Registration trials

The development of injectable risperidone included placebo-controlled trials in which risperidone LAI was compared with oral risperidone, as well as studies in which patients were switched from FGA–LAIs or from oral antipsychotics to risperidone LAI, generally with an observational period of 12 weeks. In addition to these studies, a 1-year open clinical trial, post hoc analyses and prescription audits have been carried out.

In the phase III trials fortnightly injections of risperidone LAI (25 mg, 50 mg and 75 mg) were evaluated. The 75 mg dose did not confer any advantage in efficacy over the 50 mg dose but did have a higher side-effect burden with extrapyramidal symptoms; the manufacturer therefore did not market the 75 mg dose. It is for this reason that most analyses report results for the two lower doses only.

Patients with schizophrenia (n = 440) entered a four-arm, double-blind randomised controlled trial comparing the three doses of risperidone LAI with placebo. ^{Reference Kane, Eerdekens, Lindenmayer, Keith, Lesem and Karcher2} End-point scores on all three doses differed significantly from placebo for Positive and Negative Syndrome Scale (PANSS) scores (Fig. 1). ^{Reference Kay, Fiszbein and Opler3} Higher doses of risperidone LAI were associated with more extrapyramidal symptoms, although these were generally considered mild across dose groups. Both patients and investigators rated injection site pain as low.

Fig. 1

Change in Positive and Negative Syndrome Scale (PANSS) score over 12 weeks in patients with schizophrenia randomly assigned to double-blind treatment with long-acting injectable risperidone (25 mg, 50 mg or 75 mg) or placebo. Reproduced with permission from Kane et al. ^{Reference Kane, Eerdekens, Lindenmayer, Keith, Lesem and Karcher2} a. Significantly less improvement from baseline seen with placebo than with 50 and 75 mg of risperidone (P<0.01 and P<0.05 respectively, Dunnett's multiple comparison method). b. Significantly less improvement from baseline seen with placebo than with 50 mg of risperidone (P<0.01, Dunnett's multiple comparison method). c. Significantly less improvement from baseline seen with placebo than with 25 and 50 mg of risperidone (P<0.001 and P<0.05 respectively, Dunnett's multiple comparison method). d. Significantly less improvement from baseline seen with placebo than with 25 and 50 mg of risperidone (P<0.001 and P<0.05 respectively, Dunnett's multiple comparison method).

In a second double-blind non-inferiority controlled trial (a type of study with the aim of showing that one treatment arm does no worse than another), 640 patients with schizophrenia were prescribed 1–6 mg of oral risperidone for 8 weeks and patients whose condition was stable were then randomised to be treated with risperidone LAI or to continue with oral risperidone. ^{Reference Chue, Eerdekens, Augustyns, Lachaux, Molcan and Eriksson4} Both groups demonstrated significant improvement over baseline on PANSS scores and had statistically equivalent efficacy. As in the study by Kane et al, ^{Reference Kane, Eerdekens, Lindenmayer, Keith, Lesem and Karcher2} no unexpected adverse event was observed.

The safety and efficacy of switching from four FGA–LAIs to risperidone LAI (up to 75 mg) were the main outcome variables of another 12-week study. ^{Reference Turner, Eerdekens, Jacko and Eerdekens5} Before being switched, patients received two cycles of their original FGA–LAI. Compared with previous treatment, PANSS scores and extrapyramidal symptoms both decreased.

Colleagues and I assessed over 600 patients whose medications were changed from oral risperidone to risperidone LAI over a 1-year open label trial. ^{Reference Fleischhacker, Eerdekens, Karcher, Remington, Llorca and Chrzanowski6} Those pre-treated with 1–2 mg of oral risperidone were switched to 25 mg LAI, patients given 2–4 mg received 50 mg LAI and patients originally given 4–6 mg received 75 mg LAI every 2 weeks. Sixty-five per cent of patients completed the full trial, with comparable significant improvements in PANSS scores in all three groups. This study focused on long-term tolerability, and implications for efficacy are limited owing to non-randomised treatment assignment and the lack of a control group.

Secondary and post hoc analyses of registration trials

A number of secondary analyses have been performed using data from the studies outlined above. It should be emphasised that post hoc analyses have a distinctly different quality compared with studies testing hypotheses, and so results from such studies should be interpreted with caution. The large, 1-year open label trial, referred to above, ^{Reference Fleischhacker, Eerdekens, Karcher, Remington, Llorca and Chrzanowski6} was subjected to a number of post hoc analyses as outlined below.

1. (a) Participants were grouped according to prior antipsychotic: FGA–LAI, ^{Reference Lasser, Bossie, Gharabawi and Turner7} oral FGA, ^{Reference Van Os, Bossie and Lasser8} or oral risperidone. ^{Reference Lasser, Bossie, Gharabawi and Baldessarini9} Improvements in PANSS scores and extrapyramidal symptoms were evident for all but were most compelling for those switched from stable doses of oral risperidone to risperidone LAI. These results need to be confirmed in prospective clinical trials.

2. (b) The risk–benefit ratios for patients aged 65 years or older confirmed tolerability and efficacy as seen in younger populations. ^{Reference Lasser, Bossie, Zhu, Gharabawi, Eerdekens and Davidson10}

3. (c) Results for participants with schizoaffective disorders (n = 110) were detailed separately and showed significant PANSS score improvements. ^{Reference Lasser, Bossie, Gharabawi, Eerdekens and Nasrallah11}

4. (d) Five (0.9%) of 530 participants with schizophrenia but without dyskinesia at baseline subsequently met predefined tardive dyskinesia criteria. ^{Reference Gharabawi, Bossie, Zhu, Mao and Lasser12}

5. (e) The main findings of a healthcare resource utilisation analysis were a significant decrease in hospitalisation days and out-patient consultations, which led to the hypothesis that risperidone LAI would, in the long run, decrease healthcare costs for patients with schizophrenia. ^{Reference Leal, Rosillon, Mehnert, Jarema and Remington13}

6. (f) Young adults (⩽25 years of age; n = 66) showed no clinically relevant efficacy differences from the older patients, but were more susceptible to experiencing side-effects. ^{Reference Lasser, Bossie, Zhu, Locklear and Kane14}

7. (g) Eighty-five per cent of all patients who had fulfilled symptomatic remission criteria, ^{Reference Andreasen, Carpenter, Kane, Lasser, Marder and Weinberger15–Reference Lasser, Bossie, Gharabawi and Kane17} i.e. a score of ⩽3 on eight predefined items of the PANSS, at baseline maintained remission status for at least 6 months. Additionally, 21% of the patients without symptom remission at baseline achieved full remission status by the study end-point. This analysis demonstrated that the concept of remission is clinically testable, and confirms that continuous antipsychotic maintenance treatment has a significant impact on long-term outcome in schizophrenia.

The placebo-controlled trial by Kane et al, ^{Reference Kane, Eerdekens, Lindenmayer, Keith, Lesem and Karcher2} referred to above, was also subjected to post hoc analyses as outlined below.

1. (a) An analysis of health-related quality of life revealed significant improvements in some aspects of quality of life for patients on active treatment, compared with the placebo group. ^{Reference Nasrallah, Duchesne, Mehnert, Janagap and Eerdekens18} Interestingly, patients in the 25 mg dosage group reached normal 36-item Short Form Health Survey (SF–36) values in several domains after only 3 months, and similar results were seen in the 1-year open label trial. ^{Reference Fleischhacker, Eerdekens, Karcher, Remington, Llorca and Chrzanowski6,Reference Fleischhacker, Rabinowitz, Kemmler, Eerdekens and Mehnert19} These improvements were largely independent of changes in PANSS scores.

2. (b) With combined use of data from the placebo-controlled study ^{Reference Kane, Eerdekens, Lindenmayer, Keith, Lesem and Karcher2} and the 1-year study, ^{Reference Fleischhacker, Eerdekens, Karcher, Remington, Llorca and Chrzanowski6} injection site pain (measured with a 10 cm visual analogue scale) and patient satisfaction (measured by the Drug Attitude Inventory) ^{Reference Awad and Hogan20} were explored. ^{Reference Lindenmayer, Jarboe, Bossie, Zhu, Mehnert and Lasser21} Scores were low at baseline with comparable scores for risperidone LAI (all doses) and placebo injections, and pain scores tended to decrease during the two studies. This contrasts with notably higher pain scores for four FGA–LAIs in an older study. ^{Reference Bloch, Mendlovic, Strupinsky, Altshuler, Fennig and Ratzoni22}

3. (c) Participants from the placebo-controlled study ^{Reference Kane, Eerdekens, Lindenmayer, Keith, Lesem and Karcher2} and a short open label 12-week study ^{Reference Lindenmayer, Eerdekens, Berry and Eerdekens23} were invited to enrol in an extension study to explore longer-term tolerability. ^{Reference Lindenmayer, Khan, Eerdekens, Van Hove and Kushner24} Mean durations ranged between 347 (s.d. = 239) and 504 (s.d. = 505) days and the mean modal dose of risperidone LAI was 50 mg every 2 weeks. Approximately half of both study groups completed 12 months of treatment. Psychosis, headache, insomnia and agitation were the most frequent adverse events. Extrapyramidal symptoms were reported in 22–33% of patients (proportionally more in the 75 mg dose group).

Post-registration (mostly phase IV) clinical trials

Randomised controlled clinical trials. A report from Taiwan described a single-blind randomised comparison between daily oral risperidone and risperidone injections (25 mg, 37.5 mg or 50 mg) every 2 weeks in patients pre-treated with oral risperidone (study duration 48 weeks). ^{Reference Bai, Chen, Chen, Chang, Wu and Hung25} Among 50 hospitalised patients with symptomatically stable schizophrenia, 90% of whom completed the study, PANSS score change was no different between oral risperidone and risperidone LAI groups, but the authors noted a reduced efficacy of LAI at doses below 50 mg every 2 weeks. With regard to safety, patients in the LAI group were reported to experience fewer extrapyramidal symptoms and to have a lower score on the Udvalg for Kliniske Undersøgelser (UKU) side-effect rating scale, as well as lower prolactin levels.

A randomised double-blind 52-week study compared RLAI 25 mg and 50 mg given every 2 weeks. ^{Reference Simpson, Mahmoud, Lasser, Kujawa, Bossie and Turkoz26} Only half (51%) of the 324 randomised patients completed the study. Time to relapse (the primary outcome) was comparable for the two treatment groups, but a modestly lower relapse risk on the higher dose was observed. This also translated into improved psychosocial functioning, again without significant differences between treatment groups. A post hoc analysis of these data revealed that insight was correlated with psychosocial outcomes at end-point. ^{Reference Gharabawi, Bossie, Turkos, Kujawa, Mahmoud and Simpson27} Change in negative symptoms and duration of LAI treatment also contributed to better psychosocial outcome. Prolactin levels were significantly higher in the 50 mg group at end-point (50 mg/l).

The most recent addition to the reports of randomised controlled clinical trials of risperidone LAI is a study that compares this treatment with olanzapine tablets. ^{Reference Keks, Ingham, Khan and Karcher28} After data from a large number of patients were excluded from analysis because of protocol amendments and violations, approximately 200 patients receiving a mean dose of 14.6 mg oral olanzapine daily were compared with 155 patients who received a mean modal dose of 40.7 mg risperidone LAI every 2 weeks. There was no substantial group difference in primary and secondary efficacy outcomes, including PANSS scores, Clinical Global Impression (CGI) scores and maintenance of effect. Patients on LAI showed a higher risk of extrapyramidal symptoms and were more often prescribed anticholinergic drugs. The olanzapine group had a greater risk of weight gain and body mass index increase. Two patients died in the risperidone LAI group and six died in the olanzapine group, a fact that the authors do not comment on.

Open, observational and large simple trials. Lindenmayer et al studied symptomatically stable patients with schizophrenia switched from oral haloperidol, quetiapine or olanzapine to risperidone LAI (25–50 mg) and noted small but significant reductions in total PANSS scores by week 12, with switching being well tolerated. ^{Reference Lindenmayer, Eerdekens, Berry and Eerdekens23}

A sample of 1876 patients with schizophrenia or other psychoses were enrolled in a large, open prospective trial, the Switch to Risperidone Microspheres (StoRMi) study, in which patients were switched directly from their previous antipsychotic medication to risperidone LAI (25–50 mg). ^{Reference Möller, Llorca, Sacchetti, Martin, Medori and Parellada29} Three-quarters (74%) of the sample completed the 6-month trial. Significant improvements in PANSS, CGI, Global Assessment of Functioning, patient satisfaction and quality of life (SF–36) scores were noted. Severity of extrapyramidal symptoms also decreased during the 6-month trial period. Subanalyses of StoRMi have also been conducted and more or less mirror the evidence presented in earlier clinical trials. ³⁰ Llorca et al followed up 529 StoRMi participants: at 18 months the discontinuation rate was 56%, although 45% of patients had achieved remission criteria. ^{Reference Llorca, Sacchetti, Lloyd, Kissling and Medori31} Kissling et al noted that four-fifths of the 52 StoRMI participants aged 65 years or over had active concomitant somatic disease. ^{Reference Kissling, Glue, Medori and Simpson32} No patient had had a cerebrovascular adverse event and clinicians judged most of the emerging side-effects as mild or moderate in severity. In patients switching from FGA–LAIs to risperidone LAI a significant reduction in extrapyramidal symptoms was observed.

In a 1-year open label study for clinically stable patients with schizophrenia or schizoaffective disorders, ^{Reference Gharabawi, Gearhart, Lasser, Mahmoud, Zhu and Mannaert33} participants were given two 50 mg doses of risperidone LAI with oral supplementation in the first month and monthly injections thereafter for up to 48 weeks. The intention-to-treat population comprised 67 patients of whom 18% relapsed. Mean plasma concentrations (risperidone and 9-hydroxyrisperidone) were stable throughout the study period although there appeared to be more variability compared with fortnightly dosing. A dose increase to 75 mg once-monthly in 6–8 participants who had relapsed on 50 mg led to a significant improvement in PANSS total score. The authors cautiously interpret the findings from this pilot trial as encouraging and supportive for ‘a possible role for [risperidone] LAI 50 mg once monthly in certain stable patients’. ^{Reference Gharabawi, Gearhart, Lasser, Mahmoud, Zhu and Mannaert33} One cannot fail to notice that this trial, although performed 4 years before eventual publication, was never followed up by a controlled trial as suggested by the authors.

Most recently, Emsley et al in 2008 reported results of an open-label trial with risperidone LAI in recent-onset schizophrenia, schizophreniform or schizoaffective disorder. ^{Reference Emsley, Medori, Koen, Oosthuizen, Niehaus and Rabinowitz34} The sample consisted of people who had met DSM–IV criteria for these disorders for no more than 1 year. Of 50 participants, 72% completed this 2-year trial and 64% met remission criteria at the end-point. Improvements were noted in PANSS scores, CGI scores and scales measuring psychosocial outcomes. Most of the patients received either 25 mg or 37.5 mg injections; only 16% were treated with 2-weekly injections of 50 mg at end-point. Approximately a third of the patients developed extrapyramidal symptoms, including one participant with tardive dyskinesia. As observed in other first-episode samples, ^{Reference Kahn, Fleischhacker, Boter, Davidson, Vergouwe and Keet35,Reference Strassnig, Miewald and Keshawan36} there was significant weight gain and body mass index increase in this sample. This underscores the apparent higher likelihood of younger patients developing metabolic side-effects.

Analyses of administrative data, prescription audits and chart reviews. Three reports from UK prescription audits complement the available evidence. Taylor et al evaluated the written prescriptions and reported a significant improvement in CGI scores compared with the pre-treatment period; ^{Reference Taylor, Young, Mace and Patel37} they also found a considerable reduction in antipsychotic co-prescriptions. Patel et al attempted to find prognostic indicators for early discontinuation of risperidone LAI and found that patients who were switched from an oral antipsychotic to risperidone LAI were more likely to discontinue this drug than those who had previously been treated with an LAI antipsychotic. ^{Reference Patel, Young, Samele, Taylor and David38} Whether this report was derived from the same sample as the one by Taylor et al is not clearly explained. In an extension of these studies, ^{Reference Taylor, Young, Mace and Patel37,Reference Patel, Young, Samele, Taylor and David38} the same group, in a mix of a non-interventional, observational and chart review approaches, evaluated the treatment history of 250 patients who had received prescriptions of risperidone LAI. ^{Reference Taylor, Young and Patel39} Six months after starting treatment 47% of patients were still receiving this medication. Patients over 55 years old as well as those receiving a dose of more than 25 mg were more likely to continue with risperidone LAI. On the other hand, pre-treatment with clozapine reduced the likelihood of a continuous risperidone LAI prescription.

In another UK study, Niaz & Haddad compared the total number of in-patient days before and after risperidone LAI prescriptions by reviewing case notes of 90 patients treated for a median duration of 9.5 months. ^{Reference Niaz and Haddad40} This mirror-image analysis included 74 patients; 46 patients treated with oral antipsychotics during the same period served as a control group. The latter had significantly lower baseline rates of drug and alcohol misuse as well as fewer forensic problems. On most measures (admissions, compulsory admissions, in-patient days, treatment cost) risperidone LAI was advantageous when compared with previous treatment. The authors critically discuss the results within the limitations of a retrospective case-note study and submit that although they ‘cannot prove that risperidone LAI caused the benefits in the primary mirror image analysis, we report an association’. They nevertheless suggest improved adherence to be the most plausible explanation for the advantages of risperidone LAI. A study of similar sample size and design yet with a less sophisticated set of analyses, also performed in the UK, corroborated these findings. ^{Reference Taylor, Currie, Lloyd, Price and Peperell41}

Beneficiaries of Californian Medicaid were investigated in an analysis of administrative data evaluating fluphenazine decanoate, haloperidol decanoate and risperidone LAI. ^{Reference Olfson, Marcus and Ascher-Svanum42} A mirror-image design comparing the utilisation of these three drugs was adopted, and the sample was composed of large numbers of people treated with the two FGA–LAIs (fluphenazine n = 948, haloperidol n = 1631) and 116 people treated with risperidone LAI. The authors compared the 180 days before with the 180 days after initiating long-acting antipsychotics. Although the authors do report differences between the three LAIs, these appear somewhat irrelevant from a clinical perspective given that less than 10% of the patients received injections for the full 180-day observation period, and treatment lasted on average 39–48 days. The authors duly note treatment practice differences between California and Europe, and suggest that higher levels of psychiatric comorbidity within their sample contributed to the higher rates of polypharmacy. Like the authors of the two studies reviewed above, ^{Reference Niaz and Haddad40,Reference Taylor, Currie, Lloyd, Price and Peperell41} they also acknowledge the limitations of analysing administrative data and medical records. ^{Reference Olfson, Marcus and Ascher-Svanum42}

Pharmacokinetic studies. A number of research groups have specifically investigated the pharmacokinetics of risperidone LAI. Castberg & Speakset studied drug plasma levels in 30 patients receiving risperidone LAI (12–14 days after the final injection, and no sooner than after the fourth injection) compared with 278 patients given oral risperidone. ^{Reference Castberg and Spigset43} Serum concentrations of risperidone increased from 38 nm/l (2 weeks after 25 mg/14 days) to 148nm/l (75 mg/14 days) in a dose-dependent manner. The concentration–dose ratio for oral risperidone was much more variable.

A retrospective analysis of a therapeutic drug monitoring programme in Norway found a significant percentage of patients treated with RLAI to have plasma levels below what the authors considered to be an established reference range (30–120 nm/l), and they too noted substantial individual variation in serum levels. ^{Reference Nesvag and Tanum44} One suggested reason for this was faulty injection technique (into fatty rather than muscular tissue). Polypharmacy was common in both Scandinavian studies and must be considered a limitation when interpreting the results, and no account was taken of cigarette smoking, which is known to have an impact on some antipsychotic plasma levels. ^{Reference Haring, Meise, Humpel, Saria, Fleischhacker and Hinterhuber45}

A more rigorous bioavailability study of risperidone LAI was conducted by the parent pharmaceutical company. ^{Reference Eerdekens, Van Hove, Remmerie and Mannaert46} This 15-week open label study included 86 participants with schizophrenia whose symptoms were stable. Patients were first stabilised on 2 mg, 4 mg or 6 mg of oral risperidone for at least 4 weeks and then switched to receive 25 mg, 50 mg or 75 mg of risperidone intramuscularly every 2 weeks. Plasma levels of the active moiety (risperidone plus the major active metabolite 9-hydroxy-risperidone) and risperidone were prospectively collected. There was a dose-proportional increase in steady-state plasma concentrations of the active moiety. Mean steady-state C _max concentrations of the active moiety were 25–32% lower after risperidone LAI than after oral risperidone, but this was not the case for C _min concentrations. Based on these data the authors postulate bioequivalence of the oral and intramuscular doses tested, although this clearly needs further study, as also argued by Bai et al. ^{Reference Bai, Chen, Chen, Chang, Wu and Hung25} Plasma level fluctuations were considerably lower for risperidone LAI and this is believed to enhance the safety/tolerability characteristics of risperidone. Some preliminary evidence for this is provided by Bai et al, who described reduced side-effects and serum prolactin levels associated with a ‘relative stability of serum level metabolites’. ^{Reference Bai, Chen, Chen, Chang, Wu and Hung25}

Imaging studies. Two studies investigated the pharmacodynamics of risperidone LAI using positron emission tomography (PET). ^{Reference Gefvert, Eriksson, Persson, Helldin, Björner and Mannaert47,Reference Uchida, Mamo, Kapur, Labelle, Shammi and Mannaert48} Gefvert et al found steady state of the plasma active moiety of risperidone from the fourth injection onwards, and this was maintained for 4–5 weeks after the last injection. ^{Reference Gefvert, Eriksson, Persson, Helldin, Björner and Mannaert47} Dopamine D₂ receptor occupancy evaluated 2 weeks after the fifth injection in eight participants revealed a range of occupancy of 25–83% in a more or less dose-dependent fashion (25 mg, 50 mg, 75 mg). Uchida et al assessed monthly injections of risperidone LAI, examining D₂ receptor occupancy using [¹¹C]raclopride PET. ^{Reference Uchida, Mamo, Kapur, Labelle, Shammi and Mannaert48} After at least three consecutive monthly injections of 50 mg risperidone LAI, seven participants had a PET scan within 4 days of the next scheduled injection. The mean total plasma level of the active moiety was 16.6 ng/ml and D₂ occupancy ranged between 29% and 82% with a mean value of 56%. None of the patients relapsed during the study. Thus it may not be necessary to achieve continuous striatal D₂ blockade greater than 60% to maintain efficacy.

Surguladze et al used functional magnetic resonance imaging to compare 16 people receiving risperidone LAI with 16 people treated with various FGA–LAIs. ^{Reference Surguladze, Chu, Evans, Anilkumar, Patel and Timehin49} They found that those given FGA–LAIs had worse performance on various memory tasks, with differing cortical activation patterns particularly in the medial and ventrolateral prefrontal areas, compared with those receiving risperidone LAI and the control participants.

Other studies. Generally, individuals with schizophrenia have higher rates of comorbidity of substance misuse, and a group of such patients was the subject of an open, randomised 6-month clinical trial evaluation of the efficacy of risperidone LAI and zuclopenthixol LAI. ^{Reference Rubio, Martinez, Ponce, Jimenez-Arriero, Lopez-Munoz and Alamo50} In total 115 patients were randomised to receive one of the two drugs during in-patient treatment. The main outcome variable was the number of positive urine tests, but psychopathology and adherence to a specific psychotherapeutic programme were also assessed. At end-point the mean dose of risperidone was 47.2 mg LAI per 15 days with an additional 3.4 mg daily of the oral drug. In the zuclopenthixol group patients received 200 mg LAI every 3 weeks and a mean daily oral dose of the same drug of 15 mg. Patients on zuclopenthixol had a significantly higher number of positive urine tests. Risperidone showed advantages over zuclopenthixol in PANSS scores and patients also attended more of the psychotherapeutic sessions. One of the explanations the authors offer is the reduced propensity of SGAs to induce extrapyramidal symptoms and dysphoria, both of which may serve as a motive for people with schizophrenia to misuse substances.