-0. PubMed Central (PMC)
Source: retrospective primary × observational/bibliometric.
Confidence: high for direction, moderate for P2-specific detail.
Would change mind: a methodologically broader bibliometric census showing no increase or radically different field distribution.

CAND-02

Claim: Retrospective bibliometrics place the United States, United Kingdom and Canada among the leading contributors to the broader cannabis/cannabinoid publication corpus, with U.S. NIH/HHS/NIDA prominent among funders.
Type: literature-shape.
Support: PMID 35610633, DOI 10.1186/s42238-022-00133-0. PubMed Central (PMC)
Source: retrospective primary × observational/bibliometric.
Confidence: moderate-high.
Would change mind: P2-restricted bibliometrics materially reversing the ordering.

CAND-03

Claim: In the 2003 CAMS trial, oral cannabis extract and oral Δ9-THC did not improve the prespecified Ashworth-scale primary outcome versus placebo over 15 weeks, although patient-reported spasticity/pain measures favored active treatment.
Type: clinical-finding-of-the-period.
Support: “Cannabinoids for treatment of spasticity and other symptoms related to multiple sclerosis (CAMS study),” PMID 14615106, DOI 10.1016/S0140-6736(03)14738-1. PubMed
Source: period primary × rct.
Population/product/etc.: 630 treated adults with stable MS/spasticity; oral cannabis extract or Δ9-THC vs placebo; 15 weeks; Ashworth primary outcome; cannabis extract Δ −0.32? Precise direction represented above as reported contrast 0.32, CI −1.04 to 1.67; THC 0.94, CI −0.44 to 2.31; p=0.40. AEs: not accessible to me. Registration: none identified.
Confidence: high for primary result.
Would change mind: full-text correction or integrity notice altering the primary analysis.

CAND-04

Claim: Nabiximols development introduced a standardized THC:CBD oromucosal pharmaceutical program and, by 2011, included an enriched randomized-withdrawal MS-spasticity trial in initial responders.
Type: literature-shape.
Support: Novotna et al., PMID 21362108, DOI 10.1111/j.1468-1331.2010.03328.x; NCT00681538. PubMed
Source: period primary × rct; evolving registry record.
Confidence: high.
Would change mind: registry/publication identity mismatch.

CAND-05

Claim: In a 2007 randomized trial of adults with HIV-associated sensory neuropathy, smoked cannabis containing 3.56% THC three times daily for five days produced a larger reduction in reported pain than cannabinoid-extracted placebo cannabis.
Type: clinical-finding-of-the-period.
Support: “Cannabis in painful HIV-associated sensory neuropathy,” PMID 17296917, DOI 10.1212/01.wnl.0000253187.66183.9c; NCT00046722. PubMed
Source: period primary × rct.
Population/product: HIV painful neuropathy; smoked 3.56% THC cannabis; placebo cannabis; 5 days; 34% vs 17% median daily-pain reduction, p=.03.
AEs: no serious AEs in abstract; complete AE counts not accessible.
Confidence: high for short-term comparative result.
Would change mind: full-text or registry inconsistency affecting prespecified outcome.

CAND-06

Claim: In a 2010 crossover RCT of post-traumatic/postsurgical neuropathic pain, 25 mg of 9.4%-THC herbal cannabis inhaled through a pipe three times daily for five days produced a 0.7-point lower mean pain score than 0%-THC cannabis on an 11-point scale.
Type: clinical-finding-of-the-period.
Support: PMID 20805210, DOI 10.1503/cmaj.091414, publication-reported ISRCTN68314063. PubMed
Source: period primary × rct crossover.
N: 23 recruited/21 completed.
AEs: headache, dry eyes, burning, dizziness, numbness, cough among common drug-related events at 9.4%.
Confidence: high for published result, moderate for registration verification.
Would change mind: verified registry showing outcome switching or major carryover problems.

CAND-07

Claim: In 2014, the Cochrane epilepsy review concluded that the then-available small CBD trials did not permit reliable conclusions about efficacy or long-term safety.
Type: synthesis-conclusion.
Support: “Cannabinoids for epilepsy,” PMID 24595491, DOI 10.1002/14651858.CD009270.pub3. PubMed
Source: period sr-ma × rests on small rct/interventional studies.
Confidence: high.
Would change mind: correction to the review.

CAND-08

Claim: By 2017, a randomized 120-patient Dravet trial showed pharmaceutical oral CBD 20 mg/kg/day produced a greater reduction in convulsive-seizure frequency than placebo, with more adverse events and withdrawals.
Type: clinical-finding-of-the-period.
Support: “Trial of Cannabidiol for Drug-Resistant Seizures in the Dravet Syndrome,” PMID 28538134, DOI 10.1056/NEJMoa1611618, NCT02091375. PubMed
Source: period primary × rct.
Result: adjusted median difference −22.8 percentage points, 95% CI −41.1 to −5.4; p=.01.
AEs: diarrhea, vomiting, fatigue, pyrexia, somnolence, abnormal LFTs; more withdrawals.
Confidence: high.
Would change mind: verified correction/integrity issue or outcome-registration mismatch.

CAND-09

Claim: A 2001 systematic review of 30 randomized comparisons/1,366 patients concluded that older cannabinoid antiemetic regimens outperformed several comparator antiemetics, while the underlying primary trials were largely pre-P2.
Type: synthesis-conclusion.
Support: PMID 11440936, DOI 10.1136/bmj.323.7303.16. PubMed
Source: period sr-ma × rests on rct ×30 comparisons; underlying primaries largely prior.
Confidence: high.
Would change mind: chronology of included trials showing substantial post-2000 primary evidence.

CAND-10

Claim: In 2015, Whiting et al. documented a randomized cannabinoid literature spanning multiple indications and products rather than one uniform intervention.
Type: literature-shape.
Support: PMID 26103030, DOI 10.1001/jama.2015.6358. PubMed
Source: period sr-ma × rests primarily on rct.
Confidence: high.
Would change mind: none short of correction/retraction.

CAND-11

Claim: The 2014 AAN systematic review found that the neurologic cannabinoid evidence base included 34 studies, eight rated Class I, with the strongest period conclusions concentrated in some MS symptom outcomes rather than across all neurologic disorders.
Type: synthesis-conclusion.
Support: PMID 24778283, DOI 10.1212/WNL.0000000000000363. PubMed
Source: period sr-ma/guideline-like evidence assessment × mixed rct.
Confidence: high for 2014 conclusion.
Would change mind: correction or retirement rationale materially undermining historical conclusion.

CAND-12

Claim: Health Canada issued Sativex a conditional NOC on 15 April 2005 for adjunctive symptomatic relief of neuropathic pain in adults with MS; the product was a buccal spray containing THC 27 mg/mL and CBD 25 mg/mL.
Type: product-regulatory.
Support: Health Canada NOC record, control 091289, DIN 02266121. Health Products Canada
Source: evolving official record; event 2005-04-15; retrieved 2026-10-03.
Confidence: high.
Would change mind: archived 2005 NOC contradicting current regulatory history.

CAND-13

Claim: Health Canada issued a conditional Sativex authorization on 1 August 2007 for adjunctive analgesic treatment in adults with advanced cancer pain.
Type: product-regulatory.
Support: Health Canada NOC/c record, control 109461. Canada
Source: evolving official; event 2007-08-01.
Confidence: high.
Would change mind: primary archived NOC discrepancy.

CAND-14

Claim: MHRA authorized Sativex in June 2010 as add-on treatment for symptom improvement in patients with MS spasticity.
Type: product-regulatory.
Support: UK Home Office/MHRA regulatory account. GOV.UK
Source: period-event official; current hosted record retrieved 2026-10-03.
Confidence: high.
Would change mind: original MA showing materially different indication wording/date.

CAND-15

Claim: The EU authorized rimonabant/Acomplia in 2006 for obesity/overweight with risk factors and suspended it in 2008 after CHMP concluded benefits no longer outweighed risks, including an approximate doubling of psychiatric-disorder risk versus placebo.
Type: product-regulatory.
Support: EMA EPAR EMEA/H/C/000666; 2008 CHMP communication EMEA/CHMP/537777/2008. European Medicines Agency (EMA)
Source: official/evolving record.
Confidence: high.
Would change mind: original decision documents contradicting current EMA chronology.

CAND-16

Claim: The BIA 10-2474 phase-I program demonstrated that endocannabinoid-system drug development could produce severe unexpected human toxicity: repeated 50-mg dosing was followed by acute neurologic injury including one brain death.
Type: harms-research.
Support: “Acute Neurologic Disorder from an Inhibitor of Fatty Acid Amide Hydrolase,” PMID 27806235. PubMed
Source: period primary × phase-I interventional/rct.
Population: healthy volunteers; six active in implicated cohort, detailed report on four consenting participants.
Product/route: BIA 10-2474 oral, 50 mg/day.
Outcome: severe neurologic syndrome.
Confidence: high for event; low for causal mechanism beyond study drug exposure.
Would change mind: investigation showing another cause unrelated to the drug.

CAND-17

Claim: Controlled abstinence studies in the early 2000s documented a reproducible cannabis-withdrawal symptom cluster with onset typically within days after cessation of heavy use.
Type: harms-research.
Support: Budney et al., PMID 12943018, DOI 10.1037/0021-843X.112.3.393; review PMID 15514394. PubMed
Source: period primary × observational/experimental abstinence; narrative review.
Confidence: high.
Would change mind: failed replication across well-characterized heavy users.

CAND-18

Claim: A 2007 systematic review of longitudinal population studies found cannabis exposure associated with later psychotic outcomes, with pooled adjusted OR 1.41 for any psychotic outcome among ever-users.
Type: harms-research.
Support: PMID 17662880, DOI 10.1016/S0140-6736(07)61162-3. PubMed
Source: period sr-ma × rests on observational longitudinal studies.
Confidence: high for association; not phrased as simple causation.
Would change mind: reanalysis eliminating association after adequate confounding control.

CAND-19

Claim: The Dunedin cohort reported an association between persistent cannabis use—especially adolescent-onset persistent use—and neuropsychological decline from childhood to age 38.
Type: harms-research.
Support: PMID 22927402, DOI 10.1073/pnas.1206820109. PubMed
Source: period primary × observational prospective cohort.
Population: 1,037 birth cohort.
Exposure: repeated cannabis use/dependence.
Outcome: longitudinal neuropsychological change.
Confidence: moderate because observational/confounding questions remain.
Would change mind: stronger reanalysis showing association explained by pre-existing/confounding factors.

CAND-20

Claim: Controlled driving research during P2 demonstrated measurable psychomotor/lateral-control effects after administered cannabis and evaluated interaction with alcohol.
Type: harms-research.
Support: “Cannabis effects on driving lateral control with and without alcohol,” PMID 26144593, DOI 10.1016/j.drugalcdep.2015.06.015. PubMed
Source: period primary × controlled interventional.
Confidence: high that controlled impairment research existed; candidate wording intentionally avoids translating simulator changes into a universal crash-risk magnitude.
Would change mind: study integrity issue.

CAND-21

Claim: The 2016 pregnancy systematic review/meta-analysis documented an observational literature linking prenatal cannabis exposure with maternal and neonatal outcomes but subject to exposure measurement and confounding limitations.
Type: harms-research.
Support: PMID 27048634, DOI 10.1136/bmjopen-2015-009986. PubMed
Source: period sr-ma × observational.
Confidence: high for literature characterization.
Would change mind: major methodological correction.

CAND-22

Claim: Period respiratory evidence focused primarily on smoked cannabis and associated frequent/regular smoking with bronchitic symptoms rather than establishing safety across non-smoked routes.
Type: harms-research.
Support: PMID 23802821; PMID 22497563, DOI 10.3109/15412555.2012.671868. PubMed
Source: period narrative review plus primary × observational cohort.
Confidence: moderate-high.
Would change mind: route-stratified studies showing this literature was not predominantly smoking-based.

CAND-23

Claim: CBD–antiseizure-medication interaction studies by 2015–17 identified substantially increased N-desmethylclobazam exposure and other serum AED changes during pharmaceutical CBD treatment.
Type: harms-research.
Support: PMID 26114620, DOI 10.1111/epi.13060; PMID 28782097. PubMed
Source: period primary × observational/interventional pharmacokinetic.
Confidence: high that interaction signal existed; magnitude is product/population-specific.
Would change mind: controlled PK work showing measurements were artefactual.

CAND-24

Claim: NASEM’s 2017 assessment rated evidence as conclusive/substantial for adult chronic pain, oral cannabinoids for CINV, and patient-reported MS spasticity, while assigning lower or insufficient categories to many other therapeutic endpoints.
Type: synthesis-conclusion.
Support: The Health Effects of Cannabis and Cannabinoids, DOI 10.17226/24625, PMID 28182367. NCBI
Source: period official evidence assessment × rests on systematic reviews/RCTs/observational evidence.
Confidence: high as statement of NASEM’s conclusion.
Would change mind: corrected edition.

CAND-25

Claim: NASEM’s 2017 assessment concluded that U.S. Schedule-I regulation, multiple review processes and constrained research-product supply impeded cannabis/cannabinoid research.
Type: programme.
Support: NASEM chapter 15, DOI 10.17226/24625; DEA 2016 policy statement Docket DEA-447. NCBI
Source: period official.
Confidence: high that these bodies documented the barriers; quantitative magnitude not established.
Would change mind: primary administrative evidence showing the processes described did not apply.

3. Era table and not-identified list

The requested era table is in §I-1 above, and the not-identified list is in §I-2. They are intentionally descriptive rather than a condition-specific effectiveness ranking.

4. Regulatory decisions table

Nabilone note: the FDA basis-of-approval record for NDA 18-677 is historical and largely pre-P2; I did not assign a fresh 2000–2017 approval event without a primary regulatory docket establishing one. FDA Access Data

5. Reference list

The list below includes the sources materially relied upon. “Verified live” means I opened a live PubMed/official result in this session and copied the title/identifier presented there. Access level describes this pass, not global availability.

1. Zajicek J et al. “Cannabinoids for treatment of spasticity and other symptoms related to multiple sclerosis (CAMS study): multicentre randomised placebo-controlled trial.” PMID 14615106; DOI 10.1016/S0140-6736(03)14738-1. Verified live (resolved title copied); abstract/extended PubMed record; period. PubMed
2. Abrams DI et al. “Cannabis in painful HIV-associated sensory neuropathy: a randomized placebo-controlled trial.” PMID 17296917; DOI 10.1212/01.wnl.0000253187.66183.9c. Verified live; abstract; period. PubMed
3. ClinicalTrials.gov NCT00046722. “Marijuana for HIV-Related Peripheral Neuropathy” / official-title record displayed. Verified live; registry record; evolving record, study start 2003; retrieved 2026-10-03. ClinicalTrials.gov
4. Ware MA et al. “Smoked cannabis for chronic neuropathic pain: a randomized controlled trial.” PMID 20805210; DOI 10.1503/cmaj.091414; ISRCTN68314063 reported by publication. Verified live PMID/DOI; ISRCTN not independently opened; abstract; period. PubMed
5. Novotna A et al. “A randomized, double-blind, placebo-controlled, parallel-group, enriched-design study of nabiximols (Sativex), as add-on therapy, in subjects with refractory spasticity caused by multiple sclerosis.” PMID 21362108; DOI 10.1111/j.1468-1331.2010.03328.x. Verified live; abstract; period. PubMed
6. ClinicalTrials.gov NCT00681538. Sativex MS-spasticity study. Verified live; evolving registry record; retrieved 2026-10-03. ClinicalTrials.gov
7. Blake DR et al. “Preliminary assessment of the efficacy, tolerability and safety of a cannabis-based medicine (Sativex) in the treatment of pain caused by rheumatoid arthritis.” PMID 16282192; DOI 10.1093/rheumatology/kei183. Verified live; abstract; period. PubMed
8. Devinsky O et al. “Trial of Cannabidiol for Drug-Resistant Seizures in the Dravet Syndrome.” PMID 28538134; DOI 10.1056/NEJMoa1611618; NCT02091375. Verified live title/PMID/DOI; registry ID verified through publication record; abstract; period. PubMed
9. Gloss D, Vickrey B. “Cannabinoids for epilepsy.” PMID 24595491; DOI 10.1002/14651858.CD009270.pub3. Verified live; full PMC availability indicated / abstract consulted; period. PubMed
10. Tramèr MR et al. “Cannabinoids for control of chemotherapy induced nausea and vomiting: quantitative systematic review.” PMID 11440936; DOI 10.1136/bmj.323.7303.16. Verified live; PubMed/PMC full-text availability; period, with many prior underlying trials. PubMed
11. Whiting PF et al. “Cannabinoids for Medical Use: A Systematic Review and Meta-analysis.” PMID 26103030; DOI 10.1001/jama.2015.6358. Verified live; abstract; period. PubMed
12. Koppel BS et al. “Systematic review: efficacy and safety of medical marijuana in selected neurologic disorders [RETIRED]: report of the Guideline Development Subcommittee of the American Academy of Neurology.” PMID 24778283; DOI 10.1212/WNL.0000000000000363. Verified live; PMC/abstract; period. “RETIRED” flag noted. PubMed
13. Moore THM et al. “Cannabis use and risk of psychotic or affective mental health outcomes: a systematic review.” PMID 17662880; DOI 10.1016/S0140-6736(07)61162-3. Verified live; abstract; period. PubMed
14. Budney AJ et al. “The time course and significance of cannabis withdrawal.” PMID 12943018; DOI 10.1037/0021-843X.112.3.393. Verified live; abstract; period. PubMed
15. Meier MH et al. “Persistent cannabis users show neuropsychological decline from childhood to midlife.” PMID 22927402; DOI 10.1073/pnas.1206820109. Verified live; full PubMed/PMC-linked record; period. PubMed
16. Hartman RL et al. “Cannabis effects on driving lateral control with and without alcohol.” PMID 26144593; DOI 10.1016/j.drugalcdep.2015.06.015. Verified live; abstract/record; period. PubMed
17. Gunn JKL et al. “Prenatal exposure to cannabis and maternal and child health outcomes: a systematic review and meta-analysis.” PMID 27048634; DOI 10.1136/bmjopen-2015-009986. Verified live; abstract/PMC; period. PubMed
18. Tashkin DP. “Effects of marijuana smoking on the lung.” PMID 23802821; DOI 10.1513/AnnalsATS.201212-127FR. Verified live; abstract; period. PubMed
19. Tashkin DP et al. “Impact of changes in regular use of marijuana and/or tobacco on chronic bronchitis.” PMID 22497563; DOI 10.3109/15412555.2012.671868. Verified live; abstract; period. PubMed
20. Mittleman MA et al. “Triggering myocardial infarction by marijuana.” PMID 11401936; DOI 10.1161/01.CIR.103.23.2805. Verified live; bibliographic/abstract-level result incomplete for effect estimate; period. PubMed
21. Geffrey AL et al. “Drug-drug interaction between clobazam and cannabidiol in children with refractory epilepsy.” PMID 26114620; DOI 10.1111/epi.13060. Verified live; abstract; period. PubMed
22. Gaston TE et al. “Interactions between cannabidiol and commonly used antiepileptic drugs.” PMID 28782097. Verified live; abstract; period. PubMed
23. Kerbrat A et al. “Acute Neurologic Disorder from an Inhibitor of Fatty Acid Amide Hydrolase.” PMID 27806235. Verified live title/source result; abstract; period. PubMed
24. van Esbroeck ACM et al. “Activity-based protein profiling reveals off-target proteins of the FAAH inhibitor BIA 10-2474.” PMID 28596366. Verified live title/result; abstract; period. PubMed
25. National Academies of Sciences, Engineering, and Medicine. The Health Effects of Cannabis and Cannabinoids: The Current State of Evidence and Recommendations for Research. DOI 10.17226/24625; PMID 28182367. Verified live; full NCBI Bookshelf; period. NCBI
26. Health Canada, Sativex NOC record. NOC 2005-04-15, control 091289, DIN 02266121. Verified live official record; evolving record; event 2005, retrieved 2026-10-03. Health Products Canada
27. Health Canada NOC/c list, Sativex. 2005 and 2007 conditional authorizations, control nos. 091289/109461. Verified live official record; evolving record; retrieved 2026-10-03. Canada
28. GOV.UK. “Scheduling of the cannabis-based medicine ‘Sativex’” / Circular 006/2013. Verified live official record; period regulatory history; retrieved 2026-10-03. GOV.UK
29. EMA. Acomplia EPAR, product no. EMEA/H/C/000666. Verified live official record; evolving record with 2006–09 milestones; retrieved 2026-10-03. European Medicines Agency (EMA)
30. EMA/CHMP. 2008 suspension recommendation, ref. EMEA/CHMP/537777/2008. Verified live official record; period event. European Medicines Agency (EMA)
31. EMA. Public Statement on Acomplia withdrawal, EMEA/39457/2009. Verified live official PDF result; period event. European Medicines Agency (EMA)
32. DEA. “Applications to Become Registered Under the Controlled Substances Act to Manufacture Marijuana to Supply Researchers in the United States,” Docket DEA-447, 2016. Verified live official Federal Register prepublication PDF; period. Federal Register Public Inspection
33. Yeung et al./bibliometric corpus source. “A bibliometric analysis of the cannabis and cannabinoid research literature.” PMID 35610633, DOI 10.1186/s42238-022-00133-0. Verified live; full PMC; retrospective. PubMed Central (PMC)
34. Matielo et al. “A Bibliometric Analysis of Cannabis Publications: Six Decades of Research and a Gap on Studies with the Plant.” DOI 10.3390/publications6040040. Verified live publisher page; full text; retrospective because published 2018. MDPI

6. Self-flagged weak points

1. Canonical PubMed counts are missing. The exact ESearch endpoints required by the brief returned inaccessible-tool errors. I will not estimate them from search-engine results. This is the largest explicit incompleteness in the pass.

2. Cochrane Library direct interface: I located and verified the relevant Cochrane epilepsy review through PubMed/PMC, but did not obtain a separate native Cochrane-site search-result census.

3. WHO ICTRP: direct WHO ICTRP interrogation was not successfully available through the browsing route. I searched ClinicalTrials.gov and web-indexed registry material. Consequently, “no registration identified” statements should be rechecked by the verifier against WHO ICTRP/EU registers.

4. ISRCTN68314063: the Ware publication itself reports this registration, but I did not obtain a live ISRCTN record. It is therefore not compliant with the strictest interpretation of rule 10a until the verifier opens it.

5. NCT02091375: identity was verified through the live PubMed paper, but I did not surface the standalone ClinicalTrials.gov historical record and therefore did not adjudicate registered-primary-outcome matching.

6. CAMS adverse events/dosing: the PubMed record supplied primary results and design but not enough details for complete AE/titration extraction. Those fields are marked not accessible rather than guessed.

7. Mittleman cardiovascular estimate: the identifier/title resolved, but the returned source display did not expose the study’s effect estimate. I deliberately did not reproduce a remembered “trigger” magnitude.

8. Sativex pivotal-trial linkage: the regulatory chronology is well documented, but I did not retrieve the original regulator assessment dossiers listing every pivotal trial for the 2005 Canadian and 2010 UK decisions. The regulatory table therefore names trial programs/examples rather than claiming an exhaustive regulator-defined pivotal set.

9. Cesamet/nabilone: an FDA basis-of-approval file is accessible, but the core approval evidence is substantially pre-2000. I found references to U.S. commercial/reapproval history in secondary material but did not identify a sufficiently clean P2 primary regulatory decision document to make it a candidate.

10. Bibliometric analyses are retrospective. They are used only for corpus shape/geography/source discovery, not as evidence of what clinicians or scientists knew by 2017.

11. Publication integrity: PubMed records were inspected for obvious correction/retraction flags where surfaced. Koppel 2014 is currently labeled [RETIRED] in PubMed; because it was a period professional-body conclusion, I preserve it as historical evidence but flag its current retirement status. I did not run a dedicated Retraction Watch/Crossref audit for every source; that belongs to verification under the protocol.

12. Boundary case: Blake et al. has e-publication 2005-11-09 and print 2006, both within P2, so no cross-part problem. No publication located here had an e-publication/print split across 1999/2000 or 2017/2018. The 2018 Devinsky dose-ranging publication is explicitly excluded from P2 despite describing work connected to the earlier program. PubMed

13. Later regulatory status pages: Health Canada/EMA/GOV.UK pages were retrieved in 2026 and are evolving records under rule 6b. I used them to verify dated historical events; post-2017 indication changes were not treated as evidence of what was known in 2017.

14. This was not a systematic review. Search coverage was broad but selective and evidence-map oriented, as the brief requires.

7. Search log

Run date: 2026-10-03.

PubMed/MEDLINE

Core exploratory searches included combinations of:

cannabis cannabinoids 2000 2017 clinical trial randomized

cannabinoids multiple sclerosis randomized CAMS

smoked cannabis HIV neuropathy randomized Abrams

smoked cannabis neuropathic pain Ware

nabiximols Sativex multiple sclerosis spasticity Novotna

cannabidiol Dravet randomized Devinsky

cannabinoids systematic review Whiting 2015

cannabinoids epilepsy Cochrane Gloss Vickrey

cannabinoids chemotherapy nausea vomiting Tramer

cannabis psychosis systematic review Moore 2007

cannabis withdrawal Budney

persistent cannabis neuropsychological decline Meier

cannabis driving Hartman Huestis

prenatal cannabis systematic review Gunn

marijuana bronchitis Tashkin

marijuana myocardial infarction Mittleman

cannabidiol clobazam interaction Geffrey

FAAH inhibitor BIA 10-2474

Exact live result counts for these exploratory web/PubMed searches were not consistently exposed by the search interface and therefore are not reported.

Canonical count searches

Exact canonical expression from brief:

("Cannabis"[Mesh] OR "Cannabinoids"[Mesh] OR "cannabis"[tiab] OR "marijuana"[tiab] OR "marihuana"[tiab] OR "cannabinoid"[tiab] OR "cannabinoids"[tiab])

Date windows attempted:

• 2000/01/01–2004/12/31
• 2005/01/01–2009/12/31
• 2010/01/01–2014/12/31
• 2015/01/01–2017/12/31

Result: NCBI ESearch URL inaccessible to the browsing tool in all four attempts; counts and QueryTranslation unavailable. The exact links are supplied in §A.

ClinicalTrials.gov

Searches included titles/authors/interventions for:

• Abrams / HIV peripheral neuropathy / marijuana → NCT00046722, verified.
• Novotna / Sativex / MS spasticity → NCT00681538, verified.
• Devinsky / cannabidiol / Dravet → NCT02091375 identified from the paper; standalone record not surfaced in this pass.
• Ware / smoked cannabis neuropathic pain → publication reports ISRCTN68314063; no ClinicalTrials.gov registration identified.
• General searches of nabiximols/cannabis neuropathic pain also surfaced later/out-of-window studies, which were excluded where publication dates fell after 2017.

WHO ICTRP / EU registers

Web search attempted for registry identifiers and named-product trials. Direct WHO ICTRP access not obtained. EU/EudraCT searches did not produce sufficiently reliable direct records for inclusion. This is a declared access gap.

Cochrane

Search concepts:

cannabinoids epilepsy Cochrane

cannabis cannabinoids multiple sclerosis Cochrane

cannabinoids chronic pain Cochrane

The 2014 Cannabinoids for epilepsy review was verified via PubMed/PMC. Direct Cochrane-platform result counts were not obtained.

Regulators / official sources

Health Canada:
Sativex Notice of Compliance conditions 2005 neuropathic pain multiple sclerosis
Sativex 2007 cancer pain NOC/c

Identified official records for control nos. 091289 and 109461. Canada

UK/MHRA/Home Office:
MHRA Sativex 2010 multiple sclerosis spasticity marketing authorisation
Official GOV.UK scheduling material located. GOV.UK

EMA:
Acomplia rimonabant 2006 withdrawal 2008
Official EPAR, suspension communication and withdrawal statement located. European Medicines Agency (EMA)

FDA:
Cesamet nabilone approval chemotherapy nausea FDA
Historical NDA medical review located; no clean new P2 approval milestone assigned. FDA Access Data

DEA/Federal Register:
marijuana research additional manufacturers 2016 DEA
Docket DEA-447 located. Federal Register Public Inspection

Government/international evidence assessments

Searches for:

National Academies cannabis cannabinoids 2017 therapeutic effects

National Academies cannabis research barriers Schedule I NIDA

National Academies cannabis driving pregnancy psychosis

Full NCBI Bookshelf chapters retrieved for summary, pregnancy, injury/death and research barriers. NCBI

Bibliometrics

Searches:

cannabis cannabinoid research bibliometric countries publications

bibliometric cannabis publications 2017

Located 2018 and 2022 retrospective analyses. PubMed Central (PMC)

Access failures / exclusions

• NCBI ESearch API: canonical-count endpoints inaccessible.
• Direct WHO ICTRP: not accessible through available interface.
• Some journal full texts/paywalled supplements: inaccessible; I used gap codes rather than inferring missing extraction fields.
• Some historical regulatory assessment dossiers were not surfaced in searchable text.
• Publications first publicly published 2018 onward were excluded as P2 evidence; where mentioned at all, they were used only as retrospective/context sources and labeled accordingly.
• No project Drive, previous raw pass, P1 evidence file, archive material or probe document was opened.

Training-knowledge reliance

I used prior general knowledge only to formulate exploratory search terms. No candidate claim or specific identifier in this submission is intentionally supported solely from model memory. Where a remembered-looking detail could not be live-verified—most notably the canonical counts and some registry details—I omitted it or flagged it unverified.

Stage 1 ends here. I have not answered, reconstructed, or anticipated any Stage 2 probe.