The therapy, known as foslevodopa/foscarbidopa or LDp/CDp, is a soluble prodrug formulation of levodopa and carbidopa delivered through a small pump via continuous subcutaneous infusion. Unlike oral tablets, whose absorption in the gut is erratic and easily disrupted by meals and gastrointestinal slowdown, the infusion maintains a steady level of dopamine stimulation throughout the day and night. For patients with advanced disease whose brains can no longer buffer the peaks and troughs of oral medication, that steadiness translates into smoother motor control. But the same continuous dopaminergic stimulation can, in susceptible patients, tip the mesolimbic system toward hallucinations, psychosis, and compulsive behaviors.

Because pivotal trials excluded patients with significant cognitive impairment or severe psychiatric comorbidity, clinicians have lacked solid data on who is truly at risk. To fill that gap, a team of Italian researchers from two tertiary referral centers, the Parkinson Institute of Milan and the IRCCS Mondino Foundation in Pavia, retrospectively analyzed 77 consecutive patients with advanced Parkinson’s disease who had started LDp/CDp infusion, each followed for at least six months. The results, published in the Journal of Neurology, offer one of the clearest portraits yet of the therapy’s neuropsychiatric safety in everyday clinical care.

The cohort was typical of advanced Parkinson’s patients seen in specialist centers: a mean disease duration of nearly 14 years, moderate-to-severe motor burden on the MDS-UPDRS scale, and a substantial load of pre-existing vulnerabilities. Almost half of the patients met criteria for cognitive decline, with mild cognitive impairment in 41.6 percent and Parkinson’s disease dementia in 7.8 percent. A quarter had a history of hallucinations or psychosis, and roughly 27 percent had experienced impulse control disorders, the gambling, shopping, and hypersexual behaviors that can emerge with dopaminergic drugs. Notably, eight patients were already experiencing active but well-controlled neuropsychiatric symptoms when they began the infusion.

Over six months of treatment, the therapy delivered on its motor promise. Total levodopa equivalent daily dose rose by roughly 300 milligrams, and motor scores improved significantly: MDS-UPDRS III fell from 34.4 to 27.0, while Part IV scores, which capture fluctuations and dyskinesia, dropped from 8.3 to 5.25. At the same time, clinicians simplified the medication regimen, tapering dopamine agonists, COMT inhibitors, and MAO-B inhibitors, a move supported by recent analyses suggesting that infusion monotherapy can be as effective as polytherapy.

The psychiatric picture was more nuanced. New-onset neuropsychiatric adverse events emerged in 17 patients, or 22.1 percent of the cohort, at a median of 30 days after starting the pump. Hallucinations, mostly distressing visual phenomena occurring at night, accounted for about 41 percent of these events, followed by psychosis or delusional ideas at nearly 30 percent. Isolated cases of punding, aggression, and acute psychomotor agitation also appeared. This rate is higher than the 15 to 17 percent reported in the pivotal phase III trials, but the researchers attribute that gap to their deliberately unselected population, which included many patients that regulatory studies would have screened out.

Crucially, the story does not end with the adverse events themselves. Most episodes were managed successfully through straightforward measures: lowering the infusion rate, particularly at night, reducing or withdrawing dopamine agonists, and, when needed, adding low-dose atypical antipsychotics such as quetiapine or clozapine. Overall, 82 percent of events resolved or improved. Only three patients, just under 18 percent of those affected, had to abandon the therapy because of severe psychiatric complications. Across the entire cohort, the six-month dropout rate was 16.9 percent, with adverse events, bridging to deep brain stimulation, and device intolerance listed as the leading causes. One patient also developed a length-dependent lower-limb polyneuropathy detected on nerve conduction studies, prompting discontinuation as a precaution.

Perhaps the most clinically valuable finding came from the search for risk factors. Among all demographic, pharmacological, and cognitive variables tested, the baseline score on the Frontal Assessment Battery, or FAB, a short bedside test of executive function, stood out as the strongest independent predictor of neuropsychiatric adverse events, with lower scores roughly doubling the risk in the multivariable models. Daytime and nighttime infusion rates and total daily dose also contributed to risk, but FAB retained its predictive power across every regression model tested, outperforming the widely used MMSE global cognitive screen. Kaplan-Meier curves confirmed that patients below accepted FAB and MoCA thresholds accumulated new-onset events far faster than cognitively preserved peers.

The findings carry practical weight for dosing strategy. The Italian team maintained deliberately conservative flow rates, averaging 0.33 milliliters per hour during the day and 0.20 at night, which is well below the aggressive rates of roughly 0.60 milliliters per hour reported in smaller case series where severe neuropsychiatric complications were frequent. The authors suggest that continuous round-the-clock stimulation may promote mesolimbic hyperactivation and receptor desensitization, disrupting sleep architecture and precipitating nocturnal hallucinations in vulnerable brains. Reducing or pausing the nighttime infusion proved to be an effective first-line intervention, and the team also advocates inpatient initiation, slow titration, and preventive low-dose antipsychotics or acetylcholinesterase inhibitors for patients deemed high risk.

Encouragingly, patients who entered treatment with active but controlled hallucinations or impulse control disorders generally fared well, and the shift away from oral dopamine agonists appeared to ease compulsive behaviors, echoing the favorable pattern long reported with intestinal levodopa gel. The overall message is one of reassurance tempered by vigilance: LDp/CDp infusion is generally well tolerated even in a frail, cognitively mixed population, provided that frontal executive function is formally screened before treatment, titration is gradual, nighttime doses are kept low, and patients are monitored closely during the first weeks, when most events declare themselves. The researchers caution that their study was retrospective, uncontrolled, and limited to two expert centers, and they call for prospective multicenter trials with standardized infusion protocols to confirm the findings. For now, the results give clinicians a concrete roadmap for making this increasingly popular therapy safer for the patients who need it most.

Subject of Research: Real-world neuropsychiatric safety of continuous subcutaneous foslevodopa/foscarbidopa infusion in advanced Parkinson's disease

Article Title: Real-world neuropsychiatric safety profile of foslevodopa/foscarbidopa infusion in advanced Parkinson's disease

Article References: Real-world neuropsychiatric safety profile of foslevodopa/foscarbidopa infusion in advanced Parkinson's disease. (n.d.). https://doi.org/10.1007/s00415-026-14113-4

Image Credits: AI Generated

DOI: 10.1007/s00415-026-14113-4

Keywords: Parkinson's disease, foslevodopa/foscarbidopa, subcutaneous levodopa infusion, neuropsychiatric adverse events, hallucinations, psychosis, impulse control disorders, frontal executive function, Frontal Assessment Battery, device-aided therapy, continuous dopaminergic stimulation, real-world evidence