A randomised double-blind placebo-controlled trial of minocycline and/or omega-3 fatty acids added to treatment as usual for at risk Mental States: The NAYAB study

: Inflammatory mechanisms are thought to contribute to the onset of psychosis in persons with an at-risk mental state (ARMS). We investigated whether the anti-inflammatory properties of minocycline and omega-3 polyunsaturated fatty acids (omega-3), alone or synergistically, would prevent transition to psychosis in ARMS in a randomised, double-blind, placebo-controlled trial in Pakistan. Methods: 10,173 help-seeking individuals aged 16 – 35 years were screened using the Prodromal Questionaire-16. Individuals scoring 6 and over were interviewed using the Comprehensive Assessment of At-Risk Mental States (CAARMS) to confirm ARMS. Participants (n = 326) were randomised to minocycline, omega-3, combined minocycline and omega-3 or to double placebo for 6 months. The primary outcome was transition to psychosis at 12 months. Findings: Forty-five (13.8 %) participants transitioned to psychosis. The risk of transition was greater in those randomised to omega-3 alone or in combination with minocycline (17.3.%), compared to 10.4 % in those not exposed to omega-3; a risk-ratio (RR) of 1.67, 95 % CI [0.95, 2.92] p = 0.07. The RR for transitions on mino-cycline vs. no minocycline was 0.86, 95 % CI [0.50, 1.49] p > 0.10. In participants who did not become psychotic, CAARMS and depression symptom scores were reduced at six and twelve months (mean CAARMS difference = 1.43; 95 % CI [0.33, 1.76] p < 0.01 in those exposed to omega-3. Minocycline did not affect CAARMS or depression scores. Interpretation: In keeping

transition to psychosis appear to have distinct pathogeneses which are differentially modulated by omega-3 supplementation.Funding: The study was funded by the Stanley Research Medical Institute.

Introduction
The onset of psychosis in schizophrenia is typically preceded by a prodromal phase of mild symptoms and deteriorating function.Clinical criteria have been developed to identify individuals at high risk of developing a psychotic disorder such as the at-risk mental state (ARMS) (Yung et al., 1996) or clinical high-risk (Fusar-Poli, 2017).Initial studies suggested that approximately 35 % of individuals with ARMS develop a psychotic illness within a year (Cannon et al., 2008) but these rates have fallen, with a meta-analysis suggesting that this figure is closer to 15 % (Salazar de Pablo et al., 2021).Although most individuals with ARMS will not transition to psychosis, many will experience depression and persistent ARMS symptoms (Salazar de Pablo et al., 2021).Consequently, the ARMS period provides an opportunity to intervene to prevent or delay conversion to psychosis and improve longer-term outcomes.
Several preventative treatments have been evaluated in clinical trials for the ARMS population.Cognitive behaviour therapy has the strongest evidence base for efficacy in preventing transition to psychosis compared to treatment as usual but not in comparison with other psychosocial therapies, antipsychotic or experimental drug treatments (Davies et al., 2018;Devoe et al., 2020).However, there are too few trials to draw definitive conclusions about overall or comparative efficacy of specific treatments.Three RCTs have evaluated dietary supplementation with omega-3 polyunsaturated fatty acids (omega-3) in preventing onset of psychosis (Amminger et al., 2010;Cadenhead et al., 2017;McGorry et al., 2017).The first placebo-controlled trial of omega-3 supplementation in ARMS, the Vienna study (Amminger et al., 2010), reported a transition rate of 11 of 40 (27.5 %) in the placebo group compared to 2 of 41 (4.9 %) in the omega-3 group at twelve month follow up (p = 0.007).The NAPLS study (Cadenhead et al., 2017) and the NEURAPRO study (McGorry et al., 2017), failed to show significant effects of omega-3 supplementation on transition rate although the NAPLS study found a significant association between a baseline diet low in omega-3 foods and later conversion to psychosis.
The rationale for the use of omega-3 was to reverse suspected dietary deficiencies in psychosis that affect the lipid composition and properties of cell membranes and more recently that they inhibit cyclo-oxygenase 2 and formation of inflammatory prostaglandins (Berger et al., 2019).Studies of circulating cytokines indicate schizophrenia is associated with mild activation of the peripheral immune system (Pillinger et al., 2019) which might be a cause or consequence of inflammation in the brain.The possibility that an inexpensive and safe dietary supplement might reverse inflammatory-metabolic mechanisms of psychosis was of particular interest to our evaluation of treatments in Pakistan; a country with prevalent poverty, communicable disease and food insecurity predisposing to dietary deficiency and inflammatory disorders.
We included a minocycline treatment arm as early studies in schizophrenia reported benefit on negative symptoms in persons with established and recent-onset schizophrenia (Miyaoka et al., 2008) including a study from our group in Pakistan with a centre in Brazil (Chaudhry et al., 2012).The selection of minocycline was encouraged by its specific ability to reduce inflammatory activation of microgliathe brain's resident inflammatory cells (Sellgren et al., 2019).In addition, there were early reports of microglial activation in patients with schizophrenia, detected in-vivo using positron emission tomography (PET), to image radioligand binding to activated microglia (van Berckel et al., 2008).During the course of the NAYAB study there were two large scale studies of minocycline in schizophrenia that found no effect of minocycline on negative symptoms of schizophrenia (Deakin et al., 2018) or overall symptoms of schizophrenia (Weiser et al., 2019).
However, these studies do not exclude the possibility that minocycline may be effective during a prodromal phase when inflammatory mechanisms may be most active in the pathogenesis of psychosis and there had been no studies of minocycline in an ARMS population to assess for prevention or delay to psychosis transition.
We carried out a double-blind randomised placebo-controlled trial of minocycline and omega-3 to determine their efficacy in preventing transition to psychosis in persons with ARMS.Omega-3 and minocycline are safe and widely available drugs in low and middle income countries (LMICs).Both drugs have anti-inflammatory actions and therefore we hypothesised they might, either individually or synergistically, be effective in preventing transition to psychosis by reducing neuroimmune mechanisms that may mediate the many gene-environmental factors leading to onset of symptoms.

Study design
NAYAB was a multicentre, double-blind, randomised placebocontrolled study of minocycline and omega-3 using a 2 x 2 factorial design.The four treatment groups took either minocycline, omega-3, their combination, or their matching placebos for 6 months followed by a 6-month follow-up period.The study was conducted within three large cities in Pakistan (Lahore, Karachi, and Rawalpindi) with a combined population of approximately forty-six million.The trial was conducted between March 2017 and November 2018.The hypotheses and protocol have been published (Qurashi et al., 2017).Ethical approval for the study was obtained prior to study commencement from the Ethics and Scientific Review Committee of the Karachi Medical and Dental College, Pakistan.The paper was conducted and reported according to the CONSORT checklist (Schulz and Altman, 2010).

Participants and selection of primary outcome measure
Eligible participants were aged 16-35 years who met criteria for one or more of the three sub-types of ARMS: state vulnerability, attenuated psychotic symptoms or brief limited intermittent psychotic symptoms.Symptoms and criteria were assessed using the Comprehensive Assessment of At Risk Mental States (CAARMS) (Yung et al., 2005).The CAARMS is a widely used and validated semi-structured interview for identification of individuals at increased risk of developing a firstepisode psychotic disorder.Ratings of the severity of four groups of positive psychotic symptoms can be summed to produce a continuous 'global' severity measure or added to the sum of the four frequency ratings to produce a 'combined' measure.The distress associated with severity ratings is assessed by a visual analogue rating.
Participants were excluded if there was a history of psychotic illness; diagnosis of learning disability; any pre-existing inflammatory conditions (e.g.rheumatoid arthritis); organic brain disease (e.g.epilepsy); previous treatment with an antipsychotic, mood stabilizing agent or electroconvulsive therapy; prior history of intolerance or serious side effects to tetracycline or omega-3; concomitant penicillin therapy or anticoagulant therapy (to minimise adverse drug events); active substance abuse (except nicotine or caffeine) or dependence within the last three months according to Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (Association and Association, 2013); pregnant or breastfeeding; current or previous treatment with either a tetracycline or omega-3 in the three months before study entry; prescribed antiinflammatory medication at study entry; active expression of suicidal ideation or current aggressive/dangerous behaviour; relevant current or past hematologic, hepatic, renal, neurologic or other medical disorder that might interfere with the study.All participants gave written informed consent to participate.
The study was widely advertised to the general population using radio and newspaper articles, general practitioner engagement and selfreferrals from potential participants and family members.There are no early intervention or other youth mental health services in Pakistan.We screened for the presence of prodromal symptoms using the Prodromal Questionnaire-16 (PQ-16), a brief self-report questionnaire.We used a cut-off value of 6 that has been shown to have high sensitivity (87 %) and specificity (87 %) in identifying ARMS (Ising et al., 2012).Participants with scores of 6 or above were subsequently assessed with the CAARMS to identify those meeting criteria for ARMS.Screening and recruitment to the study was undertaken by a team of trained researchers.

Randomisation and masking
Participants were randomly assigned in a 1:1:1:1 ratio to either placebo, minocycline, omega-3 or minocycline and omega-3 in combination.The trial pharmacist prepared study medicines according to the randomisation code generated by the study statistician in the UK (SL).Research assistants (RAs) delivered study medication to participants.Participants received an equivalent number of over-encapsulated tablets and placebo capsules were manufactured to match the active study drugs.Pill counts were undertaken by RAs to assess adherence and no participant received less than 75 % of the allocated intervention.Participants, research assistants undertaking the assessment measures and the trial statistician were blind to treatment allocation.

Procedures
During the first week of the study participants received a 100 mg capsule of minocycline daily (or matching placebo) then 200 mg of minocycline daily (or matching placebo) for the remaining intervention period; all participants tolerated the 200 mg.We chose to trial 200 mg as this was the dose used in a previous trial in early-phase schizophrenia (Deakin et al., 2018).Two 600 mg capsules of omega-3 gave a daily dose of 1.2 g (720 mg of eicosapentaenoic acid, EPA, and 480 mg of docosahexaenoic acid, DHA) based on trials in ARMS populations (McGorry et al., 2017).Omega-3 has a soft-shell capsule and minocycline is a hardshell capsule with different appearances, so hard and soft-shell matching placebos were provided such that all participants received 2 hard and 2 soft capsules after titration.Research assistants obtained informed consent to the procedures described in the patient information leaflet.No participant received concurrent psychotropic medication or psychological treatment during the study.Participants were seen on four occasions: at baseline, 3, 6, and 12 months after study entry for research assessments measures.

Outcomes
The primary efficacy outcome was conversion to psychotic disorder at 12 months after study entry.This was operationally defined using the positive symptom subscales in the CAARMS as one of: -. i) Unusual thought content held with delusional intensity (global score 6) occurring several times or more per week (frequency and duration score > 3).ii) Non-bizarre ideas held with delusional conviction (global score 6) occurring several times or more per week (frequency and duration score > 3).iii) Perceptual abnormalities in any modality (global score ≥ 5) occurring several times or more per week (frequency and duration score > 3).
iv) Disorganised speech (global score ≥ 5) occurring several times or more per week (frequency and duration score > 4).
Secondary outcome measures included total CAARMS global positive symptom scores post-treatment at 6 and 12 months co-varying for baseline; change in social and occupational functioning at 12 months using Social and Occupational Functioning Assessment Scale (SOFAS) (Goldman et al., 1992); change in severity of depressive symptoms between study entry and at 12 months measured by the Montgomery-Åsberg Depression Rating Scale (MADRS) (Montgomery and Åsberg, 1979).Adverse events and serious adverse events were monitored throughout the study and assessed at each visit during the intervention phase.

Statistical analysis
The study was powered based on an expected transition rate to psychosis of 30 % in the placebo group.We expected a higher transition to psychosis rate in an ARMS population in Pakistan because of the risk exposures mentioned and the poor access to psychosocial interventions and pharmacotherapy.For the study interventions to be considered clinically beneficial, a reduction in transition to psychosis to 15 % was deemed necessary.This required a sample size of 59 per treatment arm to detect placebo-active differences with a power of 80 % and at the 5 % significance level.To allow for a 25 % drop-out rate and to ensure there were 59 completing participants per arm, a minimum sample size of 80 subjects per treatment group was required requiring a total sample size of 320 participants.
The statistical analysis was overseen by study statistician (SL).There were no interim analyses, and all analyses were carried out after the database was completed.The primary outcome measures were risk of psychosis and CAARMS scores by treatment group.All statistical analyses of the clinical outcomes were performed using STATA v.14 and SPSS v.27.When the primary analyses were complete, the treatment code was revealed.
The 2 x 2 factorial analysis for omega-3 compared the proportion or 'risk' of psychosis onsets in 'all omega-3′ exposed participants whether taken alone or in combination with minocycline, versus onset risk in the 'no-omega-3′ participant group taking double placebo or minocycline alone.Chi-square analysis was used to test for significant deviation from equality of risk.Risk ratios (RR) with 95 % confidence intervals (CIs) are presented.Similarly, psychosis onsets in the 'all minocycline' exposed participant group, defined as minocycline alone or in combination with omega-3, were compared with 'no minocycline' groups taking double placebo or omega-3 alone.Pair-wise comparisons between the three active treatment groups versus double-placebo were not planned.We screened for possible additive prevention or other effects of the combined treatment using a chi-square across the 4 treatment groups to detect any significant group deviations in risk.
We used analysis of variance to analyse the post-treatment ratings at 6 and 12 months (repeated 'time' measure) in the non-transitioning group, co-varying for baseline with fixed factors for omega-3 ('all omega-3′ vs. 'no omega-3′) and minocycline ('all minocycline' vs. 'no minocycline').We looked for evidence of synergy in positive interactions between omega-3 and minocycline factors.Last observations for dropouts were carried forward to the post treatment ratings.

Role of the funding source
The study was funded by the Stanley Research Medical Institute.The funder had no role in study design, data collection, data analysis, data interpretation or writing of the report.The corresponding author had full access to all the data in the study and final responsibility for the decision to submit for publication.

Results
Between March 2016 and November 2017, 326 participants were recruited and randomised to the study from a screened sample of 10699.The last 12-month visit was in November 2018.Fig. 1 shows participant flow in the trial.80-82 participants were randomised to each of the four treatment groups.29 participants dropped out of the study resulting in a 91 % retention rate.Drop-outs were evenly distributed between the treatment groups with the exception that 4 participants in the omega-3 + placebo group self-harmed by overdose or self-cutting, requiring brief medical attention but not continued contact with health services.27 of the 29 participants who dropped out did so before their scheduled first follow-up at 3 months; their baseline clinical ratings were carried through the follow-ups for the secondary, intention-to-treat analysis of symptoms and function ratings.
The randomised sample included 60 % male participants with a mean age of 24 years at trial entry.All participants rated positive for attenuated psychotic symptoms (APS), and none had brief limited psychotic symptoms (BLIPS).15 % met criteria for the vulnerability risk group, defined by family history and declining performance.
There were no statistically significant or numerically important differences in age, sex or other social and occupational demographics between the 4 treatment groups (Table 1) nor between the 29 dropouts and study completers (Supplementary Table 1).The baseline mean ratings for psychosis (CAARMS), depression (MADRS), and functional impairment (SOFAS) point to mild-moderate severity across the sample.
Forty-five (13.8 %) participants transitioned to psychosis; 44 assessed via CAARMS criteria and one who was admitted with a psychotic illness prior to the first 3-month CAARMS interview (Fig. 1).All but 7 of the transitions to psychosis had occurred by 3 months and none occurred after 6 months.Those who developed psychosis showed closely similar baseline demographics and clinical ratings compared with those who did not become psychotic (Supplementary Table 1).However, only one third of the onset cases compared to half of the no-onset cases came from nuclear families (p = 0.03).
The rate of psychosis onsets in the double placebo group was low at 11.0 % and marginally lower at 9.8 % in the minocycline alone group whereas onsets were greater in the omega-3 alone (18.8 %) and the combined minocycline + omega-3 group (15.9 %) (Table 2).These group differences in proportion were not statistically significant x 2 (3, 326) = 3.62, p > 0.1.
In those who did not become psychotic, repeated measures analysis of variance of the post-treatment CAARMS global positive symptom ratings (6 and 12 months), with baseline a covariant, revealed a small, apparently beneficial effect of omega-3 exposure on CAARMS scores and no interaction with time (6 vs 12 months).Participants who received omega-3, either alone or combined with minocycline (the 'all omega-3′ group), had significantly lower global CAARMS scores at 6 and 12 months compared to the 'no omega-3′ group (mean difference = 1.43; 95 % CI [0.33, 1.76]; p < 0.01).Fig. 2 highlights the CAARMS total ratings of the omega-3 exposed groups in blue and the non-exposed groups in red; the baseline ratings of 27 of the 29 dropouts and the last (3 month) observation of the remaining 2 are carried forward.The omega-3 effect was equally statistically significant for the Combined and  Distress CAARMS scores.There were no main effects or interactions with time (6 vs. 12 month) for minocycline ('all minocycline' group vs 'no minocycline' group).Exposure to omega-3 was also associated with better MADRS depression ratings which were lower in the 'all omega-3′ group than in the 'no omega-3′ group at the end of treatment at 6 months but not at 12month follow-up (treatment by time interaction p = 0.04).
We explored whether risk group status changed with treatment.There were no BLIPS at entry or in follow-up visits.All participants met APS criteria at baseline and 80 % met APS criteria at 12 months.At 12 months, 15.8 % of the 'all omega-3′ group and slightly fewer (6.8 %) in the 'no omega-3′ group, had become APS free, a significant group difference against the 100 % baseline of the randomised sample x 2 (1, n = 252) = 5.17, p = 0.03).Social and occupational impairment (SOFAS ratings) showed little change in those who remained in the study over the 12 months; 13.1 % improved by 2-3 categories and 14.6 % were rated with good or slight impairment in contrast to none at baseline.SOFAS scores were unaffected by treatment group.There were no group differences in adverse events.

Discussion
The aim of the trial was to test the hypothesis that an inflammatory process mediates the onset of psychosis in those at high risk of developing psychosis.We tested the prediction that six months treatment with minocycline and/or omega-3, either alone or combined, would reduce the transition to psychosis over 12 months in comparison to placebo in an ARMS population defined by CAARMS criteria.We found no evidence that treatment with minocycline or omega-3, either alone or in combination, reduced transition to psychosis and nearly all participants at 12 months continued to fulfil criteria for ARMS.However, the power of the study to detect protective treatment effects is substantially less than planned because the overall transition rate of 14 % was half the 30 % anticipated.This results in broad confidence intervals for the estimated RRs and the possibility of false negative inferences.However, exposure to omega-3, against prediction, was associated with increased onsets in both the omega-3 alone and combination groups (Table 2) giving an overall RR (1.67) with a true RR within a 95 % CI = [0.95,2.9] from close to equal risk (RR = 1) and a 3-fold increased RR.Exploring this effect at 3 months, when all but 7 of the transitions had already occurred, revealed a doubling of transitions in those exposed to omega-3 (RR = 1.95; 95 % CI [1.03, 3.67]; p = 0.04).These CIs do not remotely encompass the predefined minimal clinically meaningful reduction of 50 % in onsets (RR = 0.50); we infer that the lack of a protective effect of omega-3 is unlikely to be a false negative.
The estimated effect of minocycline is marginally less than unity (RR = 0.85), but the CI extends to RR = 0.50 so we cannot conclude that minocycline is without a clinically relevant effect in a future trial where new methods of selection identify a risk profile with a much greater rate of onsets.A lack of a preventative effect of minocycline would be consistent with a large trial of minocycline in recent-onset schizophrenia that found no antipsychotic benefit compared to placebo (Deakin et al., 2018).Epidemiological studies have not found a protective effect of minocycline exposure in adolescence, typically for acne, on later risk of psychosis [26).In-vivo imaging of the microglial marker protein TSPO using PET increasingly suggests that microglial inflammation occurs in depression rather than in schizophrenia and recent studies suggest minocycline has antidepressant efficacy (Upmark et al., 2021;Nettis et al., 2021).In schizophrenia, reductions in TSPO expression are consistently reported which are most marked in studies with recent onset, drug-naïve patients (Plavén-Sigray et al., 2021) and in the absence of an inflamed microglial phenotype in the ARMS, a beneficial effect of the anti-microglial properties of minocycline seems unlikely.
Omega-3 was not effective in reducing rates of transition to psychosis in the present study and this is in keeping with the large NEURAPRO study (Berger et al., 2019).However, in both the NEURAPRO and NAYAB studies, the lower than expected transition rates in placebotreated participants (14 % in NAYAB study and 11 % in NEURAPRO) reduced power to detect a preventative effect.None of the three previous prevention studies showed a tendency to exacerbation of transition with omega-3.However, our sample is drawn from a very different LMIC population exposed to high rates of malnutrition in childhood and poor diet with no early intervention services.It seems possible that lifetime exposure to low dietary omega-3 might distort lipid metabolism and brain development to create a susceptibility to normal or abnormal 2. to omega-3 improved CAARMS global positive symptom scores in non-transitioning Key: Blue lines and circles show CAARMS scores in non-transitioning ARMS in the 2 groups exposed to omega-3 indicated by blue lines and circles, compared to the 2 groups who did not receive omega-3 (red lines and triangles).Closed symbols with dashed lines indicate single drug exposureblue for omega-3 and red for minocycline.6-and 12-month post-treatment scores include baseline scores of 29 dropouts carried forward.CAARMS standard errors all less than 0.5.Main effect of omega-3 on 6 &12 months CAARMS, p < 0.01.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)downstream metabolic effects of omega-3 supplementation.Some possible mechanisms are discussed in one report in which EPA supplementation increased drop-outs and psychotic symptoms during treatment for a psychotic episode (Bentsen et al., 2013) especially in those with low with low red-cell omega-3 PUFA concentrations.
Despite the lack of benefit on transitions, exposure to omega-3 in the non-transitioning ARMS sample was associated with a statistically significant improvement in CAARMS and MADRS ratings.Two studies in early active psychosis reported omega-3 improved non-psychotic symptoms including depression (Robinson et al., 2019;Pawełczyk et al., 2015) and general psychopathology and function (Robinson et al., 2019).There is increasing evidence that low PUFA levels occur as much in major depression as in schizophrenia (Hallahan et al., 2016;Liao et al., 2019).A meta-analysis of double-blind RCTs (n = 2160) found therapeutic benefits of omega-3 in major depressive disorder (Liao et al., 2019) and on depressive and general symptoms rather than on positive and negative symptoms in first episode psychosis (Bentsen et al., 2013).Whether anti-inflammatory effects of omega-3 mediate the reported benefits is not known.However, ARMS was associated with increased circulating concentrations of IL-6 in a meta-analysis of 7 studies but increases did not predict psychosis onset (Park and Miller, 2020).A recent prospective study similar in scale and design to our own, reported that baseline was only in the majority non-transitioning group (n = 219) compared with 56 controls whereas none of 19 cytokines were changed in the group that transitioned to psychosis (n = 50) versus controls (Mondelli et al., 2023).These findings together with the benefit of omega-3 on CAARMS and MADRS ratings in our nontransitioning group, point to a possible inflammatory component to the affective and general psychopathology aspects in ARMS.

4.1 Strengths
The current study is the first to trial minocycline with omega-3 PUFAs at scale in an ARMS population.A strength of this study was the large sample size in comparison to previous pharmacological intervention studies in this area that have been in the range n = 44-304 (Davies et al., 2018) and the use of a validated assessment measure in terms of ascertainment of ARMS and transition to psychosis.Recruitment to the study was high in those assessed as CAARMS positive with almost all (98 %) consenting to randomisation indicating trial acceptability and a willingness to take part.Attrition rates were less than 10 % in all study arms making for minimal possible confounding by undetected missing at random assumptions and demonstrating that the study interventions were highly acceptable.

Limitations
Longer duration of prodromal symptoms is a risk factor for transition to psychosis but was not measured.Only the positive sub-scale of CAARMS was measured which despite being of particular interest, does not capture the full range of clinical symptoms in an ARMS population.However, we supplemented the positive sub-scale of CAARMS with the MADRS.The study has no measures of inflammatory markers or of PUFAs due to lack of resource and technical infrastructure.However, there is no known or consensus pattern of cytokines that predicts risk of transition and, as noted, recent evidence points away from inflammatory cytokines as a risk factor.Without plasma or erythrocyte PUFA levels we cannot be certain that the dose of omega-3 engaged this target but increased blood PUFA markers were reported in a study using similar doses (Alqarni et al., 2020).Recent evidence suggests that considerably greater doses of omega-3 may be necessary to engage the full range of anti-inflammatory pathways and 2-4 g doses have shown preventative benefit in cardiovascular disease albeit with an increased risk of atrial fibrillation (Bhatt et al., 2019).Amminger et al. (Amminger et al., 2020) conclude from their analysis of omega-3 levels in the NEURAPRO study, that high doses may be necessary to reduce transition while lower doses are clearly sufficient to benefit broad psychopathology as in our study.However, further therapeutic studies are not indicated until greater transition rates can be predicted and in the present LMIC population, high dose omega-3 could risk exacerbating the increased risk of transition observed at the dose we used.The study could have benefitted from an assessment of dietary PUFA intake which is known to be low in Pakistan, but PUFA dietary intake validated against plasma measures are lacking.The main weakness of our study as with others, is the low transition rate and lack of statistical power to detect treatment effects.

Conclusion
We found no evidence that either minocycline or omega-3 PUFAs, either alone or in combination, reduce a putative inflammationmediated transition to psychosis in an ARMS population.However, the study was underpowered due to the low transition rate.There was evidence that omega-3 supplementation had a benefit on positive psychotic symptoms and mood ratings in those who did not become psychotic.We discuss a possible implication of this finding and recent cytokine studies that the ARMS may have an inflammatory component to pathogenesis which is discontinuous with psychosis risk and potentially amenable to anti-inflammatory treatment.Further omega-3 studies and minocycline studies are not indicated until it has been demonstrated that a substantial subgroup exists with reliably reduced erythrocyte omega-3 PUFA or with immune activation that have a high rate of transition.

Data sharing
Individual participant data collected during the trial is available after de-identification immediately after publication and will end five years after article publication.Data requestors will need to complete a data access agreement and requests sent to inti.qurashi@nhs.net.

Declaration of Competing Interest
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: IBC reports giving lectures or advice to Eli Lilly, Bristol Myers Squibb, Lundbeck, AstraZeneca, and Janssen pharmaceuticals for which he or his employing institution have been reimbursed, outside the submitted work; IBC was previously trustee of the Pakistan Institute of Living and Learning (PILL).NH has been a past Trustee of the PILL, Abaseen Foundation UK, Lancashire Mind UK and Manchester Global Foundation (MGF).He is an executive member of the Academic Faculty at the Royal College of Psychiatrists, London.He is a NIHR Senior Investigator.He is director of research and innovation at Mersey Care NHS Foundation Trust.NH/IBC/AB & TK has attended educational events organized by various pharmaceutical industries.IQ is associate director of research and innovation at Mersey Care NHS Foundation Trust.

Fig. 1 .
Fig. 1.Trial profile Note: Each time-point indicates when CAARMS assessments were carried out.'Psychosis' indicates those meeting the CAARMS threshold criteria.

Table 1
Baseline demographics and outcome variables for 4 intervention groups.

Table 2
Conversion to psychosis by intervention.

Table 3
Conversion to psychosis by total minocycline or omega-3 exposures; factorial analysis.