Recent work investigating antipsychotic drugs has consistently demonstrated that early response within 2 weeks of initiating treatment can predict longer-term treatment outcomes.1–3 This finding has raised the possibility that patients who fail to respond to a particular medication after 2 weeks may benefit from immediately switching to another antipsychotic,4 although few studies have directly tested such an approach. While a great deal of evidence has accumulated with respect to predicting improvements in psychiatric symptoms, less is known about the relationship between early response and antipsychotic-induced extrapyramidal side effects (EPSs), particularly in patients beginning antipsychotic treatment for the first time. These side effects constitute an important aspect of treatment outcome, and the ability to predict them based on early antipsychotic response would enhance physicians’ ability to rapidly and effectively treat psychotic patients.
Extrapyramidal side effects are a diverse group of movement disorders that frequently occur in patients treated with antipsychotic drugs, commonly categorized as dystonia, parkinsonism, akathisia, and dyskinesia. In the CATIE schizophrenia trial, 12-month event rates of parkinsonism and akathisia were 37% to 44% and 26% to 35%, respectively, whereas 12-month event rates for dyskinesia were lower at 8% to 12%.5 Dystonic reactions are rarely reported in major trials of patients with prior antipsychotic exposure such as the CATIE schizophrenia trial, but studies of antipsychotic-naive patients have reported incidences of dystonia from 15% to 37%.6–8 These distressing side effects impair multiple aspects of patient functioning,9 and akathisia in particular has been associated with depression and suicidality.10,11 Moreover, EPSs are an important cause of treatment discontinuation.5,12 Preventing and ameliorating these side effects are crucial aspects of antipsychotic treatment, so being able to predict EPSs based on early antipsychotic response may be an important tool for optimizing treatment of psychotic patients, particularly those experiencing first-episode psychosis. For instance, patients suspected to be at high risk of EPSs may benefit from being treated with particular antipsychotic drugs, lower antipsychotic doses, or prophylactic anticholinergic medication, thereby minimizing EPSs and maximizing treatment compliance.
Several previous articles have taken preliminary steps toward describing the relationship between early antipsychotic response and EPSs. For example, Kinon et al4 found that early response or nonresponse to risperidone (using a cutoff value of ≥20% Positive and Negative Syndrome Scale total score reduction) did not predict the incidence of akathisia or parkinsonism in a 12-week study period. Switching early nonresponders to olanzapine did result in a decreased frequency of dyskinesia, but this may have been unrelated to early nonresponse and merely a result of switching medications. Schennach-Wolff et al13 found that early nonresponse was associated with more EPSs occurring within the first 2 weeks of treatment, but did not investigate whether this association persisted beyond the early treatment period. Finally, Stauffer et al2 reported that both early responders and early nonresponders experienced parkinsonism early in treatment, but these symptoms eventually improved for early responders, resulting in more parkinsonism in early nonresponders by weeks 10 to 12 of treatment. This is a compelling result, but interpreting it is difficult because analyses in the study included pooled data from patients treated with olanzapine or haloperidol. Recent results have suggested that early response to olanzapine does not have the same predictive value as early response to other antipsychotics.3,14,15 As such, olanzapine early responders may not be equivalent to haloperidol early responders. Furthermore, none of these earlier studies specifically investigated patients with no previous antipsychotic exposure. Because these patients appear to be at higher risk of antipsychotic-induced EPSs,16,17 it is of particular interest to predict and prevent EPSs for them.
Another issue of interest is whether affective symptoms accompanying psychosis are associated with EPS risk. Some studies have found that patients with prominent affective symptoms are more likely to experience antipsychotic-induced EPSs18 or spontaneous parkinsonism prior to antipsychotic treatment.19 However, others have found that depression in schizophrenic patients is associated with akathisia but not other EPSs20 or that there is a negative correlation between parkinsonism and depression.21 These studies suggest a relationship between affective symptoms and EPSs in psychotic patients, but once again, the exact nature of that relationship is unclear. The existence of this relationship raises the question of whether early improvement in affective symptoms following antipsychotic treatment might assist in predicting EPS risk.
Early antipsychotic response has emerged as a reliable and clinically valuable predictor of psychiatric outcomes, but little is known about whether it can also predict the risk of side effects. In the current study, we address this issue in a sample of antipsychotic-naive, first-episode psychosis patients treated with haloperidol. We further investigate the relationship between psychiatric symptoms and EPSs by determining whether early improvement in affective symptoms predicts EPS risk. As the value of assessing early response becomes better recognized, we hope to clarify which signs and symptoms at this time point provide the most clinical utility, allowing physicians to optimize treatment for psychotic patients.
This was an observational longitudinal study in which first-episode psychosis patients admitted to our adult psychiatric inpatient service from 1989 to 2002 were prospectively studied to assess psychiatric outcomes and neurologic side effects. Upon admission to hospital, patients or substitute decision makers gave informed consent, and patients were fully assessed before any antipsychotic treatment was initiated. Study protocols were approved by the McMaster University Research Ethics Board.
For the current analysis, we identified those patients who were antipsychotic-naive upon admission and treated with haloperidol for at least 75% of their hospitalization. Patients were not selected based on their specific psychiatric diagnosis. Supplementary medications were used in accordance with normal clinical care. Anticholinergic medications were used to treat emergent EPSs, but were not given prophylactically.
Patients were assessed at admission, weekly throughout their hospital stay, and at hospital discharge. These assessments were conducted by a psychiatrist (P.I.R.), a neurologist (M.F.M.), or a research nurse. Psychiatric symptoms were quantified using the Brief Psychiatric Rating Scale (BPRS), Hamilton Depression Rating Scale (HAM-D), and Hamilton Anxiety Rating Scale (HAM-A).
Acute dystonia was diagnosed if a patient experienced a sustained muscle contraction that required immediate treatment with benztropine.
Akathisia was assessed using the Barnes Akathisia Rating Scale. A global clinical impression of “mild” (2) was regarded as the cutoff point for the presence of akathisia.
Our parkinsonism scale was adapted from the motor examination section of the United Parkinson’s Disease Rating Scale, including items measuring facial and vocal expression, tremor, rigidity, bradykinesia (formally assessed using an alternate motion rate task), gait (including slowing, shuffling, arm swing, and turning), and writing. Each item was scored from 0 (“normal”) to 4 (“severely impaired”). Based on these assessments, a global clinical impression of parkinsonism was scored from 1 (“no parkinsonism”) to 4 (“severe parkinsonism”). Patients with a global impression score of 2 (“mild parkinsonism”) or greater were considered to have parkinsonism.
Dyskinesia was assessed using the Abnormal Involuntary Movement Scale. Patients met the diagnostic criteria for dyskinesia if they scored “moderate” (3) for movements of 1 body part or “mild” (2) for movements of 2 or more body parts.
We calculated percent improvement on psychiatric scales by subtracting the week 2 or discharge score from the baseline score and dividing by the baseline score. Because the minimum score on the BPRS is 18, we subtracted 18 from the raw scores before making this calculation.
Binary logistic regression was used to determine whether early BPRS improvement at week 2 predicted EPSs. Brief Psychiatric Rating Scale improvement was used as a continuous variable, without setting specific criteria for response or nonresponse. The presence of any EPSs at any point during hospitalization was the primary outcome measure. Age, sex, length of hospital stay, and maximum antipsychotic dose were included in the regression models. Patients who were missing week 2 assessments were not included in this analysis. Secondary analyses were conducted using specific EPSs (dystonia, parkinsonism, akathisia, or dyskinesia) as outcome measures. We also conducted a separate analysis excluding patients who experienced EPSs within the first 2 weeks of treatment to determine whether early BPRS improvement in this subgroup predicted EPSs beginning at later time points.
To assess the predictive power of early improvement in affective symptoms, percent improvement on HAM-D or HAM-A scores at week 2 was substituted for BPRS improvement in the regression model described previously using any EPSs during hospitalization as the outcome measure. Again, patients with missing HAM-D or HAM-A scores at week 2 were not included in these analyses.
Finally, we used linear regression accounting for age, sex, length of hospital stay, and maximum antipsychotic dose to confirm that early BPRS improvement at week 2 predicted BPRS improvement at hospital discharge. To compare patients included in the analysis to patients excluded because of missing BPRS scores, we used independent t tests for continuous variables and Fisher exact test for categorical variables.
Patients and Treatment
In total, 461 patients were formally assessed for eligibility, 86 of whom declined participation in the study and were not investigated further. An additional 106 patients were found to have been exposed to an antipsychotic prior to their first assessment and were excluded from the analysis. In the remaining group, 199 patients were treated primarily with haloperidol. Of these patients, 9 patients were discharged from hospital before week 2, and 54 others were missing BPRS assessments at week 2. These patients were not included in the analysis, leaving a final sample size of 136 patients. Characteristics of the patient sample are presented in Table 1. The patients excluded from the analysis because of missing week 2 assessments did not differ from the patients included in the analysis on any measure described in Table 1.
Haloperidol doses were generally low, with 72% of patients receiving a mean dose less than 5 mg/d. A number of patients treated with antipsychotics also received other medications during their hospital stay: antiparkinsonian agents in 82 patients, lithium in 47, benzodiazepines in 115, antidepressants in 38, and β-blockers in 20 patients. In addition, 21 patients used an antipsychotic other than haloperidol at some point during their hospitalization. While most patients treated with lithium received a diagnosis of bipolar disorder, lithium was also used in several patients with schizoaffective disorder or major depressive disorder with psychotic features.
Mean BPRS improvement was 55.6% (SEM, 3.33%) at week 2 and 77.1% (SEM, 2.46%) at hospital discharge. Patients also showed dramatic improvement on measures of affective symptoms. At week 2, mean improvement on the HAM-D was 46.0% (SEM, 5.36%), and mean improvement on the HAM-A was 45.8% (SEM, 6.90%). At hospital discharge, mean improvement on the HAM-D was 68.8% (SEM, 5.48%), and mean improvement on the HAM-A was 66.8% (SEM, 5.68%). This resulted in mean raw scores of 4.35 (SEM, 0.48) on the HAM-D and 3.37 (SEM, 0.46) on the HAM-A at hospital discharge. Assessments at hospital discharge were missing for 21 patients.
Rates of EPSs occurring during hospitalization are presented in Table 2. Of the 106 patients who experienced some form of EPSs, 85 experienced EPSs within the first 2 weeks of treatment. Among patients with no EPSs during the first 2 weeks, 21 patients went on to experience EPSs beginning later in treatment.
The first issue we investigated was whether early antipsychotic response (measured by BPRS percent improvement at week 2) predicted the occurrence of EPSs at any point during hospitalization. A logistic regression model including BPRS improvement at week 2, age, sex, length of hospitalization, and maximum antipsychotic dose revealed that greater antipsychotic response at week 2 was associated with decreased rates of EPSs (odds ratio [OR], 0.971; 95% confidence interval [CI], 0.951–0.991; P = 0.004). To determine whether early response continued to predict EPSs that began after week 2 of treatment, we performed an identical analysis excluding all patients with EPSs occurring within the first 2 weeks. In this model, greater BPRS improvement at week 2 remained a significant predictor of reduced EPSs (OR, 0.960; CI, 0.933–0.988; P = 0.005). This result indicates that greater early response continues to be associated with reduced EPS risk even in patients who have not experienced EPSs by week 2 of treatment. When we repeated this analysis excluding patients who received any antipsychotic other than haloperidol, strong early response continued to predict decreased rates of EPSs (OR, 0.970; CI, 0.949–0.991 P = 0.006), even when we additionally excluded all patients with EPSs occurring within the first 2 weeks (OR, 0.962; CI, 0.932–0.992; P = 0.014). Maximum antipsychotic dose was not a significant predictor of EPSs in either of these regression models. When mean antipsychotic dose or cumulative dose throughout hospitalization was substituted for maximum antipsychotic dose in the regression models, it still did not predict EPSs.
In secondary analyses, we used specific EPSs as outcome measures. These analyses demonstrated that greater BPRS improvement at week 2 was associated with a decreased incidence of parkinsonism (OR, 0.987; CI, 0.975–0.999; P = 0.028) and dyskinesia (OR, 0.985; CI, 0.972–0.998; P = 0.025), but not dystonia (OR, 1.000; CI, 0.990–1.011; P = 0.944) or akathisia (OR, 0.996; CI, 0.986–1.007; P = 0.492). Maximum antipsychotic dose was not a significant predictor of EPSs in any of these models. Comparisons of early improvement between patients who experienced EPSs and those who did not are illustrated in Figure 1. Increasing age was associated with decreased rates of dystonia (OR, 0.958; CI, 0.935–0.981; P < 0.001) and akathisia (OR, 0.964; CI, 0.943–0.985; P = 0.001).
While recognizing that the sample was small, we nevertheless explored whether the association between early response and EPSs held across age, sex, and diagnostic categories. The association was somewhat stronger in patients at least 40 years of age (OR, 0.947; CI, 0.910–0.986) compared with those younger than 40 years (OR, 0.982; CI, 0.959–1.007), in females (OR, 0.932; CI, 0.884–0.982) compared with males (OR, 0.985; CI, 0.963–1.007), and in patients with a primary diagnosis of bipolar mania (OR, 0.937; CI, 0.876–1.002) compared with those with other diagnoses (OR, 0.973; CI, 0.951–0.996), but none of these differences reached the level of statistical significance. Brief Psychiatric Rating Scale improvement at week 2 was also a significant predictor of BPRS improvement at hospital discharge in this patient sample (B = 0.226, R2 = 0.164, P < 0.001).
Early Improvement in Affective Symptoms
We next assessed whether early improvement in HAM-D and HAM-A scores predicted EPS risk. Logistic regression indicated that EPSs during hospitalization were not significantly predicted by week 2 HAM-A improvement (OR, 1.000; CI, 0.993–1.008; P = 0.929), but there was a trend toward significance for week 2 HAM-D improvement (OR, 0.988; CI, 0.973–1.002; P = 0.094). Early HAM-D and HAM-A improvements are compared between patients with and without EPSs in Figure 2.
In this study, we investigated whether early response was associated with EPS risk in a sample of previously antipsychotic-naive patients. Most patients experienced dramatic symptom improvement even at a relatively low dose of haloperidol, reflecting the striking effectiveness of antipsychotic medication, especially when it is being given for the first time. We found that a greater early antipsychotic response—assessed by week 2 BPRS improvement—was associated with a decreased incidence of EPSs, particularly parkinsonism and dyskinesia. Importantly, this predictive value continued to apply to EPSs that began after 2 weeks of treatment. We also found that dystonia and akathisia were more common in younger patients. Early HAM-A improvement did not predict EPSs, but there was a nonsignificant trend suggesting that greater HAM-D improvement at week 2 was associated with a decreased incidence of EPSs.
One possible interpretation of these results is that patients who experienced a poor early treatment response were subsequently treated with higher antipsychotic doses and developed EPSs as a result. However, antipsychotic dose was not a significant predictor of EPSs in any regression analysis we conducted. This suggests that the relationship between early antipsychotic response and EPSs reflects some aspect of the underlying illness rather than the specific treatment regimen. The lack of any association between antipsychotic dose and EPSs is curious, but a possible explanation is that when doses are generally kept low—as they were in this study—dosing variations within this safer range may make a relatively smaller contribution to the prediction of EPS risk.
Earlier work has associated poor clinical response with EPSs.22,23 However, no previous studies have demonstrated that early antipsychotic response can predict a variety of EPSs occurring later in treatment, thereby providing useful prognostic information that could guide clinical decision making. For instance, patients at high risk may benefit from the use of prophylactic anticholinergic medication even if they have not yet exhibited any signs of EPSs by week 2 of treatment. The clinical utility of such an approach should be evaluated in future research.
A unique advantage of the current study is the use of an entirely antipsychotic-naive patient sample. While others have taken steps toward describing the relationship between early antipsychotic response and EPS risk,2,4,13 these studies primarily investigated patients with prior antipsychotic exposure. The use of antipsychotic-naive patients is important for several reasons. First, antipsychotic-induced EPSs appear to be more common in first-episode psychosis patients and those with no prior antipsychotic exposure,16,17 so it is in this population that assessing EPS risk is most relevant. Second, in patients with prior antipsychotic exposure, treatment decisions will be guided by their experience with previous antipsychotics. In antipsychotic-naive patients, no such experience is available, so the prognostic value of early antipsychotic response is more valuable. Third, in patients who have been treated with antipsychotics before the first study assessment, it is unclear whether a true “baseline” has been established. This lack of a baseline limits the interpretation of early treatment response. The lack of antipsychotic-naive patients may help to explain the conflicting results of earlier studies.2,4
The generalizability of the current results is limited to some extent by the fact that every patient was treated with haloperidol. This approach was chosen to limit the variability associated with different antipsychotic drugs and because haloperidol was the antipsychotic most used in our study population during the time period in which these data were collected. However, because the predictive value of early response appears to differ between antipsychotics,3,14,15 it is unclear to what extent the results observed here can be applied to patients treated with other drugs. As well, there is substantial geographic variation in the activity of enzymes responsible for metabolizing haloperidol.24 As it pertains to the current results, this issue is of particular interest in low-income countries where the use of haloperidol may be more common. However, studies demonstrating that first-generation antipsychotics can be as effective as newer agents, and also that some second-generation antipsychotics are associated with more severe metabolic side effects,12 have led to renewed interest in haloperidol worldwide.
Overall, this study of antipsychotic-naive, first-episode psychosis patients naturalistically treated with haloperidol further demonstrates the value of assessing early response after 2 weeks of treatment. In addition to predicting psychiatric outcomes, early response also provides clinically useful information about the risk of EPSs, even in patients who do not exhibit these side effects within the first 2 weeks of treatment. This information has the potential to guide clinical decision making in order to minimize the incidence of EPSs in patients treated with antipsychotic medications.
AUTHOR DISCLOSURE INFORMATION
The authors declare no conflicts of interest.
1. Schennach-Wolff R, Jäger M, Mayr A, et al. Predictors of response and remission in the acute treatment of first-episode schizophrenia patients—is it all about early response? Eur Neuropsychopharmacol
2. Stauffer VL, Case M, Kinon BJ, et al. Early response to antipsychotic therapy as a clinical marker of subsequent response in the treatment of patients with first-episode psychosis. Psychiatry Res
3. Leucht S, Zhao J. Early improvement as a predictor of treatment response and remission in patients with schizophrenia: a pooled, post-hoc analysis from the asenapine development program. J Psychopharmacol
4. Kinon BJ, Chen L, Ascher-Svanum H, et al. Early response to antipsychotic drug therapy as a clinical marker of subsequent response in the treatment of schizophrenia. Neuropsychopharmacology
5. Miller DD, Caroff SN, Davis SM, et al. Extrapyramidal side-effects of antipsychotics in a randomised trial. Br J Psychiatry
6. Aguilar EJ, Keshavan MS, Martínez-Quiles MD, et al. Predictors of acute dystonia in first-episode psychotic patients. Am J Psychiatry
7. Rosebush PI, Mazurek MF. Neurologic side effects in neuroleptic-naive patients treated with haloperidol or risperidone. Neurology
8. Bobes J, Gibert J, Ciudad A, et al. Safety and effectiveness of olanzapine versus conventional antipsychotics in the acute treatment of first-episode schizophrenic inpatients. Prog Neuropsychopharmacol Biol Psychiatry
9. Tandon R, Jibson MD. Extrapyramidal side effects of antipsychotic treatment: scope of problem and impact on outcome. Ann Clin Psychiatry
10. Koreen AR, Siris SG, Chakos M, et al. Depression in first-episode schizophrenia. Am J Psychiatry
11. Seemüller F, Schennach R, Mayr A, et al. Akathisia and suicidal ideation in first-episode schizophrenia. J Clin Psychopharmacol
12. Lieberman J, Stroup TS, McEvoy JP, et al. Effectiveness of antipsychotic drugs in patients with chronic schizophrenia. N Engl J Med
13. Schennach-Wolff R, Meyer S, Seemüller F, et al. Influencing factors and predictors of early improvement in the acute treatment of schizophrenia and schizophrenia spectrum disorder. J Psychiatr Res
14. Rasmussen SA, Rosebush PI, Anglin SE, et al. The predictive value of early treatment response in antipsychotic-naive patients with first-episode psychosis: haloperidol versus olanzapine. Psychiatry Res
15. Hatta K, Otachi T, Sudo Y, et al. Difference in early prediction of antipsychotic non-response between risperidone and olanzapine in the treatment of acute-phase schizophrenia. Schizophr Res
16. Sanger T, Lieberman JA, Tohen M, et al. Olanzapine versus haloperidol treatment in first-episode psychosis. Am J Psychiatry
17. McEvoy JP, Hogarty GE, Steingard S. Optimal dose of neuroleptic in acute schizophrenia. A controlled study of the neuroleptic threshold and higher haloperidol dose. Arch Gen Psychiatry
18. Docx L, Morrens M, Bervoets C, et al. Parsing the components of the psychomotor syndrome in schizophrenia. Acta Psychiatr Scand
19. Chong S, Subramaniam M, Verma S. Spontaneous parkinsonism in antipsychotic-naive patients with first-episode psychosis. Can J Psychiatry
20. Majadas S, Olivares J, Galan J, et al. Prevalence of depression and its relationship with other clinical characteristics in a sample of patients with stable schizophrenia. Compr Psychiatry
21. Rybakowski JK, Vansteelandt K, Remlinger-Molenda A, et al. Extrapyramidal symptoms during treatment of first schizophrenia episode: results from EUFEST. Eur Neuropsychopharmacol
22. Emsley R, Oosthuizen P, Koen L, et al. Remission in patients with first-episode schizophrenia receiving assured antipsychotic medication: a study with risperidone long-acting injection. Int Clin Psychopharmacol
23. Chen L, Ascher-Svanum H, Lawson A, et al. Movement disorder profile and treatment outcomes in a one-year study of patients with schizophrenia. Neuropsychiatr Dis Treat
24. Sistonen J, Sajantila A, Lao O, et al. CYP2D6
worldwide genetic variation shows high frequency of altered activity variants and no continental structure. Pharmacogenet Genomics