Risk of relapse during tapering of antipsychotic medication after a first psychotic episode: association with D2 receptor affinity but not with tapering speed

Abstract

While antipsychotic maintenance treatment effectively prevents relapse after a first psychotic episode, many remitted antipsychotic users wish to reduce or discontinue their medication, due to side effects, long‐term health concerns, stigma, or the desire to regain autonomy. Current guidelines suggest gradual tapering, but what the optimal speed of this tapering should be, especially in patients who remitted from a first psychotic episode, remains unclear. Furthermore, D2 receptor affinity of the antipsychotic drug may also affect relapse risk. This study examined relapse risk and time to relapse, within the first 18 months after remission from a first psychotic episode, in 227 individuals who tapered antipsychotic medication. Relapse was defined dichotomously using consensus criteria based on the Positive and Negative Syndrome Scale, hospitalization for psychosis, or explicit clinical judgement of the treating psychiatrist. Tapering speed (in olanzapine equivalents mg/day) was calculated as the difference of antipsychotic dose between the start and the end of tapering, divided by the number of days in between. Antipsychotics were classified into partial D2 agonists, or antagonists with low or high D2 affinity. Logistic and Cox proportional hazards regression analyses were controlled for age, sex, cannabis use, and duration of first psychotic episode, as well as for differences in clinical and sociodemographic characteristics between the three drug D2 affinity groups using inverse probability of treatment weighting. During the follow‐up period, 45.8% (N=104) of participants experienced a relapse after tapering. The average tapering speed was 10 mg olanzapine equivalents over 75 days, with an average tapering duration of 124 days (range: 6‐334 days). Logistic regression analysis showed that the tapering speed did not predict the risk of relapse (z=0.989, p=0.323). Compared with users of high D2 affinity antagonists (N=57), patients using low D2 affinity antagonists (N=116) and partial D2 agonists (N=54) had a lower risk of relapse (respectively, z=−2.104; odds ratio, OR=0.48, p=0.035; and z=−2.278, OR=0.44, p=0.023). Users of high D2 affinity antagonists had a shorter time between the end of tapering and relapse (mean: 280 days) than low D2 affinity antagonist users (mean: 351 days, p=0.027) and partial D2 agonist users (mean: 357 days, p=0.040). Thus, D2 receptor affinity of the antipsychotic was more important than tapering speed in predicting psychotic relapse risk in individuals remitted from a first psychotic episode. Users of high D2 affinity antipsychotics had an about twice higher risk of relapse than users of other types of antipsychotics. This higher risk of relapse after tapering should be regarded as a relevant factor when selecting an antipsychotic drug for people with a first psychotic episode. For patients using strong D2 antagonists after remission from a first psychotic episode, extra monitoring during tapering is required.

Preventing relapse is a crucial treatment target following a first psychotic episode, since up to 85% of individuals experience a psychotic relapse within 5 years after remission ¹ , and relapse after a first psychotic episode is associated with a poorer long‐term outcome ² . There is ample evidence documenting the efficacy of antipsychotic medication in reducing the risk of relapse ³ . Consequently, international guidelines recommend the continuation of antipsychotic medication for at least one year following remission from a first psychotic episode ⁴ .

On the other hand, many individuals who remit from a first psychotic episode have a strong desire to reduce or completely discontinue their medication. This desire may be partly fueled by side effects and long‐term health concerns, but may also reflect stigma and the need for self‐determination and autonomy ⁵ , ⁶ . On average, 50% of people who remitted from a first psychotic episode discontinue medication prematurely, often without clinical guidance ⁷ , ⁸ , ⁹ . A previous study reported better long‐term social functioning after early tapering, compared to longer use of medication ¹⁰ . Dose reduction has also been experienced to aid in the development of better stress management strategies and in regaining a sense of autonomy ¹¹ . This preference to discontinue antipsychotic treatment needs to be balanced against the increased risk for psychotic relapse that follows cessation of medication ¹² . Knowledge on how to taper antipsychotic medication with minimal risk of relapse is essential for clinicians to guide the much‐desired antipsychotic dose reduction or discontinuation as safely as possible ¹³ .

An optimal protocol for antipsychotic tapering has not yet been established. There is consensus that gradual tapering is preferable to abrupt discontinuation ¹⁴ , ¹⁵ . Currently, “gradual” is generally described as “not abrupt” ¹⁶ , and lacks a clear definition. A meta‐analysis showed that dose reduction over >10 weeks – resulting in end doses about half of the initial dose, but generally still not below therapeutic levels – was equally effective in preventing psychotic relapse compared to maintenance treatment ¹² , suggesting that a period of 10 weeks may be sufficiently gradual for reducing the dose by half. However, others have advocated even slower tapering, especially when aiming to discontinue low doses ¹⁶ , ¹⁷ .

The idea behind slow tapering is that it may reduce the risk of rebound phenomena due to the development of dopamine supersensitivity following the long‐term antagonism of striatal D2 receptors by antipsychotic treatment ¹⁸ . This supersensitivity may be more pronounced with D2 receptor blockade by high affinity antagonists, compared to low affinity antagonists and partial D2 agonists ¹⁹ , ²⁰ . So far, D2 receptor affinity has not gained particular attention in relation to relapse risk. In a single naturalistic cohort study, a significantly lower hospitalization rate was found for aripiprazole (6.3%, N=17) compared to olanzapine (37.9%, N=29) as the first antipsychotic received for first psychotic episode, but the authors attributed this effect to potential bias by indication (i.e., people who function relatively well may be more likely to be prescribed aripiprazole) ²¹ .

In this study, we aimed to investigate both antipsychotic tapering speed and D2 receptor affinity as predictors of relapse risk and time to psychotic relapse in patients in remission from a first psychotic episode. We hypothesized that a higher tapering speed would be associated with a higher relapse risk, and a shorter time between end of tapering and relapse. In addition, we hypothesized that use of high D2 affinity antagonists would be associated with a greater relapse risk than use of low D2 affinity antagonists or partial D2 agonists, and with a shorter time to relapse after complete discontinuation. We included 227 individuals recently remitted from a first psychotic episode, who tapered antipsychotic medication as part of the Handling Antipsychotic Medication: Long‐term Evaluation of Targeted Treatment (HAMLETT) trial, the largest randomized controlled trial worldwide on antipsychotic dose reduction/discontinuation after a first psychotic episode ²² .

METHODS

Study design and participants

The data for this study are part of the HAMLETT study, a multicentre single‐blind randomized controlled trial of antipsychotic medication reduction/discontinuation, whose protocol has been previously described in detail ²² . All participants provided written informed consent after oral and written explanation of the study. Ethical approval was obtained from the research and ethics committee of the University Medical Center Groningen (NL 62202.042.17).

Patients were screened by clinicians from 26 specialized psychosis centers throughout the Netherlands, and were included in the HAMLETT trial if they: a) were aged 16 to 60 years; b) had received a DSM‐5 diagnosis of first episode of schizophrenia, schizoaffective disorder, schizophreniform disorder, brief psychotic disorder, or unspecified schizophrenia spectrum and other psychotic disorder; c) had been in symptomatic remission for 3‐6 months (i.e., there had been sustained improvement of psychotic symptoms, and any remaining psychotic symptoms did not interfere with daily functioning) according to their treating psychiatrist; d) were using antipsychotic medication. Patients were excluded from the trial if they had displayed harmful behavior during the psychotic episode or if they needed coercive treatment. Eligibility criteria were assessed by trained researchers using the Comprehensive Assessment of Symptoms and History interview (CASH) ²³ . Up to one year after randomization, each diagnosis was reassessed on the basis of information collected from the clinical file, the CASH interview, and the Global Assessment of Functioning scale (GAF) ²⁴ at different time points, as well as (when needed) additional information from the treating clinician.

Participants were randomized with a 1:1 ratio to maintenance (≤25% dose reduction for at least 12 months) or dose reduction/discontinuation (>25% dose reduction within 6 months). Participants and their clinicians were informed of the randomization outcome. Those randomized to the dose reduction/discontinuation group received recommended tapering schedules based on gradual discontinuation in a period of 3‐6 months (see Table 1). Data were collected between September 2017 and May 2024. They concerned the screening, the baseline measurement at 3‐6 months after remission, and follow‐up visits at 3, 6 and 12 months after baseline, i.e. up to 18 months after remission.

Table 1.

HAMLETT antipsychotic tapering schedules (all doses are in mg/day)

	Max. dose	Available doses	Start tapering	After 2w	After 4w	After 6w	After 8w	After 10w	After 12w	After 14w	After 16w	After 18w	After 20w	After 22w	After 24w	After 26w	After 28w	After 30w
Risperidone	10	0.5, 1, 2, 3, 4, 6	10	8	6	5	4	3	2	1.5	1	0.5	0.5	0.25	0.25	0.25**	0.25**	stop
Olanzapine	20	2.5, 5, 10, 15, 20	20	17.5	15	12.5	10	7.5	5	5	2.5	2.5	1.25	1.25	1.25**	1.25**	stop	stop
Quetiapine	800	25, 100, 200, 300	800	600	500	400	300	200	150	100	75	50	25	12.5	12.5	stop	stop	stop
Aripiprazole	20	5, 10, 15	20	17.5	15	12.5	10	10	7.5	7.5	5	5	2.5	2.5	2.5**	2.5**	stop	stop
Haloperidol	20	1, 5, 10	16	12	10	8	6	4	3	2	1	1	0.5	0.5	0.5**	0.5**	stop	stop
Zuclopenthixol	40	2, 10, 25	40	32	28	24	20	16	12	8	6	4	2	1	1	1**	1**	stop
Sulpiride	800	400, 50*	800	600	500	450	400	350	300	250	200	150	100	50	50	50**	50***	stop
Paliperidone	12	3*, 6*, 9*	12	12**	9	9**	9**	6	6	6**	3	3	3**	3**	3***	stop	stop	stop
Pimozide	20	1, 4	20	16	12	10	8	6	4	3	2	1	1	0.5	0.5	0.5**	0.5**	stop
Lurasidone	148	18.5, 37, 74	148	111	92.5	92.5	74	74	55.5	55.5	37	37	18.5	18.5	9.25	9.25	9.25**	stop
Clozapine	900	12.5*, 25, 100, 200	900	700	500	400	350	300	250	200	150	100	50	25	25	12.5	12.5	stop
Amisulpride	800	50, 100, 200, 400	800	700	600	500	400	350	300	250	200	150	100	50	25	25	25**	stop
Brexpiprazole	4	1*, 2*, 3*, 4*	4	3	2	2	1	1	1**	1**	1***	1***	1***	stop	stop	stop	stop	stop
Flupentixol	18	0.5, 1, 3, 5	18	15	12	9	6	4	3	1.5	1	1	0.5	0.5	0.25	0.25	0.25**	stop
Cariprazine	6	1.5*, 3*, 4.5*, 6*	6	4.5	4.5	3	3	1.5	1.5	1.5**	1.5**	1.5***	1.5***	1.5***	stop	stop	stop	stop

w – weeks, * undividable, ** one dose on alternate days, *** one dose every 4 days

In this specific study, we included HAMLETT participants who tapered antipsychotic treatment between remission and 12‐month follow‐up, who did not discontinue antipsychotics abruptly or without supervision, and for whom sufficient pharmacy data were available to calculate the speed of the first tapering trajectory after remission (N=304). Participants were excluded from the study if: a) there was no relapse information for the entire observation period (N=1); b) dose reduction was not initiated within the first 12 months after remission (N=39); c) only one dose reduction step was taken on a single day during the first 12 months after remission (e.g., a dose correction from 15 to 12.5 mg olanzapine/day) (N=37).

Assessments

Demographic characteristics, cannabis use in the month before baseline and during the 12‐month follow‐up, and global functioning and symptomatology at all time points were assessed with the CASH interview, the GAF questionnaire, and the Positive and Negative Syndrome Scale (PANSS) ²⁵ , used by raters who were blinded to the patient's condition.

The frequency of contact with a clinical team member (psychiatrist, psychologist or nurse) in the month prior to baseline was assessed with the Treatment Inventory of Costs in Patients with psychiatric disorders (TIC‐P) questionnaire ²⁶ . The duration of the first psychotic episode was calculated as the number of days between the dates of onset and remission, which were provided by clinicians from electronic patient records. Medication use was evaluated by means of a self‐report questionnaire, and detailed data on medication dispense were provided by the Foundation for Pharmaceutical Statistics, from which olanzapine equivalents were calculated ²⁷ , ²⁸ , ²⁹ .

Relapse was defined dichotomously on the basis of hospitalization for psychosis, explicit clinical judgement of the treating psychiatrist, or an increase of ≥12 points in the total score of the PANSS, or a score ≥4 and a ≥1 point increase on at least one positive PANSS item, including P1 (delusions), P2 (conceptual disorganization), P3 (hallucinations), G5 (mannerisms/posturing) or G9 (unusual thought content), during one of the follow‐up assessments compared to baseline ³⁰ .

Tapering and categorization of antipsychotic use

The tapering trajectory was determined for the first 12 months after remission. Tapering speed (in olanzapine equivalents mg/day) was defined as the difference of antipsychotic dose between the start and the end of tapering, divided by the number of days in between. A tapering period started after remission, when the participant first assumed medication at a reduced dose relative to the prior prescribed one. This period lasted until the date on which the participant completely discontinued medication or assumed the lowest dose that either preceded dose increase or remained stable for at least four months.

The total time on antipsychotic medication was calculated as the number of days between remission and the end of tapering. In case multiple antipsychotics were used simultaneously, dose reduction was regarded as the decrease in the combined olanzapine dose equivalent of the antipsychotics. Additionally, tapering speed was categorized as either slower or faster than the recommended HAMLETT tapering schedules (see Table 1), which advise tapering 10 mg of daily oral olanzapine over 20 weeks (cut‐off: 10 mg/day / 140 days = 0.071 mg/day).

Antipsychotic drugs were classified into “high D2 affinity antagonists” (K_i < 10 nM), including flupentixol, haloperidol, pimozide, risperidone, zuclopenthixol, sulpiride, paliperidone, penfluridol, amisulpride and lurasidone; “low D2 affinity antagonists” (K_i ≥ 10 nM), including olanzapine, clozapine and quetiapine; and “partial D2 agonists”, including aripiprazole, brexpiprazole and cariprazine ³¹ .

As participants were not randomized by type of antipsychotic drug, inverse probability of treatment weighting (IPTW) was applied to correct for bias by indication and balance baseline clinical and sociodemographic characteristics between the groups of antipsychotics ³² .

Data analysis

Logistic regression analysis with IPTW was performed to assess whether the risk of relapse was associated with antipsychotic tapering speed (both continuous in olanzapine equivalents mg/day, and in the two categories of faster/slower than recommended tapering speed), and with D2 receptor affinity of the used antipsychotic. High D2 affinity antagonist use was included as the reference group, and post‐hoc comparisons were performed to assess whether the risk of relapse differed in users of low D2 affinity antagonists versus D2 partial agonists.

The model was corrected for sex at birth, duration of first psychotic episode (in days), cannabis use between baseline and 12‐month follow‐up ³³ , and the start and end dose of tapering. Mean imputation was applied in case of missing duration of first psychotic episode (N=2) or duration of antipsychotic use (N=1). The most extreme values of the duration of first psychotic episode (N=2) and time on antipsychotics (N=2) were winsorized by replacing them with the 95th percentile value, maintaining their relative ranking while reducing their statistical impact ³⁴ . Sensitivity analyses included a logistic regression with IPTW to test the relationship of antipsychotic tapering speed and D2 affinity with the risk of hospitalization.

A Cox proportional hazards analysis with IPTW was performed to assess whether tapering speed, antipsychotic D2 affinity, and the start and end dose of the tapering trajectory were associated with the time to relapse. Time to relapse was defined as the number of days between the end of the tapering trajectory and the onset of the first relapse. If participants did not relapse, they were included as censored with the time to no relapse being the number of days between the end of tapering and the end of follow‐up. Participants who experienced a relapse during the tapering trajectory (N=13) were included in the Cox proportional hazards analysis with a time to relapse of zero. The model was corrected for sex at birth, cannabis use, duration of first psychotic episode, antipsychotic dose at the start and end of the tapering trajectory, and the duration of antipsychotic medication use (in days) from remission to the end of tapering.

Statistical analyses were all performed in R (version 4.3.2) via Rstudio (version 2023.12.1.402) ³⁵ .

RESULTS

The study included 227 participants in remission from a first psychotic episode who tapered antipsychotic medication, with 32.2% females (N=73) and an average duration of the episode of 193 days. The sociodemographic and clinical characteristics of the participants are shown in Table 2.

Table 2.

Sociodemographic, clinical and tapering characteristics of the study population

	No relapse (N=123)	Relapse (N=104)	Overall (N=227)
Sex at birth, N (%)
Male	79 (64.2)	75 (72.1)	154 (67.8)
Female	44 (35.8)	29 (27.9)	73 (32.2)
Age (years), mean±SD	28.2±8.8	27.0±7.9	27.6±8.4
Education (years), mean±SD	14.4±2.2	14.2±2.0	14.3±2.1
Duration of first psychotic episode (days), mean±SD	147±178	248±617	193±440
Diagnosis, N (%)
Schizophreniform disorder	42 (34.1)	27 (26.0)	69 (30.4)
Schizophrenia	41 (33.3)	56 (53.8)	97 (42.7)
Schizoaffective disorder	36 (29.3)	17 (16.3)	53 (23.3)
Brief psychotic disorder	4 (3.3)	4 (3.8)	8 (3.5)
Mental health care contacts (per month), mean±SD	4.01±6.50	3.13±3.38	3.60±5.31
Baseline PANSS score, mean±SD
Total	43.6±8.8	45.4±10.6	44.4±9.7
Positive	8.6±2.2	9.2±2.8	8.9±2.5
Negative	11.8±4.1	12.4±4.4	12.1±4.2
General	23.1±5.0	23.8±5.9	23.4±5.5
Baseline GAF score, mean±SD	65.7±10.3	65.1±13.1	65.4±11.6
Cannabis use, N (%)
Yes	33 (26.8)	32 (30.8)	65 (28.6)
No	90 (73.2)	72 (69.2)	162 (71.4)
Antipsychotic drugs used, N (%)
High affinity D2 antagonists	24 (19.5)	33 (31.7)	57 (25.1)
Low affinity D2 antagonists	67 (54.5)	49 (47.1)	116 (51.1)
Partial agonists	32 (26.0)	22 (21.2)	54 (23.8)
Time between remission and start of tapering (days), mean±SD	103±74.4	105±65.9	103±70.5
Antipsychotic dose at the start of tapering, mean±SD	12.6±6.2	12.9±6.7	12.7±6.4
Antipsychotic dose at the end of tapering, mean±SD	1.9±3.7	1.5±2.9	1.8±3.3
Duration of tapering (days), mean±SD	128±77.8	118±76.2	124±77.0
Average dose reduction per day, mean±SD	0.121±0.126	0.147±0.143	0.133±0.135
Percentage of tapered dose, mean±SD	87.5±21.8	89.1±19.1	88.3±20.6
Tapering type, N (%)
Dose reduction	37 (30.1)	32 (30.8)	69 (30.4)
Full discontinuation	86 (69.9)	72 (69.2)	158 (69.6)

Antipsychotic doses are reported in olanzapine equivalents mg/day. PANSS – Positive and Negative Syndrome Scale, GAF – Global Assessment of Functioning

Antipsychotic tapering characteristics

Of all participants, 23.8% used partial D2 agonists (N=54), 51.1% used low D2 affinity antagonists (N=116), and 25.1% used high D2 affinity antagonists (N=57). The average antipsychotic dose was 12.7±6.4 mg/day olanzapine equivalents at the start of tapering, and 1.8±3.3 mg/day olanzapine equivalents at the end of the tapering trajectory. The average dose reduction at the end was 88.3±20.6% of the starting dose, and 69.6% (N=158) of the participants fully discontinued their antipsychotic medication. The sociodemographic and clinical characteristics of the three antipsychotic D2 affinity groups are presented in Table 3.

Table 3.

Sociodemographic, clinical and tapering characteristics per antipsychotic D2 affinity group

	High D2 affinity antagonists (N=57)	Low D2 affinity antagonists (N=116)	Partial D2 agonists (N=54)	Overall (N=227)
Sex at birth, N (%)
Male	42 (73.7)	79 (68.1)	33 (61.1)	154 (67.8)
Female	15 (26.3)	37 (31.9)	21 (38.9)	73 (32.2)
Age (years), mean±SD	27.6±10.1	27.7±8.0	27.5±7.3	27.6±8.4
Education (years), mean±SD	14.0±2.2	14.4±1.9	14.5±2.3	14.3±2.1
Duration of first psychotic episode (days), mean±SD	277±785	138±185	222±290	193±440
Baseline PANSS score, mean±SD
Total	44.9±9.7	44.3±9.9	44.2±9.6	44.4±9.7
Positive	9.1±2.7	8.7±2.3	9.0±2.6	8.9±2.5
Negative	12.8±4.6	12.1±4.1	11.4±4.1	12.1±4.2
General	23.1±4.8	23.5±5.9	23.8±5.3	23.4±5.5
Baseline GAF score, mean±SD	65.2±11.9	65.2±11.8	66.2±11.2	65.4±11.6
Antipsychotic dose at the start of tapering, mean±SD	9.9±6.6	13.4±5.1	14.2±7.7	12.7±6.4
Antipsychotic dose at the end of tapering, mean±SD	1.5±2.6	1.6±3.0	2.3±4.6	1.8±3.3
Duration of tapering (days), mean±SD	104±78.6	140±75.9	108±71.1	124±77.0
Average dose reduction per day, mean±SD	0.153±0.195	0.116±0.110	0.149±0.0989	0.133±0.135
Percentage of tapered dose, mean±SD	87.9±19.9	89.1±19.0	86.9±24.6	88.3±20.6
Tapering type, N (%)
Dose reduction	19 (33.3)	36 (31.0)	14 (25.9)	69 (30.4)
Full discontinuation	38 (66.7)	80 (69.0)	40 (74.1)	158 (69.6)

Antipsychotic doses are reported in olanzapine equivalents mg/day. PANSS – Positive and Negative Syndrome Scale, GAF – Global Assessment of Functioning

The average tapering speed of all participants was 0.133±0.135 mg/day olanzapine equivalents, corresponding to a discontinuation of 10 mg/day olanzapine over 75 days. On average, a tapering trajectory lasted 124±77 days (see Table 2). The fastest tapering trajectory was a dose reduction of 4.88 mg/day olanzapine equivalents to full discontinuation in 6 days. The longest tapering trajectory spanned 334 days, in which a dose of 9.52 mg/day olanzapine equivalents was reduced to 2.38 mg/day (see also supplementary information).

When categorizing the tapering speed, we found that 30.8% tapered slower and 69.2% tapered faster than the recommended tapering trajectory (see Table 1). Among all participants, tapering was finished by 0.9% (N=2) within one week, by 3.96% (N=9) within two weeks, by 7.5% (N=17) within one month, and by 25.1% (N=57) within two months. Tapering lasted between 3 and 6 months for 50.2% (N=114) and lasted longer than 6 months for 24.7% (N=56) of participants.

Tapering speed, antipsychotic D2 receptor affinity, and risk of relapse

Within the first 18 months after remission from the first psychotic episode, 45.8% (N=104) of participants experienced a relapse after dose reduction or complete discontinuation, and 16.3% (N=37) were hospitalized. Logistic regression analysis showed that the speed of antipsychotic tapering (in olanzapine equivalents mg/day) was not associated with the risk of relapse (z=0.989, odds ratio, OR=2.86, 95% CI: 0.38‐26.24, p=0.323) (see Table 4). When tapering speed was dichotomized as faster or slower than recom‐ mended in the tapering schedules, it still did not significantly relate to the risk of relapse (z=−1.058, OR=0.71, 95% CI: 0.37‐1.33, p=0.290) (see also supplementary information). Logistic regression an‐ alyses in participants who fully discontinued antipsychotic medication, or which used hospitalization as outcome, also showed no association with tapering speed (see supplementary information).

Table 4.

Logistic regression analysis on the relationship of tapering speed and antipsychotic D2 affinity with relapse rate in patients remitted from first psychotic episode (N=227)

	z value	Odds ratio	95% CI	p
Intercept	−0.404	0.85	0.39‐1.85	0.686
Tapering speed	0.989	2.86	0.38‐26.24	0.323
Low D2 affinity antagonist*	−2.104	0.48	0.24‐0.95	0.035
Partial D2 agonist*	−2.278	0.44	0.21‐0.88	0.023
Start dose (olanzapine equivalents mg/day)	1.267	1.03	0.98‐1.08	0.205
End dose (olanzapine equivalents mg/day)	−1.666	0.92	0.84‐1.01	0.096
Duration of first psychotic episode (days)	1.179	1.00	1.00‐1.00	0.239
Cannabis use (yes)	−0.426	0.88	0.47‐1.61	0.670
Sex at birth (female)	−0.161	0.95	0.52‐1.75	0.872

^*reference category: high D2 affinity antagonist; significant p values are highlighted in bold prints

A significantly lower risk of relapse was observed in users of low D2 affinity antagonists (z=−2.104, OR=0.48, 95% CI: 0.24‐0.95, p=0.035) and partial D2 agonists (z=−2.278, OR=0.44, 95% CI: 0.21‐0.88, p=0.023) compared to those using high D2 affinity antagonists (see Table 4). There was no difference in the risk of relapse between persons using partial D2 agonists and low D2 affinity antagonists (z=0.291, OR=1.10, 95% CI: −0.57 to 2.17, p=0.771). Overall, 57.9% (N=33) of users of high D2 affinity antagonists, 42.2% (N=49) of users of low D2 affinity antagonists, and 40.7% (N=22) of users of D2 agonists relapsed during the follow‐up period (see Figure 1).

Figure 1.

Antipsychotic D2 receptor affinity and prevalence of relapse (N=104) and no relapse (N=123) in participants who tapered antipsychotic medication within 18 months after remission

The risk of relapse was marginally associated with antipsychotic dose at the end of the tapering trajectory (z=−1.666, OR=0.92, 95% CI: 0.84‐1.01, p=0.096), but not with antipsychotic dose at the start of the trajectory (z=1.267, OR=1.03, 95% CI: 0.98‐1.08, p=0.205), duration of first psychotic episode (z=1.179, OR=1.00, 95% CI: 1.00‐1.00, p=0.239), cannabis use (z=−0.426, OR=0.88, 95% CI: 0.47‐1.61, p=0.670) and sex at birth (z=−0.161, OR=0.95, 95% CI: 0.52‐1.75, p=0.872) (see Table 4).

Tapering speed, antipsychotic D2 receptor affinity, and time to relapse

The Cox proportional hazards model showed that tapering speed did not relate to the time to relapse (z=0.755; hazard ratio, HR=1.55, 95% CI: 0.50‐4.83, p=0.450) (see Table 5). The tapering speed categorized as faster or slower than the recommended tapering speed also was not related to the time to relapse (z=−0.411, HR=0.89, 95% CI: 0.50‐1.57, p=0.681) (see also supplementary information). Logistic regression analysis including only participants who fully discontinued antipsychotic medication showed similar results (see supplementary information).

Table 5.

Cox proportional hazards regression analysis on the relationship of tapering speed and antipsychotic D2 affinity with time to relapse in patients remitted from first psychotic episode (N=227)

	z value	Hazard ratio	95% CI	p
Tapering speed	0.755	1.55	0.50‐4.83	0.450
Low D2 affinity antagonist*	−2.211	0.57	0.34‐0.94	0.027
Partial D2 agonist*	−2.054	0.54	0.30‐0.98	0.040
Start dose (olanzapine equivalents mg/day)	1.157	1.02	0.98‐1.07	0.247
End dose (olanzapine equivalents mg/day)	−1.668	0.94	0.87‐1.01	0.095
Duration of first psychotic episode (days)	0.906	1.00	1.00‐1.01	0.365
Cannabis use (yes)	−0.172	0.96	0.58‐1.58	0.863
Sex at birth (female)	0.118	1.03	0.62‐1.71	0.906
Duration of antipsychotic use (days)	−0.296	1.00	1.00‐1.00	0.767

^*reference category: high D2 affinity antagonist; significant p values are highlighted in bold prints

The time to relapse was shorter for users of high affinity D2 antagonists than users of low affinity D2 antagonists (z=−2.211, HR=0.57, 95% CI: 0.34‐0.94, p=0.027) and partial D2 agonists (z=−2.054, HR=0.54, 95% CI: 0.30‐0.98, p=0.040) (see Table 5). The time to relapse did not differ between users of low affinity D2 antagonists and partial D2 agonists (z=0.209, HR=1.06, 95% CI: 0.62‐1.80, p=0.835). The restricted mean survival time at 18 months was 280 days for high affinity D2 antagonists, 351 days for low affinity D2 antagonists, and 357 days for partial D2 agonists.

DISCUSSION

This is the first study to investigate the relationship of both tapering speed and antipsychotic D2 receptor affinity with risk of psychotic relapse and time to relapse in a large sample of individuals remitted from a first psychotic episode, who were followed over the course of 18 months with a detailed reconstruction of the first tapering trajectory based on pharmacy dispensation data. Of 227 participants who reduced (30.4%) or discontinued (69.6%) antipsychotic treatment within 18 months after remission, 45.8% (N=104) experienced a relapse and 16.3% (N=37) were hospitalized. Our results indicate that antipsychotic D2 affinity, but not tapering speed, was associated with relapse risk. Specifically, individuals who tapered off high D2 affinity antagonists were at greater risk of relapse and experienced relapse sooner after discontinuation than those who tapered off low D2 affinity antagonists or partial D2 agonists.

The greater relapse risk and shorter time to relapse observed in high D2 affinity antagonist users align with the dopamine supersensitivity hypothesis. This hypothesis states that long‐term striatal D2 receptor blockade by antipsychotics may result in a compensatory upregulation of these receptors ²⁰ , ³⁶ , ³⁷ . Discontinuation or dose reduction of antipsychotic drugs may consequently lead to enhanced dopamine‐mediated signalling and subsequent psychotic exacerbation, referred to as “supersensitivity psychosis” ²⁰ , ³⁶ , ³⁷ . Indeed, rodent studies showed that high affinity D2 antagonists, such as haloperidol and risperidone, induced dopamine supersensitivity ³⁸ , whereas the low affinity D2 antagonist olanzapine did not ³⁹ .

Partial D2 agonists may pose an exception to the dopamine supersensitivity hypothesis. Despite their high D2 affinity, they probably do not induce compensatory D2 receptor upregulation, due to their partial agonistic effect ²⁰ . Animal studies have demonstrated that exposure to aripiprazole and brexpiprazole, in contrast to haloperidol, did not relate to increased D2 receptor density ¹⁹ , ³⁸ . In fact, aripiprazole has even been suggested to prevent the development of dopamine supersensitivity ¹⁹ .

Low D2 affinity antipsychotics are available in a broader range of dosages compared to high D2 affinity medications, allowing for a more gradual tapering trajectory and lower dopamine receptor occupancy with the lowest doses, thereby possibly reducing the likelihood of dopamine supersensitivity at the end of tapering. Animal studies confirm that low‐dose antipsychotics do not seem to induce dopamine supersensitivity ⁴⁰ , ⁴¹ .

Aripiprazole stands out not only due to its partial D2 agonist properties, but also for its notably long half‐life, which may be beneficial in tapering contexts. This drug has a half‐life of approximately 75 hours, and the half‐life of its active metabolite is around 94 hours. Such a long half‐life may help reduce the risk of withdrawal and relapse symptoms compared to antipsychotics with shorter half‐lives ⁴² . With other psychiatric medications, tapering protocols often involve switching from drugs with shorter half‐lives to others with longer ones before gradually reducing the dose, in order to minimize withdrawal symptoms, as seen with the substitution of short‐acting benzodiazepines with diazepam ⁴³ . Similarly, for antidepressants with short half‐lives, discontinuation symptoms are frequently reported ⁴² , and substitution with a long‐acting antidepressant, such as fluoxetine, is recommended ⁴⁴ . Aripiprazole's long half‐life, combined with its partial agonistic effects that seem to counter dopamine supersensitivity, may contribute to its lower relapse risk observed in our study. Positron emission tomography (PET) studies may clarify whether aripiprazole's partial agonism, extended half‐life, or the combination of both, supports relapse prevention during tapering. Interestingly, all available partial agonists have similarly long half‐lives and produce active metabolites with long‐acting partial agonists effects ⁴⁵ . While further research is needed to identify the ideal pharmacological profile for tapering, current evidence supports partial D2 agonists as preferable options.

Three considerations are important for interpreting our findings. First, since participants were not randomized by type of antipsychotic drug, there is a risk of bias by indication. However, participants using partial D2 agonists, and low and high affinity D2 antagonists were comparable in sociodemographic and clinical characteristics. For instance, baseline symptom severity was similar across the groups, with an average PANSS total score of 44.9 for high D2 affinity antagonist, 44.3 for low D2 affinity antagonist, and 44.2 for partial D2 agonist users (see Table 3). Although we do not have direct data on the severity of the initial psychosis, the dose at the start of tapering – often corresponding to the dose at remission – was included as a covariate and can serve as a proxy for illness severity. To reduce the potential bias by indication, we applied IPTW, a statistical method that adjusts for differences in clinical and sociodemographic characteristics, effectively serving as a statistical alternative for patient matching. While IPTW balances observed characteristics across the three D2 affinity groups, the findings may potentially still be influenced by unmeasured characteristics.

Second, there may be concerns regarding the limited range of tapering speeds in our study, as standard tapering trajectories were recommended as part of the study. Our results show that the average tapering speed of all participants was 0.133 mg/day olanzapine equivalents (range: 0.02‐0.81), which corresponds to the discontinuation of 10 mg/day olanzapine over 75 days. This rate was nearly twice as fast as recommended by the HAMLETT tapering guide that participants and their clinicians received from the study team, which advises discontinuing 10 mg/day olanzapine in 140 days (see Table 1). Notably, 69.2% of participants tapered faster than the recommended trajectory, showing that there was a considerable range. Despite this substantial range, tapering speed was not a significant predictor of relapse.

It has been suggested that, when antipsychotics are tapered more gradually, the relapses are more evenly distributed over time ¹⁶ . This would mean that, with more gradual tapering and longer follow‐up times, the association between tapering speed and risk of relapse decreases. This is further illustrated by a recent study on paliperidone discontinuation ⁴⁶ , which did not find a relationship between the rate of discontinuation (1‐monthly versus 3‐monthy long‐acting injectable switch to placebo) and relapse. The authors concluded that the absolute levels of D2 receptor occupancy, rather than the rate of occupancy change, may be related to relapse. This implies that longer tapering trajectories may only extend the time to relapse by prolonging the period until sub‐threshold levels of antipsychotic prophylaxis are reached ⁴⁶ . However, this does not negate the risks associated with rapid tapering. Compared to abrupt discontinuation, a gradual tapering approach aiming for the lowest possible dose remains preferable ¹² , ⁴⁷ . Based on our current results, it might be unnecessary to adhere to an extremely slow tapering process, thereby prolonging exposure to antipsychotics.

Third, it is important to emphasize that the participants in this study were all in remission from a first psychotic episode, functioned relatively well, and received regular care in well‐resourced facilities in the Netherlands that could provide close monitoring during the tapering trajectory. Generally, those who reach remission after a first psychotic episode have better long‐term outcomes than those who do not ⁴⁸ , which limits the generalizability of our findings to those who are not able to reach remission. Furthermore, the relapse rate in our study (45.8%) is comparable to naturalistic antipsychotic maintenance cohorts (pooled prevalence of 43%, 1.5‐2 years after first psychotic episode) ¹ , and even lower than other discontinuation trials (pooled prevalence of 57%, 18 months after first psychotic episode) ⁴⁹ . This relatively low relapse rate could be a consequence of the frequency of monitoring and the early psychosis care in the Netherlands. For example, participants had contact with a clinical team member (i.e., a psychiatrist, psychologist or nurse) on average 3.6 times per month. Nearly all participants were recruited from early intervention services, which have demonstrated to reduce relapse rates and symptom levels, as well as improving occupational functioning ⁵⁰ .

In the present analyses, we were unable to differentiate between relapses and withdrawal psychosis. A rapid, short‐lived (i.e., <6 weeks) and usually reversible psychosis occurring soon after antipsychotic dose reduction, discontinuation or switching is considered a withdrawal psychosis ⁵¹ . Withdrawal symptoms are not only a reflection of dopamine supersensitivity, but also of imbalances in other neurotransmitter systems following dose reduction or discontinuation, including histamine, acetylcholine and serotonin. A recent pharmacovigilance study explored the relationship between neurotransmitter receptor affinities and the risk of reporting withdrawal symptoms, but did not find a significant association ⁵² .

In our study, we took several steps to minimize the occurrence of withdrawal psychosis and to reduce potential bias from conflation between withdrawal and relapse ⁵³ . For instance, our definition of relapse was based on the most recent guidelines ²⁶ , and included clinical judgement of treating clinicians as an additional source. Before tapering, we also informed clinicians, participants and their relatives about the potential for withdrawal symptoms. Nevertheless, future studies would greatly benefit from more detailed analyses that distinguish withdrawal from relapse, ideally employing standardized methods yet to be developed and validated.

It is also to be considered that we calculated the tapering speed as an average reduction in olanzapine equivalents mg/day over the entire trajectory, while the speed of the last steps and the shape of the trajectory may be important for the relapse risk as D2 receptor occupancy falls below the therapeutic threshold.

In conclusion, our study highlights several important considerations for clinicians, patients and relatives when tapering off antipsychotic medication after reaching remission from a first psychotic episode. A gradual tapering approach remains preferable, but adhering to an extremely slow tapering process may not be necessary. Particular caution is needed when tapering high D2 affinity antagonists, such as haloperidol or risperidone. For patients already stabilized on strong D2 antagonists, extra monitoring during tapering is warranted. It may be prudent to initiate treatment in first psychotic episode with a partial D2 agonist or a low D2 affinity antagonist, given their potentially more favorable relapse risk profile when considering dose reduction or discontinuation after remission.