Lorlatinib and avapritinib can affect memory, thinking and mood; lorlatinib can also change speech or cause hallucinations. These effects usually improve when the dose is paused or lowered, so tell your team about new forgetfulness, confusion, mood changes, thoughts of suicide, or seeing or hearing things that are not real. The EU lorlatinib leaflet says immediately. Family may notice first.
Lorlatinib: in the US label, central nervous system effects occurred in 52% of 476 patients (cognitive 28%, mood 21%, sleep 12%, speech 11%, psychotic 7%, seizures 1.9%), with a median time to first onset of 1.4 months; 2.1% stopped permanently for one, 10% paused and 8% had a dose reduction. The EU SmPC gives cognitive effects in 27.4%, mood 21.4%, speech 8.2% and psychotic effects 6.9%, and calls them generally mild, transient and reversible on dose delay or reduction. The reader's own team's protocol comes first.
Lorlatinib dose rules. US label: grade 1, continue at the same dose or withhold until recovery to baseline, then the same or a reduced dose; grade 2 or 3, withhold until grade 0 or 1 and resume at a reduced dose; grade 4, permanently discontinue. The EU SmPC has the same grade 2 or 3 and grade 4 rules and no grade 1 row; its definition does not include seizures or sleep. Dose levels are 75 mg and then 50 mg once daily.
Avapritinib: cognitive adverse reactions occurred in 33% of 995 patients in the US label (41% of 601 with GIST). The US and EU grade rules match, though the US table prints them under GIST or advanced systemic mastocytosis and the EU table under all three indications: grade 1, continue at the same or a reduced dose, or pause until baseline or resolution; grade 2 or 3, pause until baseline, grade 1 or resolution, then the same or a reduced dose; grade 4, permanently discontinue. The EU SmPC adds that people with GIST or advanced systemic mastocytosis are monitored clinically and must notify their healthcare professional immediately. A change in mental state on avapritinib can also be a sign of bleeding in the brain, for which both labels stop the medicine for good.
Beyond the labels, the advice for lorlatinib (preparing the person and family, looking for other causes, how early to interrupt, when to refer) comes from expert papers funded or supported by Pfizer, the maker of lorlatinib, or (the 2019 paper) written with Pfizer employees and Pfizer-funded medical writing. They differ on how early to interrupt: the Italian panel says as soon as effects emerge; the 2019 paper says to consider interrupting at grade 1; a Latin American panel says to continue at the current dose at grade 1; in the CROWN trial 58% of events needed, per protocol, no medical intervention.
Why: The 2019 paper says these effects are temporary and reversible when the dose is interrupted or reduced. The CROWN paper adds that clear examples matter (people have described sluggish thought, fogginess, feeling flat or more irritable) because the term CNS adverse effects will be insufficient for most. The 2019 paper was written with Pfizer employees and its medical writing was funded by Pfizer; the CROWN study was sponsored by Pfizer.
“Effects on mood should be discussed with patients before lorlatinib is administered. It is particularly important to review these effects with patients who have pre‐existing psychiatric conditions.”
Why: The Spanish panel (funded by Pfizer) says many of these events can also have other causes, among them existing psychiatric disease, and that it is therefore important to set out how to tell them apart both before and after treatment begins. It adds that the psychiatric symptoms of people already on psychiatric treatment should be closely monitored, and that lorlatinib reduces the effect of some antipsychotics.
“The patient’s psychiatric history should be explored, and these possible events should be discussed with the patient before initiating treatment with lorlatinib.”
Why: The paper's reason is that the person may be unaware of the change. The CROWN paper (sponsored by Pfizer) says care partners or family members will often notice these effects before the patient does, particularly when symptoms are subtle.
“Caregivers and family members, as well as patients, should also be encouraged to report any changes in patients’ mood during treatment with lorlatinib in the event that patients are unaware of such changes.”
Why: The US label says these effects can be severe and that the dose is withheld, reduced or stopped according to severity. The symptom list is the one in its patient information. The EU leaflet uses stronger words: tell your doctor immediately about speech problems, mental status changes, and mood or memory problems.
“Advise patients to notify their healthcare provider if they experience new or worsening CNS symptoms”
Why: The Spanish panel (funded by Pfizer) says many of the neurological events seen on lorlatinib can also be caused by these, and recommends reviewing new medicines and the person's psychiatric state. No source read says which tests or scans to do.
“It is important to keep in mind that many of the neurological AEs detected in patients receiving lorlatinib can also be caused by other factors, including brain metastases, brain radiation, brain surgery, preexisting psychiatric disease, and other neurotropic medications, among others”
Why: This is the US label's row; it gives no reason for offering both courses. The EU SmPC has no grade 1 row. The expert papers pull both ways. The 2019 paper (Pfizer employees among its authors) says dose interruptions should be considered at grade 1, and the Italian panel (funded by Pfizer) goes further and says interruption should be carried out as soon as these effects emerge. The Latin American panel (supported by Pfizer) says treatment should continue at the current dose at grade 1.
“Continue at the same dose or withhold the dose until recovery to baseline. Resume LORBRENA at the same dose or at a reduced dose.”
Why: Both labels give this rule (the EU SmPC: withhold until grade 1 or less, then resume at one reduced dose level) and neither states a reason beside it. The EU SmPC describes these reactions as generally reversible on dose delay or reduction. The dose levels are 75 mg and then 50 mg once daily.
“Withhold dose until Grade 0 or 1. Resume LORBRENA at a reduced dose.”
Why: Both labels say to permanently discontinue at grade 4, which the EU SmPC labels life-threatening or needing urgent intervention. Neither gives a further reason.
“Permanently discontinue lorlatinib.”
European Medicines Agency (marketing authorisation holder Pfizer) 2026
Why: The Italian panel (funded by Pfizer; two authors Pfizer employees) gives this as its advice and says people with known risk factors (brain metastases, brain radiation, psychiatric illness, medicines acting on the brain) should be strictly monitored, without a figure in that passage. The CROWN paper (sponsored by Pfizer) puts referral more weakly: visiting a specialist may be helpful to manage other causes. On risk it says there are no clear predictive factors: these effects occurred in 67% of patients with prior brain radiotherapy and 41% of those without, with patient numbers too small to draw definitive conclusions.
“If lorlatinib dose modification does not result in CNS symptom improvement, patients should be referred to psychiatrists or neurologists”
Why: The CROWN paper (sponsored by Pfizer) says these strategies may be useful, citing an earlier management guide that was not read for this page; it reports no trial of them. The 2019 paper also suggests advice such as setting reminders.
“Nonpharmacological management strategies to minimize the impact of CNS AEs, such as setting reminders, mindfulness, meditation, and cognitive behavioral therapy, may be useful.”
Why: The EU SmPC says lorlatinib has a moderate influence on the ability to drive because of these effects. The US avapritinib label tells prescribers to advise patients not to drive or operate hazardous machinery if they are experiencing cognitive adverse reactions; the EU avapritinib SmPC says such patients must take special care.
“Caution should be exercised when driving or operating machines as patients may experience CNS effects”
European Medicines Agency (marketing authorisation holder Pfizer) 2026
Why: The EU SmPC says the mechanism is not known, and that in clinical studies dose reductions or interruptions improved grade 2 or higher cognitive effects compared with no action. It says these patients must notify their healthcare professional immediately about new or worsening symptoms. The US label says to advise patients and caretakers to notify the healthcare provider, without the word immediately.
“It is recommended that patients with GIST or AdvSM are clinically monitored for signs and symptoms of cognitive events such as new or increased forgetfulness, confusion, and/or difficulty with cognitive functioning.”
Why: This is the US label's row, printed under the heading for patients with GIST or advanced systemic mastocytosis. The EU SmPC's row says the same and its heading also covers indolent systemic mastocytosis. The US label says cognitive adverse reactions were managed with dose interruption or reduction when needed; it gives no reason for choosing between the courses.
“Continue AYVAKIT at same dose or reduced dose or withhold until improvement to baseline or resolution. Resume at same dose or reduced dose.”
Blueprint Medicines, on DailyMed (US National Library of Medicine) 2026
Why: Both labels give this rule. The EU SmPC notes that grade 2 or higher means an effect on activities of daily living. How long improvement takes differs by label: among people with a grade 2 or worse effect, the US label gives a median time to improvement to grade 1 or resolution of 7.9 weeks in GIST and 8.1 weeks in advanced systemic mastocytosis; the EU SmPC gives a median time to improvement of 15 weeks in GIST and 9 weeks in advanced systemic mastocytosis.
“Withhold AYVAKIT until improvement to baseline, Grade 1, or resolution. Resume at same dose or reduced dose.”
Blueprint Medicines, on DailyMed (US National Library of Medicine) 2026
Why: Both labels say to permanently discontinue at grade 4 (the EU row also covers indolent systemic mastocytosis). Neither gives a further reason.
“Permanently discontinue AYVAKIT.”
Blueprint Medicines, on DailyMed (US National Library of Medicine) 2026
Why: The US label lists altered mental status among the symptoms of intracranial haemorrhage, which occurred in 2.9% of 749 patients with GIST or advanced systemic mastocytosis in its trials, and says to stop avapritinib permanently if it occurs at any grade. The EU SmPC has the same stopping rule, and says people with severe headache, vision problems, somnolence or focal weakness must interrupt dosing and inform their healthcare professional immediately.
“Advise patients to seek immediate medical attention for signs or symptoms of intracranial hemorrhage.”
Blueprint Medicines, on DailyMed (US National Library of Medicine) 2026
How care is usually carried out, read from the guidelines and standards cited. Your own team's protocol and their judgement come first; this page is for understanding it, not for doing it yourself.
No guideline with graded recommendations covers these effects. The firmest rules are the labels' dose tables, which the US and EU labels share for grade 2 to 4: for lorlatinib, withhold at grade 2 or 3 until grade 1 or less and resume at a reduced dose, and stop for good at grade 4; for avapritinib, withhold at grade 2 or 3 until baseline, grade 1 or resolution and resume at the same or a reduced dose, and stop for good at grade 4 (rows the US label prints for GIST and advanced systemic mastocytosis, and the EU SmPC for those and indolent systemic mastocytosis). They differ at grade 1 for lorlatinib (a US row allowing either continuing or withholding; no EU row). Everything else for lorlatinib comes from expert papers funded or supported by its maker, Pfizer, or written with its employees, and they pull in two directions on timing: the Italian panel says to interrupt as soon as these effects emerge, even if not particularly bothersome, and the 2019 paper says interruption should be considered at grade 1, while the Latin American panel says to continue at the current dose at grade 1, the CROWN paper reports that 58% of events in the trial needed, per protocol, no medical intervention, and the US label allows continuing at grade 1. Both are shown; this page does not choose. For avapritinib only the labels were found.
Pfizer, on DailyMed (US National Library of Medicine) 2026 · European Medicines Agency (marketing authorisation holder Pfizer) 2026 · Blueprint Medicines, on DailyMed (US National Library of Medicine) 2026 · European Medicines Agency 2026 · The Oncologist (Bauer et al.; phase I investigators with three Pfizer employees among the authors; medical writing funded by Pfizer Inc.) 2019 · Clinical Drug Investigation (Arriola et al.; a Spanish multidisciplinary panel; project and medical writing funded by Pfizer) 2024 · The Oncologist (Liu et al.; study sponsored by Pfizer; first draft prepared with Pfizer's AI writing tool; medical writing funded by Pfizer) 2025 · Future Oncology (Ardizzoni et al.; funded by Pfizer; two authors Pfizer employees, the other four paid consultants to Pfizer for the manuscript) 2026 · Frontiers in Oncology (Corrales et al.; a Latin American expert panel; the consensus was supported by Pfizer, which the paper says had no involvement in its content or conclusions) 2026
Where it differs: Scope: the US label counts seizures, psychotic effects and changes in cognitive function, mood (including suicidal ideation), speech, mental status and sleep; the EU SmPC's dose row covers psychotic effects and changes in cognition, mood, mental status or speech. Grade 1: the US label has a row (continue, or withhold until baseline, then the same or a reduced dose); the EU SmPC has none. Grade 2 or 3 and grade 4: the same in both. What patients are told: the US patient information says to tell the healthcare provider; the EU leaflet says to tell the doctor immediately.
Evidence: Regulatory labels. US label: 52% of 476 patients had a CNS effect; 2.1% stopped permanently, 10% paused, 8% had a dose reduction. EU SmPC: median time to onset 129 days for cognitive, 57 for mood, 58 for speech and 27 for psychotic effects.
Pfizer, on DailyMed (US National Library of Medicine) 2026 · European Medicines Agency (marketing authorisation holder Pfizer) 2026
Where it differs: Italian panel 2026: unlike other side effects, temporary interruption should be carried out as soon as these effects emerge, even if not particularly bothersome. Bauer 2019: interruptions should be considered at grade 1, and dose changes imposed at low thresholds are generally effective. US label: at grade 1, continue or withhold. Latin American panel 2026: at grade 1, treatment should continue at the current dose, and if this is not possible a lower dose should be considered. Spanish panel 2024: its recommendations, which include supportive treatments, should be used as soon as possible in an attempt to avoid reducing the dose. CROWN paper 2025: more than half (58%) of events in the trial did not, per protocol, require any medical intervention; it also says early interruption can lead to rapid reversal.
Evidence: Expert opinion and trial description; every paper was funded, sponsored or supported by Pfizer or (the 2019 paper) written with Pfizer employees. In the 2019 pooled analysis, 24 of 117 patients with these effects had a dose change and 15 of the 24 had resolution, with a median time to resolution of 12.5 days; in CROWN at 5 years 60% of events resolved overall.
Future Oncology (Ardizzoni et al.; funded by Pfizer; two authors Pfizer employees, the other four paid consultants to Pfizer for the manuscript) 2026 · The Oncologist (Bauer et al.; phase I investigators with three Pfizer employees among the authors; medical writing funded by Pfizer Inc.) 2019 · Pfizer, on DailyMed (US National Library of Medicine) 2026 · Frontiers in Oncology (Corrales et al.; a Latin American expert panel; the consensus was supported by Pfizer, which the paper says had no involvement in its content or conclusions) 2026 · Clinical Drug Investigation (Arriola et al.; a Spanish multidisciplinary panel; project and medical writing funded by Pfizer) 2024 · The Oncologist (Liu et al.; study sponsored by Pfizer; first draft prepared with Pfizer's AI writing tool; medical writing funded by Pfizer) 2025
Where it differs: The grade rules are the same; the US label says withhold where the EU SmPC says interrupt. Scope differs: the US table prints the cognitive rows under patients with GIST or advanced systemic mastocytosis, the EU table under GIST, advanced or indolent systemic mastocytosis. Dose levels: both give 300, then 200, then 100 mg once daily for GIST and 200, then 100, 50 and 25 mg once daily for advanced systemic mastocytosis; only the EU SmPC gives a reduction for indolent systemic mastocytosis (25 mg once daily to 25 mg every other day). The EU SmPC adds a footnote that grade 2 or higher means an impact on activities of daily living, asks for clinical monitoring of people with GIST or advanced systemic mastocytosis, and says they must notify their healthcare professional immediately; for indolent systemic mastocytosis it notes that cognitive effects can be a symptom of the disease itself and says to notify the healthcare professional, without the word immediately. On driving, the US label says not to drive while cognitive reactions are present; the EU SmPC says to take special care.
Evidence: Regulatory labels. US label: cognitive adverse reactions in 33% of 995 patients across indications; in GIST 41% of 601, with 2.7% stopping permanently, 13.5% pausing and 8.5% reducing the dose. EU SmPC: 33% of 585 patients with GIST. Median time to improvement of a grade 2 or worse effect: 7.9 weeks (GIST) and 8.1 weeks (advanced systemic mastocytosis) in the US label, 15 and 9 weeks in the EU SmPC.
Blueprint Medicines, on DailyMed (US National Library of Medicine) 2026 · European Medicines Agency 2026
Where it differs: The Spanish panel names choices by their interaction with lorlatinib: short or medium half-life benzodiazepines could be useful for anxiety, but antidepressants with an anxiolytic profile are advisable because benzodiazepines have a negative effect on cognition; the antidepressants of choice are duloxetine and agomelatine, with caution for several others; the effect of some antipsychotics will be reduced; and olanzapine is recommended for psychosis, mania or hallucinosis, because quetiapine and ziprasidone can interact with lorlatinib. The CROWN paper reports only that the medicines most often used in the trial were anxiolytics, antidepressants and antiepileptics. The labels name no medicine for these effects.
Evidence: Expert opinion reasoned from drug interactions (the panel points to its interaction table), in a Pfizer-funded paper; the passage reports no trial result for any of these medicines.
Clinical Drug Investigation (Arriola et al.; a Spanish multidisciplinary panel; project and medical writing funded by Pfizer) 2024 · The Oncologist (Liu et al.; study sponsored by Pfizer; first draft prepared with Pfizer's AI writing tool; medical writing funded by Pfizer) 2025
| Recommendation | Grade, as printed | Source |
|---|---|---|
| Lorlatinib, US label: the range of CNS effects “These include seizures, psychotic effects and changes in cognitive function, mood (including suicidal ideation), speech, mental status, and sleep.” | Regulatory label; not graded | Pfizer, on DailyMed (US National Library of Medicine) 2026 |
| Lorlatinib, US label: grade 4 is permanent discontinuation “Permanently discontinue LORBRENA.” | Regulatory label; not graded | Pfizer, on DailyMed (US National Library of Medicine) 2026 |
| Lorlatinib, EU SmPC: generally mild, transient and reversible on dose delay or reduction “were generally mild, transient, and reversible spontaneously upon dose delay and/or dose reduction” | Regulatory label; not graded | European Medicines Agency (marketing authorisation holder Pfizer) 2026 |
| Lorlatinib, EU SmPC: dose modification or discontinuation may be required “Dose modification or discontinuation may be required for those patients who develop CNS effects” | Regulatory label; not graded | European Medicines Agency (marketing authorisation holder Pfizer) 2026 |
| 2019 paper: consider interruption at grade 1 “Dose interruptions should be considered in the presence of grade 1 CNS effects and enforced until recovery to baseline before restarting lorlatinib at the same dose or, if required, at a lower dose.” | Expert consensus opinion; not graded (Pfizer employees among the authors; medical writing funded by Pfizer) | The Oncologist (Bauer et al.; phase I investigators with three Pfizer employees among the authors; medical writing funded by Pfizer Inc.) 2019 |
| 2019 paper: patients and caregivers should be instructed to report changes “patients and their caregivers should be instructed to report any changes in cognitive function, mood, or speech to their health care provider” | Expert consensus opinion; not graded (Pfizer employees among the authors; medical writing funded by Pfizer) | The Oncologist (Bauer et al.; phase I investigators with three Pfizer employees among the authors; medical writing funded by Pfizer Inc.) 2019 |
| Italian panel: interrupt as soon as CNS effects emerge “Unlike other AEs, lorlatinib temporary interruption should be carried out as soon as CNS AEs emerge, even if not particularly bothersome.” | Expert panel; not graded (funded by Pfizer; two authors Pfizer employees) | Future Oncology (Ardizzoni et al.; funded by Pfizer; two authors Pfizer employees, the other four paid consultants to Pfizer for the manuscript) 2026 |
| Latin American panel: continue at the current dose at grade 1 “In cases of grade 1 toxicity, treatment with lorlatinib should continue at the current dose. If this is not possible, a lower dose should be considered depending on the patient’s profile.” | Expert recommendations; not graded (supported by Pfizer) | Frontiers in Oncology (Corrales et al.; a Latin American expert panel; the consensus was supported by Pfizer, which the paper says had no involvement in its content or conclusions) 2026 |
| CROWN paper: no clear predictive factors “There are no clear predictive factors for CNS AEs.” | Trial report with expert commentary; not graded (sponsored by Pfizer) | The Oncologist (Liu et al.; study sponsored by Pfizer; first draft prepared with Pfizer's AI writing tool; medical writing funded by Pfizer) 2025 |
| CROWN paper: more than half of CNS events needed no medical intervention “The vast majority of CNS AEs were grade 1 or 2, and more than half (58%) of CNS AEs, per protocol, did not require any medical intervention.” | Trial report with expert commentary; not graded (sponsored by Pfizer) | The Oncologist (Liu et al.; study sponsored by Pfizer; first draft prepared with Pfizer's AI writing tool; medical writing funded by Pfizer) 2025 |
| CROWN paper: a specialist visit may be helpful “To potentially manage other causes of neurocognitive and psychiatric impairment, visiting a specialist (eg, neurologist, psychiatrist) may be helpful.” | Trial report with expert commentary; not graded (sponsored by Pfizer) | The Oncologist (Liu et al.; study sponsored by Pfizer; first draft prepared with Pfizer's AI writing tool; medical writing funded by Pfizer) 2025 |
| Spanish panel: closely monitor the psychiatric symptoms of people already on psychiatric treatment “If patients with psychiatric treatment are starting on lorlatinib, their psychiatric symptoms should be closely monitored.” | Expert consensus; not graded (funded by Pfizer) | Clinical Drug Investigation (Arriola et al.; a Spanish multidisciplinary panel; project and medical writing funded by Pfizer) 2024 |
| Spanish panel: antidepressants with an anxiolytic profile are advisable “it is advisable to use antidepressants with an anxiolytic profile, especially because benzodiazepines have a negative effect on cognition” | Expert consensus; not graded (funded by Pfizer) | Clinical Drug Investigation (Arriola et al.; a Spanish multidisciplinary panel; project and medical writing funded by Pfizer) 2024 |
| Avapritinib, EU SmPC: immediate notification for GIST and advanced systemic mastocytosis “Patients with GIST or AdvSM must notify their healthcare professional immediately if they experience new or worsening cognitive symptoms.” | Regulatory label; not graded | European Medicines Agency 2026 |
| Avapritinib, EU SmPC: indolent systemic mastocytosis, notify without the word immediately “Patients with ISM must notify their healthcare professional if they experience new or worsening cognitive symptoms.” | Regulatory label; not graded | European Medicines Agency 2026 |
| Avapritinib, EU SmPC: dose reductions or interruptions improved grade 2 or higher effects “dose reductions or interruptions improved Grade ≥2 cognitive effects compared to no action.” | Regulatory label; not graded | European Medicines Agency 2026 |
| Avapritinib, US label: not driving while cognitive reactions are present “Advise patients not to drive or operate hazardous machinery if they are experiencing cognitive adverse reactions” | Regulatory label; not graded | Blueprint Medicines, on DailyMed (US National Library of Medicine) 2026 |
| Avapritinib, US label: altered mental status among the symptoms of intracranial haemorrhage “Symptoms of intracranial hemorrhage may include headache, nausea, vomiting, vision changes, or altered mental status.” | Regulatory label; not graded | Blueprint Medicines, on DailyMed (US National Library of Medicine) 2026 |
| Avapritinib, US label: stop permanently for intracranial haemorrhage of any grade “Permanently discontinue AYVAKIT if intracranial hemorrhage of any grade occurs” | Regulatory label; not graded | Blueprint Medicines, on DailyMed (US National Library of Medicine) 2026 |
Shares CTCAE toxicity grading (grade 3-4 adverse events), Dose reduction, interruption and discontinuation.
Shares Fluid retention, swelling and pleural effusion on kinase inhibitors and MET-targeted drugs.
Shares CTCAE toxicity grading (grade 3-4 adverse events), Fluid retention, swelling and pleural effusion on kinase inhibitors and MET-targeted drugs.
Shares Drug, herbal and food interaction review for anticancer treatment, Dose reduction, interruption and discontinuation.