- Email: [email protected]
Clinical assessment of lurasidone benefit and risk in the treatment of bipolar I depression using number needed to treat, number needed to harm, and likelihood to be helped or harmed
Journal of Affective Disorders 155 (2014) 20–27
Contents lists available at ScienceDirect
Journal of Affective Disorders journal homepage: www.elsevier.com/locate/jad
Review
Clinical assessment of lurasidone benefit and risk in the treatment of bipolar I depression using number needed to treat, number needed to harm, and likelihood to be helped or harmed Leslie Citrome a,n, Terence A. Ketter b, Josephine Cucchiaro c, Antony Loebel c a
New York Medical College, Valhalla, NY, USA Stanford University School of Medicine, Stanford, CA, USA c Sunovion Pharmaceuticals Inc., Fort Lee, NJ, USA b
art ic l e i nf o
a b s t r a c t
Article history: Received 20 September 2013 Accepted 22 October 2013 Available online 28 October 2013
Background: Prior to recent FDA approval of lurasidone for treatment of bipolar depression there were only two approved treatments for this condition (quetiapine and olanzapine–fluoxetine combination), and these were as likely to provide therapeutic benefit as adverse effects. We assessed the efficacy, safety, and tolerability of lurasidone for major depressive episodes associated with bipolar I disorder, using number needed to treat (NNT, for benefits), number needed to harm (NNH, for harms), and likelihood of being helped or harmed (LHH, ratio of NNH to NNT, for trade-offs between benefits vs. harms). Methods: Data was collected from two 6-week multicenter, randomized, double-blind, placebo-controlled, flexibly-dosed acute bipolar I depression studies, one using lurasidone monotherapy at 20–60 mg/d or 80–120 mg/d, and the other using lurasidone 20–120 mg/d adjunctive to lithium or valproate. The NNT or NNH was calculated for lurasidone vs. placebo for the following dichotomous outcomes: response (Z 50% reduction from baseline on Montgomery Asberg Depression Rating Scale (MADRS) total score); remission (final MADRS total score r12); discontinuation due to an adverse event (AE); weight gain Z 7% from baseline; incidence of spontaneously reported AEs; and incidence of total cholesterol Z 240 mg/dl, low-density lipoprotein cholesterol Z 160 mg/dl, fasting triglycerides Z200 mg/dl and glucose Z 126 mg/dl post-baseline. Results: NNT vs. placebo for response was 5 for lurasidone monotherapy (both dose ranges) and 7 for adjunctive therapy. NNT vs. placebo for remission for lurasidone monotherapy was 6 for 20–60 mg/d and 7 for 80–120 mg/d and 7 for adjunctive lurasidone. NNH vs. placebo for discontinuation due to an AE for lurasidone monotherapy was 642 for 20–60 mg/d and 181 for 80–120 mg/d, and for adjunctive lurasidone was 54 (negative NNH denotes an advantage for lurasidone). Lurasidone was not associated with any clinically meaningful mean weight or metabolic changes compared to placebo; NNH vs. placebo for weight gain Z7% was 29 for 20–60 mg/d and 5550 for 80–120 mg/d and 42 for adjunctive lurasidone. The three most frequently occurring AEs with the largest difference in incidence for lurasidone vs. placebo were nausea, akathisia, and somnolence, with NNH values for lurasidone vs. placebo ranging from 11 (nausea with lurasidone monotherapy 80–120 mg/d) to 130 (somnolence with lurasidone monotherapy 20–60 mg/d). LHH was substantially and consistently 4 1 (indicating benefit being more likely than harm) when contrasting response or remission vs. AEs or weight gain. Limitations: Additional studies, including longer-term and open-label, “real world” data is needed to confirm the results reported here. Conclusions: NNT, NNH, and LHH help quantify relative benefits (efficacy) and harms (side effects), thus placing lurasidone findings in research studies into clinical perspective. Lurasidone, compared to other treatments approved for bipolar depression, yielded comparable benefits (all had single-digit NNT vs. placebo for response or remission), and less risk of harm (double-digit or greater NNHs with lurasidone compared to single-digit NNHs for sedation with quetiapine and for Z 7% weight gain with olanzapine– fluoxetine combination), and thus a substantially more favorable LHH (4 or >>1) with lurasidone monotherapy and adjunctive therapy, compared to quetiapine and olanzapine–fluoxetine combination (LHH oor 1). & 2013 Elsevier B.V. All rights reserved.
Keywords: Bipolar I depression Efficacy Lurasidone Number needed to harm Number needed to treat Tolerability
n
Correspondence to: Leslie Citrome, 11 Medical Park Drive, Suite 106, Pomona, NY 10970, USA. Tel.: þ 1 845 362 2081; fax: þ1 845 362 8745. E-mail addresses: [email protected], [email protected] (L. Citrome).
0165-0327/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jad.2013.10.040
L. Citrome et al. / Journal of Affective Disorders 155 (2014) 20–27
21
Contents 1. 2. 3.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Efficacy outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Discontinuation due to adverse events. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Specific safety/tolerability outcomes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. Consideration of benefit vs. harm: LHH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Role of funding source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conflict of interest. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1. Introduction Lurasidone received approval by the United States Food and Drug Administration in June 2013 for the treatment of adults with major depressive episodes associated with bipolar I disorder (bipolar depression), as either a monotherapy or as adjunctive to lithium or valproate (Sunovion, 2013). Prior to lurasidone's approval, there were only two approved treatments for bipolar depression (namely quetiapine and olanzapine–fluoxetine combination), each similarly as likely to provide therapeutic vs. adverse effects (Ketter et al., 2011), with quetiapine being more problematic with respect to sedation/somnolence and olanzapine–fluoxetine combination being more problematic with respect to weight gain and metabolic problems (Ketter et al., 2011; Citrome, 2011a). Lurasidone is a second-generation antipsychotic agent that was originally approved in 2010 for the treatment of patients with schizophrenia (Sunovion, 2013; Citrome, 2011b). Although it is similar to other atypical antipsychotics in being a dopamine D2 and serotonin 5-HT2A receptor antagonist, lurasidone also exhibits high binding affinity for the serotonin 5-HT7 receptor (antagonist) and moderate affinity for the serotonin 5-HT1A (partial agonist) and noradrenaline α2c (antagonist) receptors. Lurasidone displays little to no appreciable affinity for histamine H1 and acetylcholine M1 receptors. Lurasidone's low affinity for histaminergic receptors could yield less risk of sedation, weight gain and related metabolic effects, as suggested by clinical trials in schizophrenia (Citrome, 2011b). Number needed to treat (NNT) and number needed to harm (NNH) are measures of effect size and indicate how many patients would need to be treated with one agent instead of the comparator in order to encounter one additional beneficial or harmful outcome of interest, respectively (Citrome, 2008; Citrome, 2011c). NNT is commonly used to denote potential desirable outcomes (benefits, i.e., therapeutic effects) that clinicians and patients would want to encounter and NNH is commonly used to denote potential undesirable outcomes (harms, i.e., adverse effects) that clinicians and patients would like to avoid. NNT and NNH offer clinicians ways of quantifying clinical relevance or “clinical significance.” Low NNTs (or NNHs) are evidenced when there are large differences between the interventions in question. For example, a NNT of 2 would be a very large effect size, as a difference can be observed after treating two patients with one of the interventions vs. the other. In contrast, a NNT of 50 would mean little difference between the two interventions because it would require treating fifty patients to encounter a difference in outcome. Potentially useful medications have a low NNT and a high NNH (i.e., benefit more likely than harm) when compared to other interventions. A rule of thumb is that single-digit NNTs vs.
21 21 22 22 23 23 25 25 26 26 26 27 27
placebo for efficacy measures suggest interventions that have potentially useful advantages (indeed, approved bipolar disorder treatments generally have such favorable NNTs), whereas doubledigit or higher NNHs vs. placebo for adverse outcomes suggest interventions that have potentially adequate tolerability (unfortunately, until recently, approved bipolar disorder treatments commonly failed to have such favorable NNHs). Likelihood of being helped or harmed (LHH) can be a useful way of integrating data regarding relative benefits and risks (Citrome and Kantrowitz, 2008; Citrome and Ketter, 2013; Straus, 2002). LHH is calculated using the following formula: LHH ¼ (1/NNT)/(1/NNH) ¼NNH/NNT, with a LHH 41 meaning help is more likely than harm, and a LHH less than one meaning the opposite. Choosing which NNH and NNT to use in calculating LHH requires careful consideration so that the outcomes being assessed are well matched and consistent with clinician's and patient's values and preferences (Citrome and Kantrowitz, 2008; Citrome and Ketter, 2013; Straus, 2002). Similar to a prior report on lurasidone for the treatment of schizophrenia (Citrome, 2012), we sought to describe the efficacy, safety, tolerability, and place in therapy of lurasidone in the treatment of major depressive episodes associated with bipolar I disorder, using the metrics of NNT, NNH, and LHH.
2. Methods Study data were analyzed post hoc from two 6-week, placebocontrolled studies of lurasidone that enrolled subjects with or without rapid cycling and without psychotic features – a monotherapy study where lurasidone was flexibly dosed at 20–60 mg/d or 80– 120 mg/d and an adjunctive therapy trial where flexibly dosed lurasidone (20–120 mg/d) was added to ongoing treatment with either lithium or valproate. The methods and principal results have been previously reported in detail (Loebel et al., 2013a; Loebel et al., 2013b). Table 1 provides an overview of the subjects in the two studies. Analogous to other NNT and NNH analyses (Citrome, 2012; Citrome et al., 2013), rates for the following dichotomous outcomes for each arm of each study were used: response (Z50% reduction from baseline on Montgomery Asberg Depression Rating Scale (MADRS) total score); remission (final MADRS total score of r12); discontinuation due to an adverse event (AE); weight gain Z7% from baseline; incidence of spontaneously reported AEs; incidence of shifts of total cholesterol from o240 mg/dL to Z240 mg/dL, low-density lipoprotein cholesterol from o160 mg/dL, Z160 mg/dL, fasting triglycerides from o200 mg/dL to Z200 mg/dL and glucose from o126 mg/dL to Z126 mg/dl, post-baseline. For any metabolic outcomes that
22
L. Citrome et al. / Journal of Affective Disorders 155 (2014) 20–27
Table 1 Phase III 6-week randomized controlled trials for lurasidone in patients with major depressive episodes associated with bipolar I disorder, baseline characteristics (ITT Population). Adjunctive lurasidone 20–120 mg/d vs. adjunctive placebo
Lurasidone monotherapy 20–60 mg/d or 80–120 mg/d vs. placebo
Lurasidone
Lurasidone 20–60 mg/d
Placebo
Randomized subjects, N 183 165 166 183 163 164 Safety population, Na ITT population, Nb 179 161 161 Male, n (%) 93 (52%) 85 (53%) 70 (43%) Age, years, mean 41.0 42.6 41.3 Race, n (%) White 108 (60%) 102 (63%) 107 (66%) Black 25 (14%) 16 (10%) 21 (13%) Asian 43 (24%) 37 (23%) 23 (14%) Other 3 (2%) 6 (4%) 10 (6%) Receiving lithium or valproate at baseline, Lithium, 90 (50%); valproate, 89 (50%) Lithium, 73 (46%); valproate, 87 (54%) NA n (%) MADRS total score, mean 30.6 30.8 30.3
Lurasidone 80–120 mg/d
Placebo
169 167 162 64 (40%) 42.0
170 168 162 75 (46%) 41.2
106 (65%) 25 (15%) 23 (14%) 8 (5%) NA
107 (66%) 21 (13%) 28 (17%) 6 (4%) NA
30.6
30.5
ITT – Intent-to-Treat. MADRS – Montgomery–Asberg Depression Rating Scale. NA – not applicable. a
Safety population consisted of all randomized subjects who received at least one dose of study medication. ITT population consisted of all subjects who were randomized, received at least one dose of study medication, and had a baseline and at least one post-baseline MADRS measure. b
suggested potentially relevant shifts to the clinically abnormal range, additional analyses were conducted to examine shifts from normal levels to high levels, and borderline levels to high levels. In calculating the proportions (n/N) of subjects for each treatment arm who met criteria for response or remission at endpoint (last observation carried forward (LOCF)), N, the denominator, was the Intent-to-Treat population (all subjects who were randomized, received at least one dose of study medication, and had a baseline and at least one post-baseline MADRS assessment. When reporting rates for AEs, N reflected the safety population (all randomized subjects who received at least one dose of study medication). For metabolic shift rates, N reflected the proportion with a baseline and at least one post-baseline assessment. NNT for efficacy outcomes and NNH for safety/tolerability outcomes were calculated for each study by first computing the difference in proportions between the experimental intervention and the control intervention for the outcome of interest (risk difference) and then computing the reciprocal of the risk difference and rounding up to the next highest whole number (Citrome, 2009). Calculating the 95% confidence interval (CI) for the NNT or NNH involved first calculating the 95% CI for the risk difference; the formula for this can be found elsewhere (Citrome, 2009). When the 95% CI for the NNT or NNH estimate contained infinity, the NNT or NNH estimate was not statistically significant at the P threshold of o0.05 and under those circumstances, instead of providing the 95% CI, the notation “ns” for not significant was made. All calculations for NNT and NNH were performed using the active medication as the experimental intervention and placebo as the control intervention, with the assumption that the study medication would be superior to placebo on efficacy outcomes but inferior to placebo on safety or tolerability outcomes. Thus for study medication compared to placebo, “negative” NNTs and “negative” NNHs denoted therapeutic disadvantages and tolerability advantages, respectively. Likelihood of being helped or harmed (LHH), the ratio of NNH to NNT, was calculated to illustrate trade-offs between benefits (response/remission) vs. harms (AEs), with LHH values 41 representing a greater likelihood of encountering benefit rather than harm. To place lurasidone into clinical context using outcomes generally accepted as relevant in this type of analysis, NNT for
response and remission, NNH for weight gain Z7% from baseline, NNH for discontinuation because of an AE, NNH for commonly encountered AEs, and NNH for shifts for metabolic variables, were calculated wherever possible from published, 8-week, randomized controlled registration trials and product labeling for the other products approved for the treatment of bipolar depression: quetiapine immediate and extended-release (Calabrese et al., 2005; Thase et al., 2006; Suppes et al., 2010; AstraZeneca, 2011a, 2011b) and olanzapine–fluoxetine combination (Tohen et al., 2003; Eli Lilly, 2013). For the lurasidone monotherapy study, NNT for response and remission and LHH were calculated both separately for the 20–60 mg/d and 80–120 mg/d dose groups for the main analysis, and in aggregate to yield a pooled (20–120 mg/d) dose group for selected additional analyses.
3. Results 3.1. Efficacy outcomes Table 2 provides the rates for response and remission and the NNT vs. placebo for the different treatment arms. Lurasidone demonstrated statistically significant NNT advantages vs. placebo for both response and remission. NNT vs. placebo for response was 5 for lurasidone monotherapy (both dose ranges) and 7 for adjunctive therapy; NNT vs. placebo for remission for lurasidone monotherapy was 6 for 20–60 mg/d and 7 for 80–120 mg/d and 7 for adjunctive lurasidone (see Table 2 for the 95% CIs). For the monotherapy lurasidone study, response and remission rates for the pooled lurasidone dose arms were 168/323 (52.0%) and 132/ 323 (40.9%), respectively, yielding NNTs vs. placebo of 5 (95% CI 4–8), and 7 (95% CI 4–14), respectively. Figs. 1 and 2 graphically display the NNT vs. placebo for response and remission for the pooled doses of lurasidone monotherapy, together with data calculated from the 8-week, placebo-controlled, US registration trials of quetiapine (pooled data from Calabrese et al., 2005; Thase et al., 2006; Suppes et al., 2010) and olanzapine–fluoxetine combination (Tohen et al., 2003) for bipolar depression.
L. Citrome et al. / Journal of Affective Disorders 155 (2014) 20–27
23
Table 2 Response, remission, and number needed to treat vs. placebo (ITT population).
a
ITT population, N Responders, n (%)b NNT, response vs. placebo (95% CI) Remitters, n (%)c NNT, remission vs. placebo (95% CI)
Adjunctive lurasidone 20–120 mg/d vs. adjunctive Placebo
Lurasidone monotherapy 20–60 mg/d or 80–120 mg/d vs. placebo
Lurasidone
Placebo
Lurasidone 20–60 mg/d
Lurasidone 80–120 mg/d
Placebo
179 102 (57.0%) 7 (4–24) 90 (50.3%) 7 (4–23)
161 68 (42.2%) NA 57 (35.4%) NA
161 86 (53.4%) 5 (3–8) 68 (42.2%) 6 (4–14)
162 82 (50.6%) 5 (4–11) 64 (39.5%) 7 (4–21)
162 49 (30.2%) NA 40 (24.7%) NA
CI – confidence interval. ITT – Intent-to-Treat. MADRS – Montgomery–Asberg Depression Rating Scale. NA – not applicable. NNT – number needed to treat. a ITT population consisted of all subjects who were randomized, received at least one dose of study medication, and had a baseline and at least one post-baseline MADRS measure. b Proportion of subjects who achieved a response, defined as Z 50% reduction from baseline total score on the MADRS. c Proportion of subjects who achieved remission, defined as a MADRS total score of r 12 at endpoint.
Fig. 1. MADRS responders with lurasidone monotherapy, quetiapine immediate and extended release monotherapy, and olanzapine/fluoxetine combination – number needed to treat and 95% confidence interval vs. placebo (all doses for the respective agents pooled). IR – immediate release. MADRS – Montgomery–Asberg Depression Rating Scale. NNT – number needed to treat. XR – extended release. (Proportion of subjects who achieved a response, defined as Z 50% reduction from Baseline total score on the MADRS).
Fig. 2. MADRS remitters with lurasidone monotherapy, quetiapine immediate and extended release monotherapy, and olanzapine/fluoxetine combination – number needed to treat and 95% confidence interval vs. placebo (all doses for the respective agents pooled). IR – immediate release. MADRS – Montgomery–Asberg Depression Rating Scale. NNT – number needed to treat. XR – extended release. (Proportion of subjects who achieved remission, defined as a MADRS total score of r12 at last observation carried forward endpoint).
3.2. Discontinuation due to adverse events
baseline Z 7%, and shifts on fasting plasma glucose, fasting plasma triglycerides, and all available data for total cholesterol and low density lipoprotein (LDL). For adjunctive lurasidone vs. placebo, the incidence of weight gain Z7% from baseline was 3.1% vs. 0.7%, yielding a NNH of 42 (ns). For lurasidone monotherapy 20–60 mg/ d, the incidence of weight gain Z7% from baseline was 4.2% vs. 0.7% for placebo yielding a NNH of 29 (95% CI 15–52318). For lurasidone monotherapy 80–120 mg/day vs. placebo, the corresponding rates were essentially identical (0.7%), yielding a nonstatistically significant (and clinically irrelevant) NNH of 5550. However, for olanzapine–fluoxetine combination, weight gain Z7% from baseline was observed in 19.5% of olanzapine–fluoxetine combination subjects vs. 0.3% for placebo, yielding a NNH of 6 (95% CI 4–10) (Eli Lilly, 2013), on par with the effect size observed for efficacy of olanzapine–fluoxetine combination. For quetiapine, at all doses and all formulations, approximately 8.4% of subjects receiving quetiapine gained Z 7% of their baseline weight vs. approximately 1.9% for placebo in studies of bipolar depression (AstraZeneca, 2011a, 2011b), resulting in a NNH of 16 (95% CI 12–25). When examining shift in metabolic variables in the studies for lurasidone, the only potential signal for abnormalities was shift in fasting triglycerides to Z200 mg/dL postbaseline where lurasidone monotherapy at both doses exhibited statistically significant NNH values of 14 and 15, for lurasidone 20–60 mg/d and 80–120 mg/d, respectively. However, this appears
Table 3 provides the NNH for discontinuation due to an AE. Discontinuation due to an AE was not statistically significant different from placebo for lurasidone monotherapy 20–60 mg/ day, 80–120 mg/day, or for adjunctive lurasidone. Calculation of the NNH resulted in a negative number (a negative NNH denotes an advantage for lurasidone regarding the potential harm) for adjunctive lurasidone vs. placebo, lurasidone monotherapy 80– 120 mg/d vs. placebo, and the combined dosage arms for lurasidone monotherapy (NNH 496, ns). NNH for lurasidone monotherapy 20–60 mg/d vs. placebo was 642 (ns). In contrast, the analogous discontinuation rates for quetiapine (all doses all formulations) were 126/841 (15.0%) for active medication vs. 20/487 (4.1%) for placebo (Calabrese et al., 2005; Thase et al., 2006; Suppes et al., 2010), resulting in a NNH of 10 (95% CI 8–13). For the olanzapine–fluoxetine combination pivotal studies, a higher proportion discontinued because of an AE with placebo (19/377, 5.0%), than with olanzapine–fluoxetine combination (2/86, 2.3%) (Tohen et al., 2003), resulting in a NNH of 37 (ns). 3.3. Specific safety/tolerability outcomes Table 4 provides the rates and NNH for clinically relevant weight gain, as defined by an increase in body weight from
24
L. Citrome et al. / Journal of Affective Disorders 155 (2014) 20–27
Table 3 Rates and number needed to harm vs. placebo for discontinuation due to an adverse event (randomized population). Adjunctive lurasidone 20–120 mg/d vs. adjunctive placebo
Lurasidone monotherapy 20–60 mg/d or 80–120 mg/d vs. placebo
Lurasidone
Placebo
Lurasidone 20–60 mg/d
Lurasidone 80–120 mg/d
Placebo
165 13 (7.9%) NA
166 11 (6.6%) 642 (ns)
169 10 (5.9%) 181 (ns)
170 11 (6.5%) NA
Randomized, N 183 Discontinuation due to AE, n (%) 11 (6.0%) NNH, discontinuation due to an AE vs. placebo (95% CI) 54 (ns)
a
a
AE – adverse event. CI – confidence interval. NA – not applicable. NNT – number needed to treat. NNH – number needed to harm. ns – not significant (the 95% CI contains “infinity”). a
A negative NNH denotes an advantage for lurasidone regarding the potential harm.
Table 4 Rates and number needed to harm vs. placebo for metabolic variables (Safety population).
Safety Population, Na Weight increase Z7% from baseline, n (%) NNH (95% CI) Shift fasting glucose Z 126 mg/dL (post-baseline) NNH (95% CI) Shift fasting triglycerides Z 200 mg/dL (post-baseline) NNH (95% CI) Shift fasting triglycerides o 150 to Z200 mg/dL (post-baseline) NNH (95% CI) Shift fasting triglycerides Z 150 and o 200 to Z 200 mg/dL(post-baseline) NNH (95% CI) Shift total cholesterol Z240 mg/dL (post-baseline) NNH (95% CI) Shift LDL cholesterol Z160 mg/dL (post-baseline) NNH (95% CI)
Adjunctive lurasidone 20– 120 mg/d vs. adjunctive placebo
Lurasidone monotherapy 20–60 mg/d or 80–120 mg/d vs. placebo
Lurasidone
Placebo
Lurasidone 20–60 mg/d
Lurasidone 80–120 mg/d
Placebo
183 5/162 (3.1%) 42 (ns) 0/138 (0%) 64 (ns) b 14/118 (11.9%) 28 (ns) 9/98 (9.2%) 23 (ns) 5/20 (25.0%) 18 (ns) 7/135 (5.2%) 83 (ns) 6/144 (4.2%) 58 (ns)
163 1/150 (0.7%) NA 2/127 (1.6%) NA 9/109 (8.3%) NA 4/83 (4.8%) NA 5/26 (19.2%) NA 5/126 (4.0%) NA 3/124 (2.4%) NA
164 6/143 (4.2%) 29 (15–52318) 3/123 (2.4%) 136 (ns) b 12/108 (11.1%) 14 (7–129) 7/87 (8.0%) 20 (ns) 5/21 (23.8%) 6 (ns) 5/113 (4.4%) 534 (ns) 5/121 (4.1%) 108 (ns)
167 1/147 (0.7%) 5550 (ns) 7/130 (5.4%) 46 (ns) 12/115 (10.4%) 15 (8–252) 5/91 (5.5%) 41 (ns) 7/24 (29.2%) 5 (ns) 5/114 (4.4%) 673 (ns) 9/126 (7.1%) 26 (ns)
168 1/151 (0.7%) NA 4/126 (3.2%) NA 4/114 (3.5%) NA 3/100 (3.0%) NA 1/14 (7.1%) NA 5/118 (4.2%) NA 4/125 (3.2%) NA
CI – confidence interval. NA – not applicable. NNH – number needed to harm. ns – not significant (the 95% CI contains “infinity”). a b
Safety population consisted of all randomized subjects who received at least one dose of study medication. A negative NNH denotes an advantage for lurasidone regarding the potential harm.
to be largely driven by subjects with a borderline fasting triglyceride level of Z150 mg/dL and o200 mg/dL at study baseline (see Table 4). Equivalent shift data for olanzapine–fluoxetine combination in bipolar depression were not reported in the olanzapine–fluoxetine combination clinical trial as published (Tohen et al., 2003) or in product labeling (Eli Lilly, 2013), however olanzapine's lipid-altering profile is well known (Citrome et al., 2011). Limited shift data are available for quetiapine for bipolar depression from product labeling (AstraZeneca, 2011a, 2011b) where for all doses and all formulations, NNH vs. placebo for shifts in triglycerides Z200 mg/dL is 23 (95% CI 12–420), and for total cholesterol Z240 mg/dL, 29 (95% CI 15–396). Table 5 outlines the incidence of the most commonly encountered spontaneously reported AEs reported in the clinical trials (incidence at least 5% in any treatment arm). However, for many of these AEs, the difference from placebo was small. The three most frequently occurring AEs with the largest difference in incidence for lurasidone vs. placebo were nausea, akathisia, and
somnolence, with NNH values for lurasidone vs. placebo ranging from 11 (nausea with lurasidone monotherapy 80–120 mg/d) to 130 (somnolence with lurasidone monotherapy 20–60 mg/d). In contrast to the “double-digit” NNH values for lurasidone, olanzapine–fluoxetine combination and quetiapine, in their respective registration studies, demonstrated clinically relevant NNH values vs. placebo in the “single digits.” For example, for the studies of quetiapine in bipolar depression, regardless of formulation or dose, NNH for somnolence (includes hypersomnia, sedation and somnolence) was 3 (rounded up from 2.4), with an exceedingly narrow 95% CI of 2.2–2.7 (calculated from AstraZeneca, 2011a, 2011b). The AE of somnolence alone from pooled clinical trial data was observed in 234/835 (28.0%) subjects receiving quetiapine vs. 31/487 (6.4%) for placebo (Calabrese et al., 2005; Thase et al., 2006; Suppes et al., 2010), yielding a NNH of 5 (95% CI 4–6). The data for sedation with quetiapine were similar. Dry mouth was also common with quetiapine vs. placebo (42.5% vs. 11.1%), with a NNH of 4 (95% CI 3–4). Other statistically significant NNH values
L. Citrome et al. / Journal of Affective Disorders 155 (2014) 20–27
25
Table 5 Rates and number needed to harm vs. placebo for common adverse events (Safety population). Adjunctive lurasidone 20–120 mg/d vs. adjunctive placebo
Lurasidone monotherapy 20–60 mg/d or 80–120 mg/d vs. placebo
Lurasidone
Placebo
Lurasidone 20–60 mg/d
Lurasidone 80–120 mg/d
Placebo
163 18 (11.0%) NA 9 (5.5%) NA 16 (9.8%) NA 7 (4.3%) NA 4 (2.5%) NA 9 (5.5%) NA 11 (6.7%) NA 20 (12.3%) NA 7 (4.3%) NA 5 (3.1%) NA
164 17 (10.4%) 39 (ns) 12 (7.3%) 130 (ns) 8 (4.9%) 40 (ns) 13 (7.9%) 18 (10–124) 4 (2.4%) 154 (ns) 8 (4.9%) 29 (ns) b 8 (4.9%) 33 (ns) 23 (14.0%) 48 (ns) 4 (2.4%) 19 ( 10 to 162) 10 (6.1%) 52 (ns)
167 29 (17.4%) 11 (6–39) 23 (13.8%) 14 (8–126) 15 (9.0%) 16 (9–60) 18 (10.8%) 12 (8–32) 10 (6.0%) 24 (12–1190) 11 (6.6%) 58 (ns) b 5 (3.0%) 83 (ns) 15 (9.0%) 35 (ns) b 10 (6.0%) 58 (ns) b 6 (3.6%) 175 (ns) b
168 13 (7.7%) NA 11 (6.5%) NA 4 (2.4%) NA 4 (2.4%) NA 3 (1.8%) NA 14 (8.3%) NA 3 (1.8%) NA 20 (11.9%) NA 13 (7.7%) NA 7 (4.2%) NA
a
183 Safety population, N Nausea, n (%) 32 (17.5%) NNH (95% CI) 16 (ns) Somnolence (hypersomnia, sedation, somnolence), n (%) 20 (10.9%) NNH (95% CI) 19 (ns) Extrapyramidal symptomsc, n (%) 28 (15.3%) NNH (95% CI) 19 (ns) Akathisia, n (%) 14 (7.7%) NNH (95% CI) 30 (ns) Vomiting, n (%) 9 (4.9%) NNH (95% CI) 41 (ns) Insomnia, n (%) 13 (7.1%) NNH (95% CI) 64 (ns) Diarrhea, n(%) 8 (4.4%) NNH (95% CI) 43 (ns) b Headache, n (%) 19 (10.4%) NNH (95% CI) 53 (ns) b Dizziness, n (%) 7 (3.8%) NNH (95% CI) 214 (ns) b Dry mouth, n (%) 5 (2.7%) NNH (95% CI) 299 (ns) b
b
CI – confidence interval. NA – not applicable. NNH – number needed to harm. ns – not significant (the 95% CI contains “infinity”). a b c
Safety population consisted of all randomized subjects who received at least one dose of study medication. A negative NNH denotes an advantage for lurasidone regarding the potential harm. Term includes cogwheel rigidity, drooling, dystonia, glabellar reflex abnormal, muscle rigidity, oromandibular dystonia, parkinsonism, tremor, trismus, torticollis.
were obtained when comparing quetiapine with placebo on dizziness (16.8% vs. 8.0%, NNH 12 (95% CI 9–19)), constipation (9.9% vs. 4.5%, NNH 19 (95% CI 13–38)), AEs associated with extrapyramidal syndrome (includes akathisia, dystonia, extrapyramidal disorder, hypertonia, and tremor) (8.6% vs. 3.3%, NNH 19 (95% CI 13–35)), and fatigue (9.6% vs. 6.0%, NNH 28 (95% CI 16– 138) (Calabrese et al., 2005; Thase et al., 2006; Suppes et al., 2010). For olanzapine–fluoxetine combination, “single digit” NNH values vs. placebo were observed for the AE of weight gain (17.4% vs. 2.7%, NNH 7 (95% CI 5–16) and diarrhea (18.6% vs. 6.6%, NNH 9 (95% CI 5–30)) (Eli Lilly, 2013). Other statistically significant NNH values were obtained when comparing olanzapine–fluoxetine combination with placebo on dry mouth (16.3% vs. 6.1%, NNH 10 (95% CI 6– 50)), asthenia (12.88 vs. 3.2%, NNH 11 (95% CI 6–43)), increased appetite (12.8% vs. 5.0%, NNH 13 (95% CI 7–282)), and tremor (9.3% vs. 2.4%, NNH 15 (95% CI 8–171)) (Eli Lilly, 2013). 3.4. Consideration of benefit vs. harm: LHH For lurasidone in the treatment of bipolar I depression, LHH was substantially and consistently 4 1 (indicating benefit being more likely than harm) when contrasting response or remission vs. AEs (see Table 6). For example, LHH for response vs. somnolence was Z2.8 for lurasidone monotherapy 80–120 mg/d, meaning that lurasidone high dose monotherapy was at least 2.8 times more likely to result in a reduction of 50% from the baseline MADRS total score than a spontaneously reported AE of somnolence. This is contrast to quetiapine where the LHH for response vs. somnolence was 3/6, or 0.5, meaning that quetiapine was only half as likely to result in a reduction of 50% from the baseline MADRS total score as a spontaneously reported AE of somnolence. For weight gain Z 7% from baseline, the NNH vs. placebo for lurasidone was statistically significant only for monotherapy within the dose range of 20–60 mg/d, with a LHH for response vs. weight gain
of 5.8. In contrast, this same metric for olanzapine–fluoxetine combination when comparing response or remission vs. weight gain, LHH was very close to 1.0, meaning for olanzapine–fluoxetine combination in the treatment of bipolar I depression response or remission was equally likely as clinically significant weight gain.
4. Discussion For patients with bipolar I depression, we found the therapeutic effects of lurasidone (monotherapy or adjunctive to lithium or valproate) vs. placebo resulted in single digit NNT values (5–7 for response, 6–7 for remission). These were comparable to the singledigit NNT values for the previously approved bipolar depression treatments quetiapine (6 for response, 6 for remission) and olanzapine–fluoxetine combination (4 for response, 5 for remission). Of note, the single-digit NNT for adjunctive lurasidone was evident even though the comparison was with adjunctive placebo added to a standing regimen of either lithium or valproate, either of which may have (albeit modest) potential therapeutic benefits for the treatment of acute bipolar depression (Ketter, 2010). Unapproved agents such as lamotrigine monotherapy and adjunctive antidepressants are commonly used to treat acute bipolar depression. However, evidence regarding their efficacy profiles for this indication is not encouraging. For example, for lamotrigine vs. placebo, NNT for response in bipolar depression is 12, and that for antidepressants vs. placebo, NNT is 29 (calculated from Sidor and Macqueen, 2011). Aripiprazole and ziprasidone have been tested in clinical trials for the treatment of bipolar depression but have not demonstrated efficacy, with NNT values for response vs. placebo of 44 (Ketter, 2010) and 163 (calculated from Lombardo et al., 2012). In addition to a favorable NNT for efficacy for lurasidone, we found that tolerability outcome differences for lurasidone vs. placebo generated NNH values in the double or triple digits (14–130 for
26
L. Citrome et al. / Journal of Affective Disorders 155 (2014) 20–27
Table 6 Likelihood of being helped or harmed: Response vs. nausea, akathisia, somnolence, extrapyramidal symptoms, and weight gain Z7%. Adverse event
Adjunctive
Monotherapy
Lurasidone
Lurasidone 20–60 mg/d
Lurasidone 80–120 mg/d
Monotherapy groups combined
NNT
NNT Nausea Akathisia Somnolence Extrapyramidal symptomsb Weight gain Z 7%
7 7 7 7 7
NNH a
16 30a 19a 19a 42a
LHH 2.3 4.3 2.7 2.7 6.0
5 5 5 5 5
NNH
LHH
NNT
NNH
LHH
NNT
NNH
LHH
a
7.8 3.6 26.0 8.0 5.8
5 5 5 5 5
11 12 14 16 5550a
2.2 2.4 2.8 3.2 1110
5 5 5 5 5
17 15 25a 22 58a
3.4 3.0 5.0 4.4 11.6
39 18 130a 40a 29
LHH – likelihood of being helped or harmed; calculated by NNH/NNT. NNH – number needed to harm. NNT – number needed to treat. a b
Not statistically significant; interpret the resultant LHH with caution. Term includes cogwheel rigidity, drooling, dystonia, glabellar reflex abnormal, muscle rigidity, oromandibular dystonia, parkinsonism, tremor, trismus, torticollis.
somnolence, 29–5550 for Z7% weight gain). These favorable NNH values were similar for lurasidone monotherapy and for adjunctive lurasidone when combined with lithium or valproate. These NNH values were substantially more favorable than the single-digit NNHs for the previously approved bipolar depression treatments quetiapine (NNH for somnolence, 3) and olanzapine–fluoxetine combination (NNH for Z7% weight gain, 6). The above-mentioned pattern of benefits relative to harms yielded LHH values that were consistently more favorable for lurasidone in contrast to the previously approved treatments for bipolar depression. For example, for lurasidone high dose monotherapy, LHH for response vs. somnolence was 2.8, which compared favorably with the analogous LHH for quetiapine which was 0.5. Lurasidone also demonstrated LHH values substantially 41.0 (range was 5.8 to 1100) when contrasting response or remission with weight gain Z7%. In contrast this same metric for olanzapine–fluoxetine combination was 1. Clinically, additional considerations include the time to onset of the AE vs. time to onset of a therapeutic response, as well as the severity and duration of the AE. The AE in question may be easily manageable if it is non-serious and short-lived. With respect to metabolic outcomes, no clinically relevant adverse safety signal was evident for lurasidone in the examination of clinically relevant weight gain or laboratory abnormalities; however those with preexisting abnormalities require careful monitoring. Despite their tolerability challenges, olanzapine–fluoxetine combination and quetiapine monotherapy may still have utility in highurgency situations, particularly in persons who have demonstrated good outcomes with these interventions in the past, where a pressing clinical need for efficacy mitigates their tolerability shortcomings. In addition, despite their efficacy challenges, the older unapproved treatments, specifically lamotrigine and antidepressants, may have utility in low-urgency situations, in which a compelling need for tolerability might mitigate their efficacy shortcomings. Lurasidone may ultimately prove to have utility in a broad spectrum of situations, independent of the degree of urgency, because of evidence suggesting not only adequate efficacy, but also adequate tolerability. Limitations: study populations and durations differed in the clinical trials for lurasidone, quetiapine and olanzapine–fluoxetine combination. The lurasidone trials were 6-weeks in duration and included only bipolar I disorder patients and excluded those with psychotic features. The quetiapine trials included subjects with both bipolar I disorder and bipolar II disorder and were 8-weeks in duration. The olanzapine–fluoxetine combination trials were also 8-weeks in duration and included subjects with bipolar I disorder (some of whom had psychotic symptoms), but did not include patients with bipolar II disorder.
A major limitation of NNT and NNH is that these metrics are limited to dichotomous outcomes. Other effect size measures are necessary when describing continuous outcomes without dichotomization (Citrome, 2010). Data were obtained from carefully structured studies and hence the results may not be generalizable to patients outside the confines of clinical trials. Long-term efficacy, safety and tolerability, including maintenance of response are also important considerations in the selection of treatments for bipolar depression. Dichotomous outcomes from long-term trials, such as relapse and weight gain exceeding a predefined threshold, will be of interest in further appraising the clinical profile of lurasidone.
5. Conclusions Lurasidone has recently received approval for the treatment of bipolar I depression not only as monotherapy but also as adjunctive (added to lithium or valproate) therapy. NNT, NNH, and LHH can help quantify relative benefits (efficacy) and harms (side effects), thus placing lurasidone findings in research studies into clinical perspective. Compared to the other approved bipolar depression treatments (quetiapine, olanzapine–fluoxetine combination), lurasidone yielded comparable benefits (all had single-digit NNTs vs. placebo for response or remission), less risk of harm (double-digit or greater NNHs with lurasidone compared to single-digit NNHs for somnolence with quetiapine and for weight gain Z7% with olanzapine–fluoxetine combination), and a more favorable ratio of benefit to harm likelihood (LHH much 41 with lurasidone compared to close to one with quetiapine and olanzapine–fluoxetine combination).
Role of funding source This study was supported Sunovion Pharmaceuticals Inc., Marlborough, MA and Fort Lee, NJ.
Conflict of interest In the past 36 months L. Citrome has engaged in collaborative research with, or received consulting or speaking fees, from Alexza, Alkermes, AstraZeneca, Avanir, Bristol-Myers Squibb, Eli Lilly, Envivo, Forest, Genentech, Janssen, Lundbeck, Merck, Mylan, Novartis, Noven, Otsuka, Pfizer, Reckitt Benckiser, Reviva, Shire, Sunovion, Takeda, and Valeant. In the past 36 months T. Ketter has engaged in collaborative research with, or received consulting or speaking fees, from Abbott, Allergan, AstraZeneca, Avanir, Bristol-Myers Squibb, Cephalon, Eli Lilly, Forest, GlaxoSmithKline, Janssen, Otsuka, Pfizer, Sunovion, and Teva; T. Ketter's spouse is an employee of and holds stock in Janssen Pharmaceuticals. J. Cucchiaro and A. Loebel are fulltime employees of Sunovion Pharmaceuticals.
L. Citrome et al. / Journal of Affective Disorders 155 (2014) 20–27
Acknowledgment This study was supported Sunovion Pharmaceuticals Inc., Marlborough, MA and Fort Lee, NJ.
References AstraZeneca, 2011a. Seroquel (quetiapine fumarate). Revised December 2011. Available from: 〈http://www1.astrazeneca-us.com/pi/Seroquel.pdf〉 accessed 28.01.13. AstraZeneca, 2011b. Seroquel XR (quetiapine fumarate extended release). Revised December 2011. Available from: 〈http://www1.astrazeneca-us.com/pi/sero quelxr.pdf〉 accessed 28.01.13. Calabrese, J.R., Keck Jr., P.E., Macfadden, W., Minkwitz, M., Ketter, T.A., Weisler, R.H., Cutler, A.J., McCoy, R., Wilson, E., Mullen, J., 2005. A randomized, double-blind, placebo-controlled trial of quetiapine in the treatment of bipolar I or II depression. Am. J. Psychiatry 162, 1351–1360. Citrome, L., 2008. Compelling or irrelevant? Using number needed to treat can help decide. Acta Psychiatr. Scand. 117, 412–419. Citrome, L., Kantrowitz, J., 2008. Antipsychotics for the treatment of schizophrenia: likelihood to be helped or harmed, understanding proximal and distal benefits and risks. Expert Rev. Neurother. 8, 1079–1091. Citrome, L., 2009. Quantifying risk: the role of absolute and relative measures in interpreting risk of adverse reactions from product labels of antipsychotic medications. Curr. Drug Saf. 4, 229–237. Citrome, L., 2010. Relative vs. absolute measures of benefit and risk: what is the difference? Acta Psychiatr. Scand. 121, 94–102. Citrome, L., 2011a. Olanzapine–fluoxetine combination for the treatment of bipolar depression. Expert Opin. Pharmacother. 12, 2751–2758. Citrome, L., 2011b. Lurasidone for schizophrenia: a review of the efficacy and safety profile for this newly approved second-generation antipsychotic. Int. J. Clin. Pract. 65, 189–210. Citrome, L., 2011c. Number needed to treat: what it is and what it is not, and why every clinician should know how to calculate it. J. Clin. Psychiatry 72, 412–413. Citrome, L., Holt, R.I., Walker, D.J., Hoffmann, V.P., 2011. Weight gain and changes in metabolic variables following olanzapine treatment in schizophrenia and bipolar disorder. Clin. Drug Investig. 31, 455–482. Citrome, L., 2012. Lurasidone for the acute treatment of adults with schizophrenia: what is the number needed to treat, number needed to harm, and likelihood to be helped or harmed? Clin. Schizophr. Relat. Psychoses 6, 76–85. Citrome, L., Ketter, T.A., 2013. When does a difference make a difference? Interpretation of number needed to treat, number needed to harm, and likelihood to be helped or harmed. Int. J. Clin. Pract. 67, 407–411.
27
Citrome, L., Goldberg, J.F., Portland, K.B., 2013. Placing transdermal selegiline for major depressive disorder into clinical context: number needed to treat, number needed to harm, and likelihood to be helped or harmed. J. Affect. Disord. 151, 409–417. Eli Lilly, 2013. Symbyax (olanzapine and fluoxetine hydrochloride). Revised January 2013. Available from: 〈http://pi.lilly.com/us/symbyax-pi.pdf〉 accessed 28.01.13. Ketter, T.A., 2010. Handbook of Diagnosis and Treatment of Bipolar Disorder. American Psychiatric Publishing, Inc., Washington, DC. Ketter, T.A., Citrome, L., Wang, P.W., Culver, J.L., Srivastava, S., 2011. Treatments for bipolar disorder: can number needed to treat/harm help inform clinical decisions? Acta Psychiatr. Scand. 123, 175–189. Loebel, A., Cucchiaro, J., Silva, R., Kroger, H., Sarma, K., Xu, J., Calabrese, J.R., 2013a. Lurasidone as adjunctive therapy with lithium or valproate for the treatment of bipolar I depression: a randomized, double-blind, placebo-controlled study. Am. J. Psychiatry. (in press). Loebel, A., Cucchiaro, J., Silva, R., Kroger, H., Hsu, J., Sarma, K., Sachs, G., 2013b. Lurasidone monotherapy in the treatment of bipolar I depression: a randomized, double-blind, placebo-controlled study. Am. J. Psychiatry. (in press). Lombardo, I., Sachs, G., Kolluri, S., Kremer, C., Yang, R., 2012. Two 6-week, randomized, double-blind, placebo-controlled studies of ziprasidone in outpatients with bipolar I depression: did baseline characteristics impact trial outcome? J. Clin. Psychopharmacol. 32, 470–478. Sidor, M.M., Macqueen, G.M., 2011. Antidepressants for the acute treatment of bipolar depression: a systematic review and meta-analysis. J. Clin. Psychiatry 72, 156–167. Straus, S.E., 2002. Individualizing treatment decisions. The likelihood of being helped or harmed. Eval. Health Prof. 25, 210–224. Sunovion, 2013. Latuda (lurasidone). Revised July 2013. Available from: 〈http:// www.latuda.com/LatudaPrescribingInformation.pdf〉 accessed 07.08.13. Suppes, T., Datto, C., Minkwitz, M., Nordenhem, A., Walker, C., Darko, D., 2010. Effectiveness of the extended release formulation of quetiapine as monotherapy for the treatment of acute bipolar depression. J. Affect. Disord. 121, 106–115. Thase, M.E., Macfadden, W., Weisler, R.H., Chang, W., Paulsson, B., Khan, A., Calabrese, J.R., 2006. BOLDER II Study Group, 2006. Efficacy of quetiapine monotherapy in bipolar I and II depression: a double-blind, placebo-controlled study (the BOLDER II study). J. Clin. Psychopharmacol. 26, 600–609. Tohen, M., Vieta, E., Calabrese, J., Ketter, T.A., Sachs, G., Bowden, C., Mitchell, P.B., Centorrino, F., Risser, R., Baker, R.W., Evans, A.R., Beymer, K., Dube, S., Tollefson, G.D., Breier, A., 2003. Efficacy of olanzapine and olanzapine–fluoxetine combination in the treatment of bipolar I depression. Arch. Gen. Psychiatry 60, 1079–1088.
Contents lists available at ScienceDirect
Journal of Affective Disorders journal homepage: www.elsevier.com/locate/jad
Review
Clinical assessment of lurasidone benefit and risk in the treatment of bipolar I depression using number needed to treat, number needed to harm, and likelihood to be helped or harmed Leslie Citrome a,n, Terence A. Ketter b, Josephine Cucchiaro c, Antony Loebel c a
New York Medical College, Valhalla, NY, USA Stanford University School of Medicine, Stanford, CA, USA c Sunovion Pharmaceuticals Inc., Fort Lee, NJ, USA b
art ic l e i nf o
a b s t r a c t
Article history: Received 20 September 2013 Accepted 22 October 2013 Available online 28 October 2013
Background: Prior to recent FDA approval of lurasidone for treatment of bipolar depression there were only two approved treatments for this condition (quetiapine and olanzapine–fluoxetine combination), and these were as likely to provide therapeutic benefit as adverse effects. We assessed the efficacy, safety, and tolerability of lurasidone for major depressive episodes associated with bipolar I disorder, using number needed to treat (NNT, for benefits), number needed to harm (NNH, for harms), and likelihood of being helped or harmed (LHH, ratio of NNH to NNT, for trade-offs between benefits vs. harms). Methods: Data was collected from two 6-week multicenter, randomized, double-blind, placebo-controlled, flexibly-dosed acute bipolar I depression studies, one using lurasidone monotherapy at 20–60 mg/d or 80–120 mg/d, and the other using lurasidone 20–120 mg/d adjunctive to lithium or valproate. The NNT or NNH was calculated for lurasidone vs. placebo for the following dichotomous outcomes: response (Z 50% reduction from baseline on Montgomery Asberg Depression Rating Scale (MADRS) total score); remission (final MADRS total score r12); discontinuation due to an adverse event (AE); weight gain Z 7% from baseline; incidence of spontaneously reported AEs; and incidence of total cholesterol Z 240 mg/dl, low-density lipoprotein cholesterol Z 160 mg/dl, fasting triglycerides Z200 mg/dl and glucose Z 126 mg/dl post-baseline. Results: NNT vs. placebo for response was 5 for lurasidone monotherapy (both dose ranges) and 7 for adjunctive therapy. NNT vs. placebo for remission for lurasidone monotherapy was 6 for 20–60 mg/d and 7 for 80–120 mg/d and 7 for adjunctive lurasidone. NNH vs. placebo for discontinuation due to an AE for lurasidone monotherapy was 642 for 20–60 mg/d and 181 for 80–120 mg/d, and for adjunctive lurasidone was 54 (negative NNH denotes an advantage for lurasidone). Lurasidone was not associated with any clinically meaningful mean weight or metabolic changes compared to placebo; NNH vs. placebo for weight gain Z7% was 29 for 20–60 mg/d and 5550 for 80–120 mg/d and 42 for adjunctive lurasidone. The three most frequently occurring AEs with the largest difference in incidence for lurasidone vs. placebo were nausea, akathisia, and somnolence, with NNH values for lurasidone vs. placebo ranging from 11 (nausea with lurasidone monotherapy 80–120 mg/d) to 130 (somnolence with lurasidone monotherapy 20–60 mg/d). LHH was substantially and consistently 4 1 (indicating benefit being more likely than harm) when contrasting response or remission vs. AEs or weight gain. Limitations: Additional studies, including longer-term and open-label, “real world” data is needed to confirm the results reported here. Conclusions: NNT, NNH, and LHH help quantify relative benefits (efficacy) and harms (side effects), thus placing lurasidone findings in research studies into clinical perspective. Lurasidone, compared to other treatments approved for bipolar depression, yielded comparable benefits (all had single-digit NNT vs. placebo for response or remission), and less risk of harm (double-digit or greater NNHs with lurasidone compared to single-digit NNHs for sedation with quetiapine and for Z 7% weight gain with olanzapine– fluoxetine combination), and thus a substantially more favorable LHH (4 or >>1) with lurasidone monotherapy and adjunctive therapy, compared to quetiapine and olanzapine–fluoxetine combination (LHH oor 1). & 2013 Elsevier B.V. All rights reserved.
Keywords: Bipolar I depression Efficacy Lurasidone Number needed to harm Number needed to treat Tolerability
n
Correspondence to: Leslie Citrome, 11 Medical Park Drive, Suite 106, Pomona, NY 10970, USA. Tel.: þ 1 845 362 2081; fax: þ1 845 362 8745. E-mail addresses: [email protected], [email protected] (L. Citrome).
0165-0327/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jad.2013.10.040
L. Citrome et al. / Journal of Affective Disorders 155 (2014) 20–27
21
Contents 1. 2. 3.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Efficacy outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Discontinuation due to adverse events. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Specific safety/tolerability outcomes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. Consideration of benefit vs. harm: LHH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Role of funding source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conflict of interest. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1. Introduction Lurasidone received approval by the United States Food and Drug Administration in June 2013 for the treatment of adults with major depressive episodes associated with bipolar I disorder (bipolar depression), as either a monotherapy or as adjunctive to lithium or valproate (Sunovion, 2013). Prior to lurasidone's approval, there were only two approved treatments for bipolar depression (namely quetiapine and olanzapine–fluoxetine combination), each similarly as likely to provide therapeutic vs. adverse effects (Ketter et al., 2011), with quetiapine being more problematic with respect to sedation/somnolence and olanzapine–fluoxetine combination being more problematic with respect to weight gain and metabolic problems (Ketter et al., 2011; Citrome, 2011a). Lurasidone is a second-generation antipsychotic agent that was originally approved in 2010 for the treatment of patients with schizophrenia (Sunovion, 2013; Citrome, 2011b). Although it is similar to other atypical antipsychotics in being a dopamine D2 and serotonin 5-HT2A receptor antagonist, lurasidone also exhibits high binding affinity for the serotonin 5-HT7 receptor (antagonist) and moderate affinity for the serotonin 5-HT1A (partial agonist) and noradrenaline α2c (antagonist) receptors. Lurasidone displays little to no appreciable affinity for histamine H1 and acetylcholine M1 receptors. Lurasidone's low affinity for histaminergic receptors could yield less risk of sedation, weight gain and related metabolic effects, as suggested by clinical trials in schizophrenia (Citrome, 2011b). Number needed to treat (NNT) and number needed to harm (NNH) are measures of effect size and indicate how many patients would need to be treated with one agent instead of the comparator in order to encounter one additional beneficial or harmful outcome of interest, respectively (Citrome, 2008; Citrome, 2011c). NNT is commonly used to denote potential desirable outcomes (benefits, i.e., therapeutic effects) that clinicians and patients would want to encounter and NNH is commonly used to denote potential undesirable outcomes (harms, i.e., adverse effects) that clinicians and patients would like to avoid. NNT and NNH offer clinicians ways of quantifying clinical relevance or “clinical significance.” Low NNTs (or NNHs) are evidenced when there are large differences between the interventions in question. For example, a NNT of 2 would be a very large effect size, as a difference can be observed after treating two patients with one of the interventions vs. the other. In contrast, a NNT of 50 would mean little difference between the two interventions because it would require treating fifty patients to encounter a difference in outcome. Potentially useful medications have a low NNT and a high NNH (i.e., benefit more likely than harm) when compared to other interventions. A rule of thumb is that single-digit NNTs vs.
21 21 22 22 23 23 25 25 26 26 26 27 27
placebo for efficacy measures suggest interventions that have potentially useful advantages (indeed, approved bipolar disorder treatments generally have such favorable NNTs), whereas doubledigit or higher NNHs vs. placebo for adverse outcomes suggest interventions that have potentially adequate tolerability (unfortunately, until recently, approved bipolar disorder treatments commonly failed to have such favorable NNHs). Likelihood of being helped or harmed (LHH) can be a useful way of integrating data regarding relative benefits and risks (Citrome and Kantrowitz, 2008; Citrome and Ketter, 2013; Straus, 2002). LHH is calculated using the following formula: LHH ¼ (1/NNT)/(1/NNH) ¼NNH/NNT, with a LHH 41 meaning help is more likely than harm, and a LHH less than one meaning the opposite. Choosing which NNH and NNT to use in calculating LHH requires careful consideration so that the outcomes being assessed are well matched and consistent with clinician's and patient's values and preferences (Citrome and Kantrowitz, 2008; Citrome and Ketter, 2013; Straus, 2002). Similar to a prior report on lurasidone for the treatment of schizophrenia (Citrome, 2012), we sought to describe the efficacy, safety, tolerability, and place in therapy of lurasidone in the treatment of major depressive episodes associated with bipolar I disorder, using the metrics of NNT, NNH, and LHH.
2. Methods Study data were analyzed post hoc from two 6-week, placebocontrolled studies of lurasidone that enrolled subjects with or without rapid cycling and without psychotic features – a monotherapy study where lurasidone was flexibly dosed at 20–60 mg/d or 80– 120 mg/d and an adjunctive therapy trial where flexibly dosed lurasidone (20–120 mg/d) was added to ongoing treatment with either lithium or valproate. The methods and principal results have been previously reported in detail (Loebel et al., 2013a; Loebel et al., 2013b). Table 1 provides an overview of the subjects in the two studies. Analogous to other NNT and NNH analyses (Citrome, 2012; Citrome et al., 2013), rates for the following dichotomous outcomes for each arm of each study were used: response (Z50% reduction from baseline on Montgomery Asberg Depression Rating Scale (MADRS) total score); remission (final MADRS total score of r12); discontinuation due to an adverse event (AE); weight gain Z7% from baseline; incidence of spontaneously reported AEs; incidence of shifts of total cholesterol from o240 mg/dL to Z240 mg/dL, low-density lipoprotein cholesterol from o160 mg/dL, Z160 mg/dL, fasting triglycerides from o200 mg/dL to Z200 mg/dL and glucose from o126 mg/dL to Z126 mg/dl, post-baseline. For any metabolic outcomes that
22
L. Citrome et al. / Journal of Affective Disorders 155 (2014) 20–27
Table 1 Phase III 6-week randomized controlled trials for lurasidone in patients with major depressive episodes associated with bipolar I disorder, baseline characteristics (ITT Population). Adjunctive lurasidone 20–120 mg/d vs. adjunctive placebo
Lurasidone monotherapy 20–60 mg/d or 80–120 mg/d vs. placebo
Lurasidone
Lurasidone 20–60 mg/d
Placebo
Randomized subjects, N 183 165 166 183 163 164 Safety population, Na ITT population, Nb 179 161 161 Male, n (%) 93 (52%) 85 (53%) 70 (43%) Age, years, mean 41.0 42.6 41.3 Race, n (%) White 108 (60%) 102 (63%) 107 (66%) Black 25 (14%) 16 (10%) 21 (13%) Asian 43 (24%) 37 (23%) 23 (14%) Other 3 (2%) 6 (4%) 10 (6%) Receiving lithium or valproate at baseline, Lithium, 90 (50%); valproate, 89 (50%) Lithium, 73 (46%); valproate, 87 (54%) NA n (%) MADRS total score, mean 30.6 30.8 30.3
Lurasidone 80–120 mg/d
Placebo
169 167 162 64 (40%) 42.0
170 168 162 75 (46%) 41.2
106 (65%) 25 (15%) 23 (14%) 8 (5%) NA
107 (66%) 21 (13%) 28 (17%) 6 (4%) NA
30.6
30.5
ITT – Intent-to-Treat. MADRS – Montgomery–Asberg Depression Rating Scale. NA – not applicable. a
Safety population consisted of all randomized subjects who received at least one dose of study medication. ITT population consisted of all subjects who were randomized, received at least one dose of study medication, and had a baseline and at least one post-baseline MADRS measure. b
suggested potentially relevant shifts to the clinically abnormal range, additional analyses were conducted to examine shifts from normal levels to high levels, and borderline levels to high levels. In calculating the proportions (n/N) of subjects for each treatment arm who met criteria for response or remission at endpoint (last observation carried forward (LOCF)), N, the denominator, was the Intent-to-Treat population (all subjects who were randomized, received at least one dose of study medication, and had a baseline and at least one post-baseline MADRS assessment. When reporting rates for AEs, N reflected the safety population (all randomized subjects who received at least one dose of study medication). For metabolic shift rates, N reflected the proportion with a baseline and at least one post-baseline assessment. NNT for efficacy outcomes and NNH for safety/tolerability outcomes were calculated for each study by first computing the difference in proportions between the experimental intervention and the control intervention for the outcome of interest (risk difference) and then computing the reciprocal of the risk difference and rounding up to the next highest whole number (Citrome, 2009). Calculating the 95% confidence interval (CI) for the NNT or NNH involved first calculating the 95% CI for the risk difference; the formula for this can be found elsewhere (Citrome, 2009). When the 95% CI for the NNT or NNH estimate contained infinity, the NNT or NNH estimate was not statistically significant at the P threshold of o0.05 and under those circumstances, instead of providing the 95% CI, the notation “ns” for not significant was made. All calculations for NNT and NNH were performed using the active medication as the experimental intervention and placebo as the control intervention, with the assumption that the study medication would be superior to placebo on efficacy outcomes but inferior to placebo on safety or tolerability outcomes. Thus for study medication compared to placebo, “negative” NNTs and “negative” NNHs denoted therapeutic disadvantages and tolerability advantages, respectively. Likelihood of being helped or harmed (LHH), the ratio of NNH to NNT, was calculated to illustrate trade-offs between benefits (response/remission) vs. harms (AEs), with LHH values 41 representing a greater likelihood of encountering benefit rather than harm. To place lurasidone into clinical context using outcomes generally accepted as relevant in this type of analysis, NNT for
response and remission, NNH for weight gain Z7% from baseline, NNH for discontinuation because of an AE, NNH for commonly encountered AEs, and NNH for shifts for metabolic variables, were calculated wherever possible from published, 8-week, randomized controlled registration trials and product labeling for the other products approved for the treatment of bipolar depression: quetiapine immediate and extended-release (Calabrese et al., 2005; Thase et al., 2006; Suppes et al., 2010; AstraZeneca, 2011a, 2011b) and olanzapine–fluoxetine combination (Tohen et al., 2003; Eli Lilly, 2013). For the lurasidone monotherapy study, NNT for response and remission and LHH were calculated both separately for the 20–60 mg/d and 80–120 mg/d dose groups for the main analysis, and in aggregate to yield a pooled (20–120 mg/d) dose group for selected additional analyses.
3. Results 3.1. Efficacy outcomes Table 2 provides the rates for response and remission and the NNT vs. placebo for the different treatment arms. Lurasidone demonstrated statistically significant NNT advantages vs. placebo for both response and remission. NNT vs. placebo for response was 5 for lurasidone monotherapy (both dose ranges) and 7 for adjunctive therapy; NNT vs. placebo for remission for lurasidone monotherapy was 6 for 20–60 mg/d and 7 for 80–120 mg/d and 7 for adjunctive lurasidone (see Table 2 for the 95% CIs). For the monotherapy lurasidone study, response and remission rates for the pooled lurasidone dose arms were 168/323 (52.0%) and 132/ 323 (40.9%), respectively, yielding NNTs vs. placebo of 5 (95% CI 4–8), and 7 (95% CI 4–14), respectively. Figs. 1 and 2 graphically display the NNT vs. placebo for response and remission for the pooled doses of lurasidone monotherapy, together with data calculated from the 8-week, placebo-controlled, US registration trials of quetiapine (pooled data from Calabrese et al., 2005; Thase et al., 2006; Suppes et al., 2010) and olanzapine–fluoxetine combination (Tohen et al., 2003) for bipolar depression.
L. Citrome et al. / Journal of Affective Disorders 155 (2014) 20–27
23
Table 2 Response, remission, and number needed to treat vs. placebo (ITT population).
a
ITT population, N Responders, n (%)b NNT, response vs. placebo (95% CI) Remitters, n (%)c NNT, remission vs. placebo (95% CI)
Adjunctive lurasidone 20–120 mg/d vs. adjunctive Placebo
Lurasidone monotherapy 20–60 mg/d or 80–120 mg/d vs. placebo
Lurasidone
Placebo
Lurasidone 20–60 mg/d
Lurasidone 80–120 mg/d
Placebo
179 102 (57.0%) 7 (4–24) 90 (50.3%) 7 (4–23)
161 68 (42.2%) NA 57 (35.4%) NA
161 86 (53.4%) 5 (3–8) 68 (42.2%) 6 (4–14)
162 82 (50.6%) 5 (4–11) 64 (39.5%) 7 (4–21)
162 49 (30.2%) NA 40 (24.7%) NA
CI – confidence interval. ITT – Intent-to-Treat. MADRS – Montgomery–Asberg Depression Rating Scale. NA – not applicable. NNT – number needed to treat. a ITT population consisted of all subjects who were randomized, received at least one dose of study medication, and had a baseline and at least one post-baseline MADRS measure. b Proportion of subjects who achieved a response, defined as Z 50% reduction from baseline total score on the MADRS. c Proportion of subjects who achieved remission, defined as a MADRS total score of r 12 at endpoint.
Fig. 1. MADRS responders with lurasidone monotherapy, quetiapine immediate and extended release monotherapy, and olanzapine/fluoxetine combination – number needed to treat and 95% confidence interval vs. placebo (all doses for the respective agents pooled). IR – immediate release. MADRS – Montgomery–Asberg Depression Rating Scale. NNT – number needed to treat. XR – extended release. (Proportion of subjects who achieved a response, defined as Z 50% reduction from Baseline total score on the MADRS).
Fig. 2. MADRS remitters with lurasidone monotherapy, quetiapine immediate and extended release monotherapy, and olanzapine/fluoxetine combination – number needed to treat and 95% confidence interval vs. placebo (all doses for the respective agents pooled). IR – immediate release. MADRS – Montgomery–Asberg Depression Rating Scale. NNT – number needed to treat. XR – extended release. (Proportion of subjects who achieved remission, defined as a MADRS total score of r12 at last observation carried forward endpoint).
3.2. Discontinuation due to adverse events
baseline Z 7%, and shifts on fasting plasma glucose, fasting plasma triglycerides, and all available data for total cholesterol and low density lipoprotein (LDL). For adjunctive lurasidone vs. placebo, the incidence of weight gain Z7% from baseline was 3.1% vs. 0.7%, yielding a NNH of 42 (ns). For lurasidone monotherapy 20–60 mg/ d, the incidence of weight gain Z7% from baseline was 4.2% vs. 0.7% for placebo yielding a NNH of 29 (95% CI 15–52318). For lurasidone monotherapy 80–120 mg/day vs. placebo, the corresponding rates were essentially identical (0.7%), yielding a nonstatistically significant (and clinically irrelevant) NNH of 5550. However, for olanzapine–fluoxetine combination, weight gain Z7% from baseline was observed in 19.5% of olanzapine–fluoxetine combination subjects vs. 0.3% for placebo, yielding a NNH of 6 (95% CI 4–10) (Eli Lilly, 2013), on par with the effect size observed for efficacy of olanzapine–fluoxetine combination. For quetiapine, at all doses and all formulations, approximately 8.4% of subjects receiving quetiapine gained Z 7% of their baseline weight vs. approximately 1.9% for placebo in studies of bipolar depression (AstraZeneca, 2011a, 2011b), resulting in a NNH of 16 (95% CI 12–25). When examining shift in metabolic variables in the studies for lurasidone, the only potential signal for abnormalities was shift in fasting triglycerides to Z200 mg/dL postbaseline where lurasidone monotherapy at both doses exhibited statistically significant NNH values of 14 and 15, for lurasidone 20–60 mg/d and 80–120 mg/d, respectively. However, this appears
Table 3 provides the NNH for discontinuation due to an AE. Discontinuation due to an AE was not statistically significant different from placebo for lurasidone monotherapy 20–60 mg/ day, 80–120 mg/day, or for adjunctive lurasidone. Calculation of the NNH resulted in a negative number (a negative NNH denotes an advantage for lurasidone regarding the potential harm) for adjunctive lurasidone vs. placebo, lurasidone monotherapy 80– 120 mg/d vs. placebo, and the combined dosage arms for lurasidone monotherapy (NNH 496, ns). NNH for lurasidone monotherapy 20–60 mg/d vs. placebo was 642 (ns). In contrast, the analogous discontinuation rates for quetiapine (all doses all formulations) were 126/841 (15.0%) for active medication vs. 20/487 (4.1%) for placebo (Calabrese et al., 2005; Thase et al., 2006; Suppes et al., 2010), resulting in a NNH of 10 (95% CI 8–13). For the olanzapine–fluoxetine combination pivotal studies, a higher proportion discontinued because of an AE with placebo (19/377, 5.0%), than with olanzapine–fluoxetine combination (2/86, 2.3%) (Tohen et al., 2003), resulting in a NNH of 37 (ns). 3.3. Specific safety/tolerability outcomes Table 4 provides the rates and NNH for clinically relevant weight gain, as defined by an increase in body weight from
24
L. Citrome et al. / Journal of Affective Disorders 155 (2014) 20–27
Table 3 Rates and number needed to harm vs. placebo for discontinuation due to an adverse event (randomized population). Adjunctive lurasidone 20–120 mg/d vs. adjunctive placebo
Lurasidone monotherapy 20–60 mg/d or 80–120 mg/d vs. placebo
Lurasidone
Placebo
Lurasidone 20–60 mg/d
Lurasidone 80–120 mg/d
Placebo
165 13 (7.9%) NA
166 11 (6.6%) 642 (ns)
169 10 (5.9%) 181 (ns)
170 11 (6.5%) NA
Randomized, N 183 Discontinuation due to AE, n (%) 11 (6.0%) NNH, discontinuation due to an AE vs. placebo (95% CI) 54 (ns)
a
a
AE – adverse event. CI – confidence interval. NA – not applicable. NNT – number needed to treat. NNH – number needed to harm. ns – not significant (the 95% CI contains “infinity”). a
A negative NNH denotes an advantage for lurasidone regarding the potential harm.
Table 4 Rates and number needed to harm vs. placebo for metabolic variables (Safety population).
Safety Population, Na Weight increase Z7% from baseline, n (%) NNH (95% CI) Shift fasting glucose Z 126 mg/dL (post-baseline) NNH (95% CI) Shift fasting triglycerides Z 200 mg/dL (post-baseline) NNH (95% CI) Shift fasting triglycerides o 150 to Z200 mg/dL (post-baseline) NNH (95% CI) Shift fasting triglycerides Z 150 and o 200 to Z 200 mg/dL(post-baseline) NNH (95% CI) Shift total cholesterol Z240 mg/dL (post-baseline) NNH (95% CI) Shift LDL cholesterol Z160 mg/dL (post-baseline) NNH (95% CI)
Adjunctive lurasidone 20– 120 mg/d vs. adjunctive placebo
Lurasidone monotherapy 20–60 mg/d or 80–120 mg/d vs. placebo
Lurasidone
Placebo
Lurasidone 20–60 mg/d
Lurasidone 80–120 mg/d
Placebo
183 5/162 (3.1%) 42 (ns) 0/138 (0%) 64 (ns) b 14/118 (11.9%) 28 (ns) 9/98 (9.2%) 23 (ns) 5/20 (25.0%) 18 (ns) 7/135 (5.2%) 83 (ns) 6/144 (4.2%) 58 (ns)
163 1/150 (0.7%) NA 2/127 (1.6%) NA 9/109 (8.3%) NA 4/83 (4.8%) NA 5/26 (19.2%) NA 5/126 (4.0%) NA 3/124 (2.4%) NA
164 6/143 (4.2%) 29 (15–52318) 3/123 (2.4%) 136 (ns) b 12/108 (11.1%) 14 (7–129) 7/87 (8.0%) 20 (ns) 5/21 (23.8%) 6 (ns) 5/113 (4.4%) 534 (ns) 5/121 (4.1%) 108 (ns)
167 1/147 (0.7%) 5550 (ns) 7/130 (5.4%) 46 (ns) 12/115 (10.4%) 15 (8–252) 5/91 (5.5%) 41 (ns) 7/24 (29.2%) 5 (ns) 5/114 (4.4%) 673 (ns) 9/126 (7.1%) 26 (ns)
168 1/151 (0.7%) NA 4/126 (3.2%) NA 4/114 (3.5%) NA 3/100 (3.0%) NA 1/14 (7.1%) NA 5/118 (4.2%) NA 4/125 (3.2%) NA
CI – confidence interval. NA – not applicable. NNH – number needed to harm. ns – not significant (the 95% CI contains “infinity”). a b
Safety population consisted of all randomized subjects who received at least one dose of study medication. A negative NNH denotes an advantage for lurasidone regarding the potential harm.
to be largely driven by subjects with a borderline fasting triglyceride level of Z150 mg/dL and o200 mg/dL at study baseline (see Table 4). Equivalent shift data for olanzapine–fluoxetine combination in bipolar depression were not reported in the olanzapine–fluoxetine combination clinical trial as published (Tohen et al., 2003) or in product labeling (Eli Lilly, 2013), however olanzapine's lipid-altering profile is well known (Citrome et al., 2011). Limited shift data are available for quetiapine for bipolar depression from product labeling (AstraZeneca, 2011a, 2011b) where for all doses and all formulations, NNH vs. placebo for shifts in triglycerides Z200 mg/dL is 23 (95% CI 12–420), and for total cholesterol Z240 mg/dL, 29 (95% CI 15–396). Table 5 outlines the incidence of the most commonly encountered spontaneously reported AEs reported in the clinical trials (incidence at least 5% in any treatment arm). However, for many of these AEs, the difference from placebo was small. The three most frequently occurring AEs with the largest difference in incidence for lurasidone vs. placebo were nausea, akathisia, and
somnolence, with NNH values for lurasidone vs. placebo ranging from 11 (nausea with lurasidone monotherapy 80–120 mg/d) to 130 (somnolence with lurasidone monotherapy 20–60 mg/d). In contrast to the “double-digit” NNH values for lurasidone, olanzapine–fluoxetine combination and quetiapine, in their respective registration studies, demonstrated clinically relevant NNH values vs. placebo in the “single digits.” For example, for the studies of quetiapine in bipolar depression, regardless of formulation or dose, NNH for somnolence (includes hypersomnia, sedation and somnolence) was 3 (rounded up from 2.4), with an exceedingly narrow 95% CI of 2.2–2.7 (calculated from AstraZeneca, 2011a, 2011b). The AE of somnolence alone from pooled clinical trial data was observed in 234/835 (28.0%) subjects receiving quetiapine vs. 31/487 (6.4%) for placebo (Calabrese et al., 2005; Thase et al., 2006; Suppes et al., 2010), yielding a NNH of 5 (95% CI 4–6). The data for sedation with quetiapine were similar. Dry mouth was also common with quetiapine vs. placebo (42.5% vs. 11.1%), with a NNH of 4 (95% CI 3–4). Other statistically significant NNH values
L. Citrome et al. / Journal of Affective Disorders 155 (2014) 20–27
25
Table 5 Rates and number needed to harm vs. placebo for common adverse events (Safety population). Adjunctive lurasidone 20–120 mg/d vs. adjunctive placebo
Lurasidone monotherapy 20–60 mg/d or 80–120 mg/d vs. placebo
Lurasidone
Placebo
Lurasidone 20–60 mg/d
Lurasidone 80–120 mg/d
Placebo
163 18 (11.0%) NA 9 (5.5%) NA 16 (9.8%) NA 7 (4.3%) NA 4 (2.5%) NA 9 (5.5%) NA 11 (6.7%) NA 20 (12.3%) NA 7 (4.3%) NA 5 (3.1%) NA
164 17 (10.4%) 39 (ns) 12 (7.3%) 130 (ns) 8 (4.9%) 40 (ns) 13 (7.9%) 18 (10–124) 4 (2.4%) 154 (ns) 8 (4.9%) 29 (ns) b 8 (4.9%) 33 (ns) 23 (14.0%) 48 (ns) 4 (2.4%) 19 ( 10 to 162) 10 (6.1%) 52 (ns)
167 29 (17.4%) 11 (6–39) 23 (13.8%) 14 (8–126) 15 (9.0%) 16 (9–60) 18 (10.8%) 12 (8–32) 10 (6.0%) 24 (12–1190) 11 (6.6%) 58 (ns) b 5 (3.0%) 83 (ns) 15 (9.0%) 35 (ns) b 10 (6.0%) 58 (ns) b 6 (3.6%) 175 (ns) b
168 13 (7.7%) NA 11 (6.5%) NA 4 (2.4%) NA 4 (2.4%) NA 3 (1.8%) NA 14 (8.3%) NA 3 (1.8%) NA 20 (11.9%) NA 13 (7.7%) NA 7 (4.2%) NA
a
183 Safety population, N Nausea, n (%) 32 (17.5%) NNH (95% CI) 16 (ns) Somnolence (hypersomnia, sedation, somnolence), n (%) 20 (10.9%) NNH (95% CI) 19 (ns) Extrapyramidal symptomsc, n (%) 28 (15.3%) NNH (95% CI) 19 (ns) Akathisia, n (%) 14 (7.7%) NNH (95% CI) 30 (ns) Vomiting, n (%) 9 (4.9%) NNH (95% CI) 41 (ns) Insomnia, n (%) 13 (7.1%) NNH (95% CI) 64 (ns) Diarrhea, n(%) 8 (4.4%) NNH (95% CI) 43 (ns) b Headache, n (%) 19 (10.4%) NNH (95% CI) 53 (ns) b Dizziness, n (%) 7 (3.8%) NNH (95% CI) 214 (ns) b Dry mouth, n (%) 5 (2.7%) NNH (95% CI) 299 (ns) b
b
CI – confidence interval. NA – not applicable. NNH – number needed to harm. ns – not significant (the 95% CI contains “infinity”). a b c
Safety population consisted of all randomized subjects who received at least one dose of study medication. A negative NNH denotes an advantage for lurasidone regarding the potential harm. Term includes cogwheel rigidity, drooling, dystonia, glabellar reflex abnormal, muscle rigidity, oromandibular dystonia, parkinsonism, tremor, trismus, torticollis.
were obtained when comparing quetiapine with placebo on dizziness (16.8% vs. 8.0%, NNH 12 (95% CI 9–19)), constipation (9.9% vs. 4.5%, NNH 19 (95% CI 13–38)), AEs associated with extrapyramidal syndrome (includes akathisia, dystonia, extrapyramidal disorder, hypertonia, and tremor) (8.6% vs. 3.3%, NNH 19 (95% CI 13–35)), and fatigue (9.6% vs. 6.0%, NNH 28 (95% CI 16– 138) (Calabrese et al., 2005; Thase et al., 2006; Suppes et al., 2010). For olanzapine–fluoxetine combination, “single digit” NNH values vs. placebo were observed for the AE of weight gain (17.4% vs. 2.7%, NNH 7 (95% CI 5–16) and diarrhea (18.6% vs. 6.6%, NNH 9 (95% CI 5–30)) (Eli Lilly, 2013). Other statistically significant NNH values were obtained when comparing olanzapine–fluoxetine combination with placebo on dry mouth (16.3% vs. 6.1%, NNH 10 (95% CI 6– 50)), asthenia (12.88 vs. 3.2%, NNH 11 (95% CI 6–43)), increased appetite (12.8% vs. 5.0%, NNH 13 (95% CI 7–282)), and tremor (9.3% vs. 2.4%, NNH 15 (95% CI 8–171)) (Eli Lilly, 2013). 3.4. Consideration of benefit vs. harm: LHH For lurasidone in the treatment of bipolar I depression, LHH was substantially and consistently 4 1 (indicating benefit being more likely than harm) when contrasting response or remission vs. AEs (see Table 6). For example, LHH for response vs. somnolence was Z2.8 for lurasidone monotherapy 80–120 mg/d, meaning that lurasidone high dose monotherapy was at least 2.8 times more likely to result in a reduction of 50% from the baseline MADRS total score than a spontaneously reported AE of somnolence. This is contrast to quetiapine where the LHH for response vs. somnolence was 3/6, or 0.5, meaning that quetiapine was only half as likely to result in a reduction of 50% from the baseline MADRS total score as a spontaneously reported AE of somnolence. For weight gain Z 7% from baseline, the NNH vs. placebo for lurasidone was statistically significant only for monotherapy within the dose range of 20–60 mg/d, with a LHH for response vs. weight gain
of 5.8. In contrast, this same metric for olanzapine–fluoxetine combination when comparing response or remission vs. weight gain, LHH was very close to 1.0, meaning for olanzapine–fluoxetine combination in the treatment of bipolar I depression response or remission was equally likely as clinically significant weight gain.
4. Discussion For patients with bipolar I depression, we found the therapeutic effects of lurasidone (monotherapy or adjunctive to lithium or valproate) vs. placebo resulted in single digit NNT values (5–7 for response, 6–7 for remission). These were comparable to the singledigit NNT values for the previously approved bipolar depression treatments quetiapine (6 for response, 6 for remission) and olanzapine–fluoxetine combination (4 for response, 5 for remission). Of note, the single-digit NNT for adjunctive lurasidone was evident even though the comparison was with adjunctive placebo added to a standing regimen of either lithium or valproate, either of which may have (albeit modest) potential therapeutic benefits for the treatment of acute bipolar depression (Ketter, 2010). Unapproved agents such as lamotrigine monotherapy and adjunctive antidepressants are commonly used to treat acute bipolar depression. However, evidence regarding their efficacy profiles for this indication is not encouraging. For example, for lamotrigine vs. placebo, NNT for response in bipolar depression is 12, and that for antidepressants vs. placebo, NNT is 29 (calculated from Sidor and Macqueen, 2011). Aripiprazole and ziprasidone have been tested in clinical trials for the treatment of bipolar depression but have not demonstrated efficacy, with NNT values for response vs. placebo of 44 (Ketter, 2010) and 163 (calculated from Lombardo et al., 2012). In addition to a favorable NNT for efficacy for lurasidone, we found that tolerability outcome differences for lurasidone vs. placebo generated NNH values in the double or triple digits (14–130 for
26
L. Citrome et al. / Journal of Affective Disorders 155 (2014) 20–27
Table 6 Likelihood of being helped or harmed: Response vs. nausea, akathisia, somnolence, extrapyramidal symptoms, and weight gain Z7%. Adverse event
Adjunctive
Monotherapy
Lurasidone
Lurasidone 20–60 mg/d
Lurasidone 80–120 mg/d
Monotherapy groups combined
NNT
NNT Nausea Akathisia Somnolence Extrapyramidal symptomsb Weight gain Z 7%
7 7 7 7 7
NNH a
16 30a 19a 19a 42a
LHH 2.3 4.3 2.7 2.7 6.0
5 5 5 5 5
NNH
LHH
NNT
NNH
LHH
NNT
NNH
LHH
a
7.8 3.6 26.0 8.0 5.8
5 5 5 5 5
11 12 14 16 5550a
2.2 2.4 2.8 3.2 1110
5 5 5 5 5
17 15 25a 22 58a
3.4 3.0 5.0 4.4 11.6
39 18 130a 40a 29
LHH – likelihood of being helped or harmed; calculated by NNH/NNT. NNH – number needed to harm. NNT – number needed to treat. a b
Not statistically significant; interpret the resultant LHH with caution. Term includes cogwheel rigidity, drooling, dystonia, glabellar reflex abnormal, muscle rigidity, oromandibular dystonia, parkinsonism, tremor, trismus, torticollis.
somnolence, 29–5550 for Z7% weight gain). These favorable NNH values were similar for lurasidone monotherapy and for adjunctive lurasidone when combined with lithium or valproate. These NNH values were substantially more favorable than the single-digit NNHs for the previously approved bipolar depression treatments quetiapine (NNH for somnolence, 3) and olanzapine–fluoxetine combination (NNH for Z7% weight gain, 6). The above-mentioned pattern of benefits relative to harms yielded LHH values that were consistently more favorable for lurasidone in contrast to the previously approved treatments for bipolar depression. For example, for lurasidone high dose monotherapy, LHH for response vs. somnolence was 2.8, which compared favorably with the analogous LHH for quetiapine which was 0.5. Lurasidone also demonstrated LHH values substantially 41.0 (range was 5.8 to 1100) when contrasting response or remission with weight gain Z7%. In contrast this same metric for olanzapine–fluoxetine combination was 1. Clinically, additional considerations include the time to onset of the AE vs. time to onset of a therapeutic response, as well as the severity and duration of the AE. The AE in question may be easily manageable if it is non-serious and short-lived. With respect to metabolic outcomes, no clinically relevant adverse safety signal was evident for lurasidone in the examination of clinically relevant weight gain or laboratory abnormalities; however those with preexisting abnormalities require careful monitoring. Despite their tolerability challenges, olanzapine–fluoxetine combination and quetiapine monotherapy may still have utility in highurgency situations, particularly in persons who have demonstrated good outcomes with these interventions in the past, where a pressing clinical need for efficacy mitigates their tolerability shortcomings. In addition, despite their efficacy challenges, the older unapproved treatments, specifically lamotrigine and antidepressants, may have utility in low-urgency situations, in which a compelling need for tolerability might mitigate their efficacy shortcomings. Lurasidone may ultimately prove to have utility in a broad spectrum of situations, independent of the degree of urgency, because of evidence suggesting not only adequate efficacy, but also adequate tolerability. Limitations: study populations and durations differed in the clinical trials for lurasidone, quetiapine and olanzapine–fluoxetine combination. The lurasidone trials were 6-weeks in duration and included only bipolar I disorder patients and excluded those with psychotic features. The quetiapine trials included subjects with both bipolar I disorder and bipolar II disorder and were 8-weeks in duration. The olanzapine–fluoxetine combination trials were also 8-weeks in duration and included subjects with bipolar I disorder (some of whom had psychotic symptoms), but did not include patients with bipolar II disorder.
A major limitation of NNT and NNH is that these metrics are limited to dichotomous outcomes. Other effect size measures are necessary when describing continuous outcomes without dichotomization (Citrome, 2010). Data were obtained from carefully structured studies and hence the results may not be generalizable to patients outside the confines of clinical trials. Long-term efficacy, safety and tolerability, including maintenance of response are also important considerations in the selection of treatments for bipolar depression. Dichotomous outcomes from long-term trials, such as relapse and weight gain exceeding a predefined threshold, will be of interest in further appraising the clinical profile of lurasidone.
5. Conclusions Lurasidone has recently received approval for the treatment of bipolar I depression not only as monotherapy but also as adjunctive (added to lithium or valproate) therapy. NNT, NNH, and LHH can help quantify relative benefits (efficacy) and harms (side effects), thus placing lurasidone findings in research studies into clinical perspective. Compared to the other approved bipolar depression treatments (quetiapine, olanzapine–fluoxetine combination), lurasidone yielded comparable benefits (all had single-digit NNTs vs. placebo for response or remission), less risk of harm (double-digit or greater NNHs with lurasidone compared to single-digit NNHs for somnolence with quetiapine and for weight gain Z7% with olanzapine–fluoxetine combination), and a more favorable ratio of benefit to harm likelihood (LHH much 41 with lurasidone compared to close to one with quetiapine and olanzapine–fluoxetine combination).
Role of funding source This study was supported Sunovion Pharmaceuticals Inc., Marlborough, MA and Fort Lee, NJ.
Conflict of interest In the past 36 months L. Citrome has engaged in collaborative research with, or received consulting or speaking fees, from Alexza, Alkermes, AstraZeneca, Avanir, Bristol-Myers Squibb, Eli Lilly, Envivo, Forest, Genentech, Janssen, Lundbeck, Merck, Mylan, Novartis, Noven, Otsuka, Pfizer, Reckitt Benckiser, Reviva, Shire, Sunovion, Takeda, and Valeant. In the past 36 months T. Ketter has engaged in collaborative research with, or received consulting or speaking fees, from Abbott, Allergan, AstraZeneca, Avanir, Bristol-Myers Squibb, Cephalon, Eli Lilly, Forest, GlaxoSmithKline, Janssen, Otsuka, Pfizer, Sunovion, and Teva; T. Ketter's spouse is an employee of and holds stock in Janssen Pharmaceuticals. J. Cucchiaro and A. Loebel are fulltime employees of Sunovion Pharmaceuticals.
L. Citrome et al. / Journal of Affective Disorders 155 (2014) 20–27
Acknowledgment This study was supported Sunovion Pharmaceuticals Inc., Marlborough, MA and Fort Lee, NJ.
References AstraZeneca, 2011a. Seroquel (quetiapine fumarate). Revised December 2011. Available from: 〈http://www1.astrazeneca-us.com/pi/Seroquel.pdf〉 accessed 28.01.13. AstraZeneca, 2011b. Seroquel XR (quetiapine fumarate extended release). Revised December 2011. Available from: 〈http://www1.astrazeneca-us.com/pi/sero quelxr.pdf〉 accessed 28.01.13. Calabrese, J.R., Keck Jr., P.E., Macfadden, W., Minkwitz, M., Ketter, T.A., Weisler, R.H., Cutler, A.J., McCoy, R., Wilson, E., Mullen, J., 2005. A randomized, double-blind, placebo-controlled trial of quetiapine in the treatment of bipolar I or II depression. Am. J. Psychiatry 162, 1351–1360. Citrome, L., 2008. Compelling or irrelevant? Using number needed to treat can help decide. Acta Psychiatr. Scand. 117, 412–419. Citrome, L., Kantrowitz, J., 2008. Antipsychotics for the treatment of schizophrenia: likelihood to be helped or harmed, understanding proximal and distal benefits and risks. Expert Rev. Neurother. 8, 1079–1091. Citrome, L., 2009. Quantifying risk: the role of absolute and relative measures in interpreting risk of adverse reactions from product labels of antipsychotic medications. Curr. Drug Saf. 4, 229–237. Citrome, L., 2010. Relative vs. absolute measures of benefit and risk: what is the difference? Acta Psychiatr. Scand. 121, 94–102. Citrome, L., 2011a. Olanzapine–fluoxetine combination for the treatment of bipolar depression. Expert Opin. Pharmacother. 12, 2751–2758. Citrome, L., 2011b. Lurasidone for schizophrenia: a review of the efficacy and safety profile for this newly approved second-generation antipsychotic. Int. J. Clin. Pract. 65, 189–210. Citrome, L., 2011c. Number needed to treat: what it is and what it is not, and why every clinician should know how to calculate it. J. Clin. Psychiatry 72, 412–413. Citrome, L., Holt, R.I., Walker, D.J., Hoffmann, V.P., 2011. Weight gain and changes in metabolic variables following olanzapine treatment in schizophrenia and bipolar disorder. Clin. Drug Investig. 31, 455–482. Citrome, L., 2012. Lurasidone for the acute treatment of adults with schizophrenia: what is the number needed to treat, number needed to harm, and likelihood to be helped or harmed? Clin. Schizophr. Relat. Psychoses 6, 76–85. Citrome, L., Ketter, T.A., 2013. When does a difference make a difference? Interpretation of number needed to treat, number needed to harm, and likelihood to be helped or harmed. Int. J. Clin. Pract. 67, 407–411.
27
Citrome, L., Goldberg, J.F., Portland, K.B., 2013. Placing transdermal selegiline for major depressive disorder into clinical context: number needed to treat, number needed to harm, and likelihood to be helped or harmed. J. Affect. Disord. 151, 409–417. Eli Lilly, 2013. Symbyax (olanzapine and fluoxetine hydrochloride). Revised January 2013. Available from: 〈http://pi.lilly.com/us/symbyax-pi.pdf〉 accessed 28.01.13. Ketter, T.A., 2010. Handbook of Diagnosis and Treatment of Bipolar Disorder. American Psychiatric Publishing, Inc., Washington, DC. Ketter, T.A., Citrome, L., Wang, P.W., Culver, J.L., Srivastava, S., 2011. Treatments for bipolar disorder: can number needed to treat/harm help inform clinical decisions? Acta Psychiatr. Scand. 123, 175–189. Loebel, A., Cucchiaro, J., Silva, R., Kroger, H., Sarma, K., Xu, J., Calabrese, J.R., 2013a. Lurasidone as adjunctive therapy with lithium or valproate for the treatment of bipolar I depression: a randomized, double-blind, placebo-controlled study. Am. J. Psychiatry. (in press). Loebel, A., Cucchiaro, J., Silva, R., Kroger, H., Hsu, J., Sarma, K., Sachs, G., 2013b. Lurasidone monotherapy in the treatment of bipolar I depression: a randomized, double-blind, placebo-controlled study. Am. J. Psychiatry. (in press). Lombardo, I., Sachs, G., Kolluri, S., Kremer, C., Yang, R., 2012. Two 6-week, randomized, double-blind, placebo-controlled studies of ziprasidone in outpatients with bipolar I depression: did baseline characteristics impact trial outcome? J. Clin. Psychopharmacol. 32, 470–478. Sidor, M.M., Macqueen, G.M., 2011. Antidepressants for the acute treatment of bipolar depression: a systematic review and meta-analysis. J. Clin. Psychiatry 72, 156–167. Straus, S.E., 2002. Individualizing treatment decisions. The likelihood of being helped or harmed. Eval. Health Prof. 25, 210–224. Sunovion, 2013. Latuda (lurasidone). Revised July 2013. Available from: 〈http:// www.latuda.com/LatudaPrescribingInformation.pdf〉 accessed 07.08.13. Suppes, T., Datto, C., Minkwitz, M., Nordenhem, A., Walker, C., Darko, D., 2010. Effectiveness of the extended release formulation of quetiapine as monotherapy for the treatment of acute bipolar depression. J. Affect. Disord. 121, 106–115. Thase, M.E., Macfadden, W., Weisler, R.H., Chang, W., Paulsson, B., Khan, A., Calabrese, J.R., 2006. BOLDER II Study Group, 2006. Efficacy of quetiapine monotherapy in bipolar I and II depression: a double-blind, placebo-controlled study (the BOLDER II study). J. Clin. Psychopharmacol. 26, 600–609. Tohen, M., Vieta, E., Calabrese, J., Ketter, T.A., Sachs, G., Bowden, C., Mitchell, P.B., Centorrino, F., Risser, R., Baker, R.W., Evans, A.R., Beymer, K., Dube, S., Tollefson, G.D., Breier, A., 2003. Efficacy of olanzapine and olanzapine–fluoxetine combination in the treatment of bipolar I depression. Arch. Gen. Psychiatry 60, 1079–1088.