Skip to content

Gut–brain neuromodulators

TL;DR — This is the best-evidenced drug class in IBS that works across subtypes, and its flagship result comes from primary care rather than a referral clinic. ATLANTIS (ISRCTN48075063) randomised 463 adults with Rome IV IBS at 55 English general practices to self-titrated amitriptyline 10–30 mg once daily or placebo for 6 months: IBS-SSS difference −27.0 (95% CI −46.9 to −7.10, p=0.0079) and subjective global assessment of relief OR 1.78 (1.19–2.66, p=0.005) (Ford 2023, PMID 37858323; Wright-Hughes 2024, PMID 39397570). Note that −27.0 falls short of the trial's own prespecified 35-point minimum clinically important difference, while the responder outcome was clearly positive — a tension the trial reports openly. The updated meta-analysis (28 RCTs, 2,475 patients) gives a relative risk of global symptoms not improving of 0.77 (0.69–0.87) for all neuromodulators, 0.70 (0.62–0.80) for tricyclics specifically (11 trials, 1,144 patients) — the only endpoint rated moderate certainty; abdominal pain RR 0.72 (0.62–0.83) overall, 0.69 (0.54–0.87) for tricyclics, 0.74 (0.56–0.99) for SSRIs and 0.22 (0.08–0.59) for SNRIs on just two trials and 94 patients (Khasawneh 2025, PMID 40258375). Tricyclics also ranked first for abdominal pain in the traditional-therapies network meta-analysis, on four RCTs and 92 patients (Black 2020, PMID 31859183). The class costs tolerability: adverse-event withdrawal was significantly higher than placebo (Khasawneh 2025, PMID 40258375), 12.9% vs 8.7% in ATLANTIS (Wright-Hughes 2024, PMID 39397570), and the NNH for discontinuation is 24 (Busam 2026, PMID 40471839). Amitriptyline did not improve anxiety, depression, somatoform symptom reporting or work and social adjustment in ATLANTIS — its effect on the gut is not mediated by its effect on mood (Wright-Hughes 2024, PMID 39397570).

ATLANTIS in detail

The largest tricyclic trial ever conducted in IBS, and the only one designed for the setting where most IBS is managed.

Feature Detail
Design Pragmatic, randomised, multicentre, two-arm, double-blind, placebo-controlled; nested qualitative study
Setting 55 general practices across Wessex, West of England and West Yorkshire
Recruitment 18 Oct 2019 – 11 Apr 2022
Participants 463 adults, Rome IV IBS of any subtype, IBS-SSS ≥75 despite dietary change and first-line therapy, normal FBC and CRP, negative coeliac serology, no suicidal ideation. Mean age 48.5 (SD 16.1); 315 (68%) female, 148 (32%) male
Intervention Amitriptyline 10 mg od, patient-led self-titration over 3 weeks to a maximum of 30 mg od, or matched placebo, for 6 months (most had the option of a further blinded 6 months)
Primary outcome IBS-SSS at 6 months; prespecified MCID 35 points
Result −27.0 (95% CI −46.9 to −7.10), p=0.0079 (p=0.008 in the HTA report)
Key secondary Subjective global assessment of relief at 6 months: OR 1.78 (1.19–2.66), p=0.005
Null secondaries No effect on somatoform symptom reporting, anxiety, depression, or work and social adjustment
Discontinuation 46 (20%) stopped amitriptyline (30, 13%, for adverse events); 59 (26%) stopped placebo (20, 9%, for adverse events). Adverse-event withdrawals 12.9% vs 8.7%
Serious events Five serious adverse reactions (2 amitriptyline, 3 placebo); five serious adverse events unrelated to trial medication
Qualitative 77 semi-structured interviews with 42 participants and 16 GPs; most found self-titration "acceptable and empowering"

Sources: Ford 2023, PMID 37858323; Wright-Hughes 2024, PMID 39397570.

Two implementation findings from the HTA report matter as much as the effect size. First, GPs rarely prescribe amitriptyline for IBS despite existing NICE guidance, and the trial's conclusion is a direct instruction to change that. Second, the authors call for "guidance and resources to support GP–patient communication to distinguish amitriptyline for irritable bowel syndrome from use as an antidepressant" — the drug's identity as an antidepressant is itself a barrier, which connects to the labelling effect demonstrated elsewhere in this condition (antispasmodics-and-peppermint; Rexwinkel 2025, PMID 40074185).

Who responds

Post-hoc analyses of ATLANTIS (Wright-Hughes 2025, PMID 39863398):

Subgroup Effect Interaction p
Age ≥50 vs <50 IBS-SSS mean difference −46.5 (−74.2 to −18.8), p=0.0010; SGA relief OR 2.59 (1.47–4.55), p=0.0010 0.048 (IBS-SSS); 0.068 (SGA)
70% most deprived areas of England vs 30% least deprived ≥30% abdominal-pain improvement OR 2.70 (1.52–4.77), p=0.0007 0.021
Men Stronger treatment effect —
Higher PHQ-12 (somatic symptom) scores Stronger treatment effect —
IBS-D Stronger treatment effect —

The authors' conclusion is deliberately cautious: these exploratory analyses show "there are unlikely to be deleterious effects of amitriptyline in any particular IBS subtype and could help identify patients in whom amitriptyline may be more effective." Mean differences in individual IBS-SSS components favoured amitriptyline and side effects were similar across almost all subtypes. The socioeconomic-deprivation interaction is unusual in a drug trial and has not been replicated.

The class evidence

Class Global symptoms RR (not improving) Abdominal pain RR Trials / patients Certainty
All gut–brain neuromodulators 0.77 (0.69–0.87) 0.72 (0.62–0.83) 22 / 2,222 (global); 19 / 1,792 (pain) —
Tricyclic antidepressants 0.70 (0.62–0.80) 0.69 (0.54–0.87) 11 / 1,144 (global); 7 / 708 (pain) Moderate for global symptoms
SSRIs — 0.74 (0.56–0.99) 7 / 324 Low to very low
SNRIs — 0.22 (0.08–0.59) 2 / 94 Low to very low
Azapirones, tetracyclics insufficient data insufficient data — "More data required"

Source: Khasawneh 2025, PMID 40258375 (search to 1 January 2025; 28 eligible RCTs, 2,475 patients, of which 10 trials and 1,348 patients were new since the previous synthesis). Adverse events were not significantly more common overall, but withdrawal due to adverse events was significantly higher.

The predecessor synthesis found antidepressants versus placebo RR 0.66 (0.57–0.76), with similar effects for tricyclics and SSRIs, though with heterogeneity among SSRI trials (I²=49%, p=0.07) (Ford 2019, PMID 30177784). An older tricyclic-only meta-analysis of seven placebo-controlled trials (amitriptyline, imipramine, desipramine, doxepin, trimipramine) gave a pooled RR for clinical improvement of 1.93 (1.44–2.60, p<0.0001) (Rahimi 2009, PMID 19340896).

The SNRI figure deserves a warning. RR 0.22 (0.08–0.59) for abdominal pain is the largest effect estimate for any drug in this knowledge base, and it rests on two trials and 94 patients. It should be read as a signal requiring confirmation, not as evidence that SNRIs are four times better than tricyclics.

Agent by agent

The class label hides substantial heterogeneity of evidence quality. What exists, drug by drug:

Agent Best evidence Result
Amitriptyline 10–30 mg ATLANTIS, 463 patients, primary care, 6 months (Ford 2023, PMID 37858323) IBS-SSS −27.0 (−46.9 to −7.10); SGA relief OR 1.78 (1.19–2.66)
Desipramine Drossman 2003, PMID 12851867 — 431 women with moderate-to-severe functional bowel disorders, 12 weeks, four-arm comparator-controlled multicentre trial (CBT vs education; desipramine vs placebo) Intention-to-treat: not significant (60% vs 47% responders, p=0.16, NNT 8.1). Per-protocol: significant (73% vs 49%, p=0.01, NNT 5.2), strengthening further when participants with undetectable desipramine blood levels were excluded (p=0.002). Benefit concentrated in moderate rather than severe symptoms, in those with an abuse history, in those without depression, and in diarrhoea-predominant symptoms
TCAs, pooled Rahimi 2009, PMID 19340896 — 7 placebo-controlled trials of amitriptyline, imipramine, desipramine, doxepin, trimipramine RR for clinical improvement 1.93 (1.44–2.60), p<0.0001
SSRIs Khasawneh 2025, PMID 40258375 (7 RCTs, 324 patients); Bundeff 2014, PMID 24651166 (review) Abdominal pain RR 0.74 (0.56–0.99); low-to-very-low certainty. AGA recommends against — see below
SNRIs (duloxetine, venlafaxine) Khasawneh 2025, PMID 40258375 — 2 trials, 94 patients Abdominal pain RR 0.22 (0.08–0.59) — the largest effect estimate for any drug in this knowledge base, on the smallest evidence base. Duloxetine's IBS literature is otherwise open-label pilots (Brennan 2009, PMID 19548294; Kaplan 2014, PMID 23980534; Lewis-Fernández 2016, PMID 27755218)
Mirtazapine Khalilian 2021, PMID 33536043 — 67 Rome IV IBS-D patients, 15 mg then 30 mg for 8 weeks, double-blind placebo-controlled Greater reduction in IBS-SSS (p=0.002); all diary symptoms except bloating improved; quality of life (p=0.04) and anxiety (p=0.005) improved. A 2026 systematic review found only five eligible publications in total — three case reports, this RCT, and one prospective study — and performed no meta-analysis because of heterogeneity (Prodan 2026, PMID 41809155)
α2δ ligands (pregabalin) Houghton 2007, PMID 17446306 — 26 Rome II patients with rectal hypersensitivity (pain threshold ≤28 mmHg), 3 weeks, randomised placebo-controlled Raised sensory thresholds for first sensation (p=0.045), desire to defecate (p=0.008) and pain (p=0.048) to normal levels, and raised rectal compliance (p<0.0001), apparently independently. Mild dizziness and somnolence. A mechanistic proof-of-concept, not an efficacy trial
α2δ ligand PD-217,014 Houghton 2025, PMID 39812493 — 330 Rome II-defined participants, 150 or 300 mg twice daily versus placebo for 4 weeks Neither dose improved the primary adequate-relief endpoint or secondary symptom endpoints; 321 participants also met Rome IV criteria, and analyses by Rome II/IV bowel-habit subtype were likewise null. This is class-level evidence, not a pregabalin trial
Azapirones, tetracyclics — "More data required" (Khasawneh 2025, PMID 40258375)

Two observations follow. First, the desipramine result is the field's cleanest demonstration that adherence determines apparent efficacy: the same trial is negative by intention-to-treat and positive per-protocol, with the gap explained by measured blood levels (Drossman 2003, PMID 12851867). Second, the agents with the largest point estimates — SNRIs at RR 0.22 — have the smallest evidence bases, which is the classic signature of small-study effects.

What the trials say about mechanism

  • Not mood. ATLANTIS improved bowel symptoms with no change in anxiety, depression, somatoform symptom reporting or work and social adjustment (Wright-Hughes 2024, PMID 39397570). Desipramine's benefit was concentrated in patients without depression (Drossman 2003, PMID 12851867). Antidepressant effects in IBS patients with comorbid depression have been examined separately (Friedrich 2010, PMID 20678672).
  • Plausibly afferent. Pregabalin normalised rectal sensory thresholds in hypersensitive patients (Houghton 2007, PMID 17446306), and mirtazapine's rationale in IBS-D is explicitly its 5-HT3 antagonism rather than its antidepressant action (Khalilian 2021, PMID 33536043) — the same receptor targeted by alosetron and ramosetron (diarrhoea-predominant-pharmacotherapy).
  • Partly anticholinergic. Tricyclics slow transit, which is useful in IBS-D and unhelpful in IBS-C; ATLANTIS post-hoc analyses found stronger effects in IBS-D and no deleterious effect in any subtype (Wright-Hughes 2025, PMID 39863398).
  • Partly cognitive. In the Drossman programme, cognitive factors predicted response to both medical and psychological treatment (Weinland 2010, PMID 20087332).

Why "neuromodulator", not "antidepressant"

Three lines of evidence in this build support the reframing:

  1. Dose. ATLANTIS used 10–30 mg amitriptyline; antidepressant dosing is several-fold higher (Ford 2023, PMID 37858323).
  2. Mood-independence. Amitriptyline improved IBS-SSS and global relief while leaving anxiety, depression, somatoform symptom reporting and work/social adjustment unchanged (Wright-Hughes 2024, PMID 39397570).
  3. Peripheral targets exist. Tricyclics slow transit (useful in IBS-D) and have anticholinergic and sodium-channel actions independent of monoamine reuptake; the class ranked first for abdominal pain in a network of traditional therapies alongside peppermint oil (Black 2020, PMID 31859183).

Against that, the genetic correlation between IBS and anxiety/depression exceeds rg 0.5 (Eijsbouts 2021, PMID 34741163), so a shared-substrate explanation cannot be dismissed — see brain-gut-axis-and-visceral-hypersensitivity.

Khasawneh 2025 found SSRIs beat placebo for abdominal pain (RR 0.74, 0.56–0.99; 7 trials, 324 patients) and explicitly "highlight a potential for SSRIs to be modestly effective for abdominal pain" (PMID 40258375). AGA nonetheless makes a conditional recommendation against SSRIs in both IBS-C and IBS-D, at low certainty (Chang 2022, PMID 35738724; Lembo 2022, PMID 35738725). Both positions are defensible on the same data — the confidence interval touches 1.00, the trials are small and heterogeneous, and AGA weighted the evidence conservatively — but the disagreement is real and unresolved. It is catalogued on guidelines.

Safety and tolerability

Metric Value Source
NNH for discontinuation due to adverse event, tricyclics 24 (p<0.01) Busam 2026, PMID 40471839
Adverse-event withdrawal, ATLANTIS 12.9% vs 8.7% placebo Wright-Hughes 2024, PMID 39397570
Adverse events vs placebo, network meta-analysis RR 1.59 (1.26–2.06) Black 2020, PMID 31859183
Class-level adverse events Not significantly more common; withdrawals significantly higher Khasawneh 2025, PMID 40258375

Tricyclics at higher doses are named among the three highest-absolute-risk IBS drugs for adverse-event discontinuation, alongside tenapanor and alosetron (Busam 2026, PMID 40471839). ATLANTIS excluded participants with suicidal ideation at entry (Ford 2023, PMID 37858323) — a screening step that should carry over into practice and is noted on red-flags-and-safety-concerns.

A negative trial worth recording

Ethosuximide, a T-type (Cav3.2) calcium channel blocker with experimental support for a role in visceral hypersensitivity, was tested in a multicentre French proof-of-concept RCT (NCT02973542; 124 randomised, 12 weeks). Responder rates were 26.6% vs 23.3% (RR 1.14, 0.61–2.11) — no benefit — with markedly worse tolerability (discontinuation 46.9% vs 21.7%, p=0.003; 56.4% of all 463 adverse events attributed to ethosuximide, p<0.001; commonest headaches, sleep disturbance, nausea) (Kerckhove 2026, PMID 41505133). It is included here because a mechanistically motivated neuromodulator failing cleanly is informative about how little the visceral-hypersensitivity mechanism currently predicts drug success.

Open questions

  • Is a 27-point IBS-SSS difference clinically meaningful when the trial prespecified 35 points as the minimum important difference (Ford 2023, PMID 37858323)? The responder analysis says yes; the continuous primary outcome says not quite.
  • Why does amitriptyline work better in older, poorer, male and IBS-D patients (Wright-Hughes 2025, PMID 39863398)? These are exploratory subgroups awaiting replication.
  • Should SSRIs be used for abdominal pain? Meta-analysis says modestly effective (Khasawneh 2025, PMID 40258375); AGA recommends against (Lembo 2022, PMID 35738725).
  • Are SNRIs really the most effective analgesics in IBS (RR 0.22, 0.08–0.59 on 94 patients; Khasawneh 2025, PMID 40258375)? A targeted PubMed and registry search on 2026-09-02 retrieved no adequately powered SNRI trial.
  • Neuromodulator or psychological therapy first in primary care? No head-to-head pragmatic comparison exists — see psychological-therapy and OPEN-QUESTIONS.md.
  • Does amitriptyline's benefit persist beyond 12 months, and what happens on withdrawal? ATLANTIS followed to 12 months with blinded extension (Wright-Hughes 2024, PMID 39397570); nothing longer exists.
  • Is desipramine effective? The same trial is null by intention-to-treat and positive per-protocol once blood levels are accounted for (Drossman 2003, PMID 12851867), and no adequately powered adherence-controlled replication exists.
  • Is mirtazapine's 5-HT3 antagonism the active mechanism in IBS-D, and does it match ramosetron's efficacy (Khalilian 2021, PMID 33536043; Prodan 2026, PMID 41809155)?
  • Do α2δ ligands translate from normalising rectal sensory thresholds (Houghton 2007, PMID 17446306) into symptom benefit? A 330-participant efficacy trial of the related ligand PD-217,014 found no significant benefit at either dose (Houghton 2025, PMID 39812493); pregabalin itself still lacks a symptom-efficacy trial retrieved by the targeted PubMed search on 2026-09-02.
  • Does calling the drug an antidepressant reduce its effect, as the labelling literature would predict (Rexwinkel 2025, PMID 40074185)? The ATLANTIS qualitative substudy raises this but does not test it.

References

  1. Ford AC, et al. Amitriptyline at Low-Dose and Titrated for Irritable Bowel Syndrome as Second-Line Treatment in primary care (ATLANTIS): a randomised, double-blind, placebo-controlled, phase 3 trial. Lancet. 2023;402(10414):1773-1785. PMID 37858323
  2. Wright-Hughes A, et al. Low-dose titrated amitriptyline as second-line treatment for adults with irritable bowel syndrome in primary care: the ATLANTIS RCT. Health Technol Assess. 2024;28(66):1-161. PMID 39397570
  3. Wright-Hughes A, et al. Predictors of response to low-dose amitriptyline for irritable bowel syndrome and efficacy and tolerability according to subtype: post hoc analyses from the ATLANTIS trial. Gut. 2025;74(5):728-739. PMID 39863398
  4. Khasawneh M, Mokhtare M, Moayyedi P, Black CJ, Ford AC. Efficacy of gut-brain neuromodulators in irritable bowel syndrome: an updated systematic review and meta-analysis. Lancet Gastroenterol Hepatol. 2025;10(6):537-549. PMID 40258375
  5. Ford AC, Lacy BE, Harris LA, Quigley EMM, Moayyedi P. Effect of Antidepressants and Psychological Therapies in Irritable Bowel Syndrome: An Updated Systematic Review and Meta-Analysis. Am J Gastroenterol. 2019;114(1):21-39. PMID 30177784
  6. Rahimi R, Nikfar S, Rezaie A, Abdollahi M. Efficacy of tricyclic antidepressants in irritable bowel syndrome: a meta-analysis. World J Gastroenterol. 2009;15(13):1548-53. PMID 19340896
  7. Drossman DA, et al. Cognitive-behavioral therapy versus education and desipramine versus placebo for moderate to severe functional bowel disorders. Gastroenterology. 2003;125(1):19-31. PMID 12851867
  8. Weinland SR, et al. Cognitive factors affect treatment response to medical and psychological treatments in functional bowel disorders. Am J Gastroenterol. 2010;105(6):1397-406. PMID 20087332
  9. Houghton LA, Fell C, Whorwell PJ, Jones I, Sudworth DP, Gale JD. Effect of a second-generation alpha2delta ligand (pregabalin) on visceral sensation in hypersensitive patients with irritable bowel syndrome. Gut. 2007;56(9):1218-25. PMID 17446306
  10. Houghton LA, Gao S, Gilbert SA, Coffin B, Simrén M, Gale JD, A4451007 Study Investigators. Clinical Trial: Study to Investigate the Efficacy and Safety of the Alpha-2-Delta Ligand PD-217,014 in Patients With Irritable Bowel Syndrome. Aliment Pharmacol Ther. 2025;61(5):803-813. PMID 39812493
  11. Khalilian A, Ahmadimoghaddam D, Saki S, Mohammadi Y, Mehrpooya M. A randomized, double-blind, placebo-controlled study to assess efficacy of mirtazapine for the treatment of diarrhea predominant irritable bowel syndrome. Biopsychosoc Med. 2021;15(1):3. PMID 33536043
  12. Prodan R, Soldera J. Mirtazapine for the treatment of irritable bowel syndrome: A systematic review. World J Methodol. 2026;16(1):107908. PMID 41809155
  13. Bundeff AW, Woodis CB. Selective serotonin reuptake inhibitors for the treatment of irritable bowel syndrome. Ann Pharmacother. 2014;48(6):777-84. PMID 24651166
  14. Brennan BP, Fogarty KV, Roberts JL, Reynolds KA, Pope HG Jr, Hudson JI. Duloxetine in the treatment of irritable bowel syndrome: an open-label pilot study. Hum Psychopharmacol. 2009;24(5):423-8. PMID 19548294
  15. Kaplan A, Franzen MD, Nickell PV, Ransom D, Lebovitz PJ. An open-label trial of duloxetine in patients with irritable bowel syndrome and comorbid generalized anxiety disorder. Int J Psychiatry Clin Pract. 2014;18(1):11-5. PMID 23980534
  16. Lewis-Fernández R, et al. An Open-Label Pilot Study of Duloxetine in Patients With Irritable Bowel Syndrome and Comorbid Major Depressive Disorder. J Clin Psychopharmacol. 2016;36(6):710-715. PMID 27755218
  17. Friedrich M, Grady SE, Wall GC. Effects of antidepressants in patients with irritable bowel syndrome and comorbid depression. Clin Ther. 2010;32(7):1221-33. PMID 20678672
  18. Black CJ, Yuan Y, Selinger CP, Camilleri M, Quigley EMM, Moayyedi P, Ford AC. Efficacy of soluble fibre, antispasmodic drugs, and gut-brain neuromodulators in irritable bowel syndrome: a systematic review and network meta-analysis. Lancet Gastroenterol Hepatol. 2020;5(2):117-131. PMID 31859183
  19. Kerckhove N, et al. Ethosuximide and Irritable Bowel Syndrome-Related Abdominal Pain: A Randomized Clinical Trial. JAMA Netw Open. 2026;9(1):e2551368. PMID 41505133
  20. Busam JA, et al. The Safety of Pharmacotherapy for Irritable Bowel Syndrome: A Systematic Review and Meta-Analysis. Am J Gastroenterol. 2026;121(3):745-753. PMID 40471839
  21. Chang L, Sultan S, Lembo A, Verne GN, Smalley W, Heidelbaugh JJ. AGA Clinical Practice Guideline on the Pharmacological Management of Irritable Bowel Syndrome With Constipation. Gastroenterology. 2022;163(1):118-136. PMID 35738724
  22. Lembo A, Sultan S, Chang L, Heidelbaugh JJ, Smalley W, Verne GN. AGA Clinical Practice Guideline on the Pharmacological Management of Irritable Bowel Syndrome With Diarrhea. Gastroenterology. 2022;163(1):137-151. PMID 35738725
  23. Rexwinkel R, et al. Mebeverine and the Influence of Labeling in Adolescents With Irritable Bowel Syndrome or Functional Abdominal Pain Not Otherwise Specified: A 2 x 2 Randomized, Placebo-Controlled Trial. Gastroenterology. 2025;169(1):94-103. PMID 40074185
  24. Eijsbouts C, et al. Genome-wide analysis of 53,400 people with irritable bowel syndrome highlights shared genetic pathways with mood and anxiety disorders. Nat Genet. 2021;53(11):1543-1552. PMID 34741163