Schizophrenia is a chronic, disabling illness in which most patients experience frequent periods of psychotic exacerbation and relapse (Csernansky et al., 2002). Relapse in schizophrenia can be associated with progressive functional deterioration, declining treatment response, worsening clinical outcome, escalating caregiver burden, and an increased economic burden for families and society (Wiersma et al., 1998; Almond et al., 2004; Awad and Voruganti, 2008; Hong et al., 2009). Therefore, the prevention of relapse is a primary treatment goal for the successful long-term management of schizophrenia.
In patients with schizophrenia, nonadherence to treatment is a common occurrence driven by multiple factors and is considered to be a key contributor to relapse (Jeong and Lee, 2013). A number of studies have reported an association between poor or inconsistent adherence to antipsychotic (AP) medications and relapse in patients with schizophrenia (Robinson et al., 1999; Valenstein et al., 2002; Ascher-Svanum et al., 2006; Morken et al., 2008). Adherence issues make long-acting injectable AP formulations an attractive option because they provide convenient and reliable medication delivery (Kane, 2011). This ensures that physicians have full knowledge of whether adherence is achieved in patients with schizophrenia (Kane, 2006) and thus aids in relapse prevention. However, an observational cohort study recently showed that a subgroup of patients with schizophrenia will relapse despite uninterrupted long-acting therapy (LAT) (Ciudad et al., 2012).
Understanding the characteristics of patients who relapse despite receiving uninterrupted LAT may provide physicians with greater insights into more effective management of maintenance treatment and identify areas on which to focus future research. To this end, a post-hoc analysis of a long-term study of patients with schizophrenia treated with LATs was carried out to identify the factors that lead to relapse despite adherence to LAT.
Aim of the study
The primary objective of this exploratory post-hoc analysis of patients with schizophrenia or schizoaffective disorder is to identify variables associated with relapse despite continuous treatment with AP LAT.
This is an exploratory post-hoc analysis of a randomized, double-blind, multicenter, international, 1-year study (SCH-401; NCT00297388) of risperidone long-acting injection (RLAI) in patients with stable schizophrenia or schizoaffective disorder (Simpson et al., 2006). The methods of the study included in the current analysis have been described in detail elsewhere (Simpson et al., 2006). The original study was approved by the ethics committee or institutional review board at each site, and was carried out in accordance with the ethical principles in the Declaration of Helsinki. Written informed consent was obtained from all patients. This study was chosen to evaluate predictors of relapse because it included the following criteria: enrollment of patients with a schizophrenia diagnosis, active treatment with a long-acting injectable AP medication, use of relapse as a study end point, a study duration of 1 year or more, and the availability of patient-level data. Patients (N=342) were discontinued from previous oral AP and randomly assigned to biweekly intramuscular injections of RLAI 50 (n=161) or 25 mg (n=163) for a planned treatment duration of 52 weeks (Simpson et al., 2006). This analysis assessed data on the intent-to-treat population (n=323), which was defined as all randomly assigned patients who received at least one dose of study medication.
The details of the patients included in the original study have been described elsewhere (Simpson et al., 2006). Patients were outpatients aged 18 to 70 years with a diagnosis of schizophrenia or schizoaffective disorder [according to Diagnostic and Statistical Manual of Mental Disorders, 4th ed. (DSM-IV) criteria] (American Psychiatric Association, 1994). Before baseline in the original study, patients were required to have been clinically stable (i.e. they had stable symptoms without hospitalization for worsening symptoms or aggressive behavior) for a minimum of 4 months and receiving stable doses of oral APs, excluding clozapine (Simpson et al., 2006).
Assessment of relapse
The primary end point of the original study was time to relapse in the intent-to-treat population (Simpson et al., 2006). Relapse was defined as any one of the following after the initial 20 days of the study (i.e. the time point at which therapeutic blood levels of RLAI are reached): psychiatric hospitalization because of worsening symptoms; an increase in the level of psychiatric care needed (i.e. significant crisis intervention was required to avert hospitalization), and at least 25% worsening or an increase from baseline of 25% in Positive and Negative Syndrome Scale (PANSS) score from baseline (Kay et al., 1987), occurring within 2 weeks of each other, substantial clinical deterioration, as indicated by a score of 6 or 7 on the Clinical Global Impressions–Severity scale (Guy, 1976), deliberate self-injury, clinically significant suicidal or homicidal ideation (as determined by the investigator), or violent behavior leading to clinically significant injury of another individual or property damage; or a requirement for more than three 1-week periods of oral risperidone therapy within a 3-month period after day 21 (Simpson et al., 2006).
Variables associated with the risk of relapse were initially identified using univariate Cox proportional hazards regression models. Variables assessed included duration of illness [years, continuous and categorical (≤5, 5–10, ≥10)], diagnosis (schizophrenia vs. schizoaffective disorder), number of previous hospitalizations (≥1 vs. 0), RLAI treatment (50 vs. 25 mg), previous AP dosing (>4 mg/day vs. ≤4 mg/day, risperidone equivalent), sex (female vs. male), race (black vs. white, other vs. white, Hispanic vs. white), country (Canada vs. USA and Argentina/Chile vs. USA), age at onset (years), current age (years), baseline BMI, baseline Personal and Social Performance scale score, and baseline PANSS total and factor scores. Statistical significance was defined as P value less than 0.05 and a trend was defined as P value more than or equal to 0.05 and P value less than 0.10.
Although univariate regression analysis is useful for showing basic relationships with time to relapse, clinical practice suggests that multiple factors can influence the risk of relapse. As a result, stepwise multiple Cox proportional hazards models were used to examine the results of the initial univariate models. Study design variables, treatment, and previous AP dosing (>4 vs. ≤4 mg/day, risperidone equivalent) were retained in each step. Predictor variables were entered sequentially in the model; variables that became nonsignificant were subsequently checked and removed from the model (i.e. entry P≤0.1 and stay P≤0.1). The stepwise selection process consists of a series of alternating forward selection and backward elimination steps. The former adds variables to the model whereas the latter removes variables. Model fits and diagnostics were examined for violation of the assumption of proportional hazards, for influential data points, and for nonlinearity. Multicollinearity among the predictors was also assessed using various statistical tools. The importance of predictors was examined using differences on −2 log likelihoods from the Cox regression models to prevent inclusion of correlated predictors in the final model. Summaries of reasons for relapse by duration of illness and country were also assessed. In addition, comparisons of continuous variables were performed using Student’s t-test and comparisons of categorical variables were performed using the χ2-test between relapsers and nonrelapsers. No adjustments were made for multiplicity.
Overall, 59 of the 323 (18.3%) patients included in the analysis developed relapse despite continuous treatment with RLAI. Baseline demographic and clinical characteristics for the overall population and for patients who did and did not relapse are summarized in Table 1. The [mean (SD)] age in the overall population was 41.0 (11.9) years, with the majority of patients being male (62.2%) and white or black (49.2 and 26.0%, respectively). In general, patients who relapsed had a longer duration of illness [mean (SD) 19.5 (11.7) vs. 14.6 (10.7) years], had a greater number of previous psychiatric hospitalizations (98.3 vs. 89.0% of patients with ≥1 previous psychiatric hospitalization), and received higher doses of previous APs (62.7 vs. 44.7% of patients received doses of >4 mg/day risperidone equivalents) (Table 1). Furthermore, the relapse subgroup versus the nonrelapse subgroup included a higher percentage of patients from Canada (25.4 vs. 9.5%) and a lower percentage of patients from the USA (57.6 vs. 73.9%). Baseline mean Personal and Social Performance and PANSS scores appeared to be similar between the relapse and nonrelapse subgroups. In the overall population, the most common reasons for relapse were psychiatric hospitalization (8.1%) and change in the Clinical Global Impressions–Change score (4.7%) (Table 2). Overall, 20% of the study population had a diagnosis of schizoaffective disorder. However, the percentages of relapsers and nonrelapsers were similar in patients who had a diagnosis of schizoaffective disorder (15.3 vs. 21.2%).
Drug-exposure data indicated that the [mean (SD)] RLAI drug doses appeared to be similar between patients who did and did not relapse: 35.2 (12.4) versus 38.0 mg (12.5), respectively. As expected, the [mean (SD)] duration of RLAI treatment was significantly shorter among patients who relapsed versus those who did not relapse [138.3 (97.6) vs. 263.1 (119.9) days; P<0.0001]; the median durations were 111.0 and 340.5 days, respectively.
Univariate predictors of relapse
Univariate regression analyses showed that duration of illness (>10 vs. ≤5 years; P=0.0255), previous AP dosing (>4 vs. ≤4 mg/day, risperidone equivalents; P=0.0243), treatment in Canada versus the USA (P=0.0016), and negative symptoms (PANSS negative factor; P=0.0384) were significantly associated with greater risk of relapse at P value less than 0.05 level (Table 3). In addition, previous number of psychiatric hospitalizations (≥1 vs. 0; P=0.0642) and current age (P=0.0511) were also identified to be associated with relapse at P value less than 0.10 level. Although 20% of patients in the study population had a diagnosis of schizoaffective disorder, the analysis found that there was no unique, additional association between risk of relapse and diagnosis type (i.e. schizophrenia vs. schizoaffective disorder).
Stepwise multiple regression analyses
Correlational analyses were run among related variables of interest; duration of illness, number of previous hospitalizations, current age, and age at onset were shown to be highly correlated. Numerous regression models were run to determine whether one of these variables emerged as more important than the others. The relative importance of predictors was examined using differences on −2 log likelihoods from the Cox regression models. In addition, the regression coefficients from univariate regression models were considered in deciding the importance of predictors. No single variable emerged as being most important. Duration of illness was selected not only for numerical reasons for inclusion in the stepwise models but also because it was considered to be the most clinically relevant factor. On the basis of entry and stay criteria of less than or equal to 0.10, the stepwise multiple Cox proportional hazards model showed that duration of illness of >10 versus ≤5 years (P=0.0181) and being in Canada versus the USA (P=0.0008) were significantly associated with an increased risk of relapse (Table 4). The duration of illness in years as a continuous predictor was also included in a separate stepwise multiple Cox proportional hazards model and showed that each 1-year increase in duration of illness was associated with a 6.0% increase in the risk of relapse (P=0.0003) in the population sample studied.
The smoothed hazard functions for relapse were graphed by duration of illness (Fig. 1a and b). The risk of relapse was lowest in patients with a duration of illness of less than or equal to 5 years and highest for those with a duration of illness of more than 10 years (Fig. 1a and b).
Psychiatric hospitalization was the most common outcome accounting for relapse in all three duration of illness categorical groups (≤5 years, 3.9%; 6–10 years, 11.8%; >10 years, 8.9%) (Table 2). The most common outcomes accounting for relapse by country of origin were administration of rescue medication in Canada (17.5%), psychiatric hospitalization in the USA (9.2%), and more than 25% change in the PANSS total score in Argentina/Chile (9.3%).
Adverse events for relapse versus nonrelapse groups
The majority of the patients in the relapse (100%) and nonrelapse (86.7%) groups experienced at least one treatment-emergent adverse event (AE). The most common AEs (≥10%) for the relapse group were psychotic disorder not otherwise specified (NOS; 47.5%), insomnia (42.4%), schizophrenia NOS (39.0%), anxiety (18.6%), headache (15.3%), auditory hallucination (11.9%), suicidal ideation (10.2%), agitation (10.2%), and upper respiratory infection NOS (10.2%). The most common (≥10%) AEs for the nonrelapse group were insomnia (24.2%), headache (19.3%), anxiety (15.9%), psychotic disorder NOS (14.4%), and nasopharyngitis (10.2%). The percentage of patients with extrapyramidal symptom-related AEs was 20.3% for the relapse group and 20.5% for the nonrelapse group. The most common extrapyramidal symptom-related AEs for the relapse and nonrelapse groups were akathisia (5.1 vs. 7.6%) and tremor (8.5 vs. 4.9%).
The aim of this exploratory post-hoc analysis was to identify factors associated with an increased risk of relapse in patients with schizophrenia or schizoaffective disorder who were receiving RLAI. The main results of this analysis indicate that in the population sample studied, a longer duration of illness (especially >10 years) was associated with an increased risk of relapse compared with a shorter duration of disease, despite continuous LAT.
Previous studies have identified additional factors that may be associated with remission or relapse in patients with schizophrenia. An analysis of data from a multicenter cohort study suggests that lower quality of life, as assessed by the 36-item Short Form Health Survey, may be an independent predictor of increased risk for relapse in patients with schizophrenia (Boyer et al., 2013). Other predictors of remission in patients with schizophrenia have also been identified. These include lower PANSS global, negative, or general psychopathology subscores; a shorter duration of untreated psychosis; and early treatment response (Schennach-Wolff et al., 2009, 2011). A study measuring symptomatic and functional remission in patients with schizophrenia treated with RLAI indicates that baseline symptom severity, country of origin, higher mode doses of RLAI, male sex, and the additional use of psychoactive medications may predict relapse (Lambert et al., 2010). A systematic literature review also identified numerous potential treatment-related, disease-related, and patient-related drivers of relapse, such as adherence problems, substance abuse, and hospitalization/relapse history (Olivares et al., 2013). Finally, a long-term prognostic study found a positive correlation between the duration of untreated psychosis and the number of relapses over 12 years of follow-up in patients with schizophrenia (Cechnicki et al., 2011).
Such predictors may aid in the identification of persons who are at risk of relapse and may suggest populations to target for interventions that might decrease this risk. These predictors may also support treatment selection because a post-hoc analysis of AP drug trials in schizophrenia indicates that chronicity of illness may have an impact on between-drug treatment differences (Leucht et al., 2007).
Emerging models investigating the stages of schizophrenia suggest that the neurobiology of the disease may change with each successive relapse. These models for disease progression might explain why duration of illness and the number of previous relapses would predict response to AP treatment as observed in this study. It is speculated that psychotic episodes may produce neurotoxic effects (through neuroinflammation and oxidative stress) that lead to changes in the brain’s neurochemical infrastructure and/or connectivity (i.e. loss of white matter integrity). Thus, a long cumulative duration of psychosis because of inadequate or lack of treatment or multiple episodes of psychosis may result in the gradual loss of responsiveness to dopamine blockade and make patients more vulnerable to environmental stressors that lead to relapse. Such changes might be the basis for increased treatment resistance and further clinical and functional deterioration (Shenton et al., 2010; Bartzokis et al., 2011; Reis et al., 2014). This model is supported by a recent study that has identified an association between relapse and changes in both cognition and intracortical myelin and by a randomized study showing the robust value of uninterrupted treatment of treating patients with first-episode schizophrenia with long-acting APs compared with oral APs (Bartzokis et al., 2011; Subotnik et al., 2015).
Another consideration is that a longer duration of illness may be indicative of greater progression of the disease process. This hypothesis is supported by the identification of certain items identified in the univariate model, including greater severity of negative symptoms and higher doses of previous APs in individuals with a longer duration of illness.
In this analysis, receipt of LAT in Canada was identified as a significant predictor of relapse. Although this may reflect country-based differences in treatment practice, caution must be exercised when interpreting this result because it may represent chance findings arising from the limited sampling of Canadian patients and sites (n=40 at two sites) and differences in how relapses were identified at these sites. Although not evaluated as predictors for relapse, the marked differences in two of the most common AEs in the relapse group versus the nonrelapse group suggest that insomnia (42.4 vs. 24.2%) and agitation (10.2 vs. 3.8%) may be associated with relapse and their onset may predict impending relapse.
This work has several limitations that affect interpretation of the results. The variables tested in this analysis and its overall findings are based on data from a single, controlled, clinical trial database with an explanatory design that included multiple inclusion and exclusion criteria. Therefore, the results may not be generalizable to broader patient groups, such as those with active substance abuse (a known contributor to relapse) who were excluded from the studies. Further, the participants were not selected randomly and were treated in the context of a highly supervised clinical trial that may not reflect results in a naturalistic setting. Both design considerations increase uncertainty about the generalizability of the results.
The relatively small number of relapses and the 1-year duration of the study limit a robust examination of all variables that might be related to the explanation of relapse risk. Furthermore, these relapses were defined by several component outcomes, ranging from hospitalization to changes in PANSS scores. These components, although standard for most explanatory-type studies, such as study NCT00297388, almost certainly differ from the definitions of relapse used in most clinical settings. As a consequence, the identified predictors of relapse found in this study may differ from those that predict relapse in a naturalistic study.
The main findings of this research suggest that patients with a longer duration of illness (i.e. particularly >10 years) have a higher risk of relapse despite ensured adherence to LAT than those with more recent disease onset. These exploratory results suggest that increasing chronicity may increase the likelihood of future relapses despite certain knowledge of treatment at effective doses. Although these exploratory results are purely hypothesis generating, they do raise a number of important questions. Does repeated relapse result in treatment resistance to dopamine antagonists? Does relapse prevention throughout the course of illness decrease the risk of future relapse and possibly modify the course of schizophrenia? Is the early psychotic episode a distinct treatment-responsive stage of the illness that moves to a different stage after subsequent episodes? Additional prospective studies are required to answer these questions and to show that the results identified here are generalizable to the broad population of patients with schizophrenia when they are treated in a real-world environment.
The authors thank Maxwell Chang and Matthew Grzywacz, PhD, of ApotheCom (Yardley, Pennsylvania, USA) for their writing and editorial assistance, which was funded by Janssen Scientific Affairs, LLC. The analysis was also supported by Janssen Scientific Affairs.
Author contributions: Larry Alphs, Henry A. Nasrallah, Cynthia A. Bossie, and Dong-Jing Fu contributed to the design and interpretation of the post-hoc analysis and to manuscript preparation; Ibrahim Turkoz contributed to the statistical analysis of the data, the design and interpretation of the post-hoc analysis, and manuscript preparation; Srihari Gopal and David Hough contributed to the design of the study, the design and interpretation of the post-hoc analysis, and manuscript preparation. All authors approved the final version of the manuscript before submission.
Conflicts of interest
Larry Alphs, Cynthia Bossie, and Dong-Jing Fu are employees of Janssen Scientific Affairs, LLC, and Johnson & Johnson stockholders. Henry Nasrallah has received honoraria from Acadia, Alkermes, Allergan, Grünenthal, Boehringer-Ingelheim, Genentech, Janssen, Merck, Lundbeck, Novartis, Otsuka, Roche, Shire, Teva, and Sunovion, and has received research grants provided through his university from Forest Laboratories, Janssen, Merck, Novartis, Otsuka, and Sunovion. Srihari Gopal, David Hough, and Ibrahim Turkoz are employees of Janssen Research & Development, LLC, and Johnson & Johnson stockholders.
Almond S, Knapp M, Francois C, Toumi M, Brugha T (2004). Relapse in schizophrenia
: costs, clinical outcomes and quality of life. Br J Psychiatry 184:346–351.
American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders, 4th ed. Washington, DC: American Psychiatric Association.
Ascher-Svanum H, Faries DE, Zhu B, Ernst FR, Swartz MS, Swanson JW (2006). Medication adherence and long-term functional outcomes in the treatment of schizophrenia
in usual care. J Clin Psychiatry 67:453–460.
Awad AG, Voruganti LN (2008). The burden of schizophrenia
on caregivers: a review. Pharmacoeconomics 26:149–162.
Bartzokis G, Lu PH, Amar CP, Raven EP, Detore NR, Altshuler LL, et al. (2011). Long acting injection versus oral risperidone in first-episode schizophrenia
: differential impact on white matter myelination trajectory. Schizophr Res 132:35–41.
Boyer L, Millier A, Perthame E, Aballea S, Auquier P, Toumi M (2013). Quality of life is predictive of relapse in schizophrenia
. BMC Psychiatry 13:15.
Cechnicki A, Hanuszkiewicz I, Polczyk R, Bielańska A (2011). Prognostic value of duration of untreated psychosis in long-term outcome of schizophrenia
. Med Sci Monit 17:CR277–CR283.
Ciudad A, San L, Bernardo M, Olivares JM, Polavieja P, Valladares A, Gilaberte I (2012). Relapse and therapeutic interventions in a 1-year observational cohort study of nonadherent outpatients with schizophrenia
. Prog Neuropsychopharmacol Biol Psychiatry 36:245–250.
Csernansky JG, Mahmoud R, Brenner R (2002). A comparison of risperidone and haloperidol for the prevention of relapse in patients with schizophrenia
. N Engl J Med 346:16–22.
Guy W (1976). ECDEU Assessment Manual for Psychopharmacology [028 Clinical Global Impressions (CGI)]. Rockville, MD: National Institutes of Health. 218–222.
Hong J, Windmeijer F, Novick D, Haro JM, Brown J (2009). The cost of relapse in patients with schizophrenia
in the European SOHO (Schizophrenia
Outpatient Health Outcomes) study. Prog Neuropsychopharmacol Biol Psychiatry 33:835–841.
Jeong HG, Lee MS (2013). Long-acting injectable antipsychotics in first-episode schizophrenia
. Clin Psychopharmacol Neurosci 11:1–6.
Kane JM (2006). Review of treatments that can ameliorate nonadherence in patients with schizophrenia
. J Clin Psychiatry 67 (Suppl 5):9–14.
Kane JM (2011). Improving treatment adherence in patients with schizophrenia
. J Clin Psychiatry 72:e28.
Kay SR, Fiszbein A, Opler LA (1987). The positive and negative syndrome scale (PANSS) for schizophrenia
. Schizophr Bull 13:261–276.
Lambert M, de Marinis T, Pfeil J, Naber D, Schreiner A (2010). Establishing remission and good clinical functioning in schizophrenia
: predictors of best outcome with long-term risperidone long-acting injectable treatment. Eur Psychiatry 25:220–229.
Leucht S, Davis JM, Engel RR, Kane JM, Wagenpfeil S (2007). Defining ‘response’ in antipsychotic drug trials: recommendations for the use of scale-derived cutoffs. Neuropsychopharmacology 32:1903–1910.
Morken G, Widen JH, Grawe RW (2008). Non-adherence to antipsychotic medication, relapse and rehospitalisation in recent-onset schizophrenia
. BMC Psychiatry 8:32.
Olivares JM, Sermon J, Hemels M, Schreiner A (2013). Definitions and drivers of relapse in patients with schizophrenia
: a systematic literature review. Ann Gen Psychiatry 12:32.
Reis Marques T, Taylor H, Chaddock C, Dell’acqua F, Handley R, Reinders AA, et al. (2014). White matter integrity as a predictor of response to treatment in first episode psychosis. Brain 137 (Pt 1):172–182.
Robinson D, Woerner MG, Alvir JM, Bilder R, Goldman R, Geisler S, et al. (1999). Predictors of relapse following response from a first episode of schizophrenia
or schizoaffective disorder. Arch Gen Psychiatry 56:241–247.
Schennach-Wolff R, Jäger M, Seemüller F, Obermeier M, Messer T, Laux G, et al. (2009). Defining and predicting functional outcome in schizophrenia
spectrum disorders. Schizophr Res 113:210–217.
Schennach-Wolff R, Jager M, Mayr A, Meyer S, Kühn KU, Klingberg S, et al. (2011). Predictors of response and remission in the acute treatment of first-episode schizophrenia
patients – is it all about early response? Eur Neuropsychopharmacol 21:370–378.
Shenton ME, Whitford TJ, Kubicki M (2010). Structural neuroimaging in schizophrenia
: from methods to insights to treatments. Dialogues Clin Neurosci 12:317–332.
Simpson GM, Mahmoud RA, Lasser RA, Kujawa M, Bossie CA, Turkoz I, et al. (2006). A 1-year double-blind study of 2 doses of long-acting risperidone in stable patients with schizophrenia
or schizoaffective disorder. J Clin Psychiatry 67:1194–1203.
Subotnik KL, Casaus LR, Ventura J, Luo JS, Hellemann GS, Gretchen-Doorly D, et al. (2015). Long-acting injectable risperidone for relapse prevention and control of breakthrough symptoms after a recent first episode of schizophrenia
. a randomized clinical trial. JAMA Psychiatry 72:822–829.
Valenstein M, Copeland LA, Blow FC, McCarthy JF, Zeber JE, Gillon L, et al. (2002). Pharmacy data identify poorly adherent patients with schizophrenia
at increased risk for admission. Med Care 40:630–639.
Wiersma D, Nienhuis FJ, Slooff CJ, Giel R (1998). Natural course of schizophrenic disorders: a 15-year follow-up of a Dutch incidence cohort. Schizophr Bull 24:75–85.