Inflammatory bowel disease is chronic immune-mediated inflammation of the gut. It comes in two main forms, and the distinction matters because it changes almost everything downstream.
Crohn's disease can affect anywhere from mouth to anus, does so in patches with normal bowel in between, and inflames the full thickness of the wall — which is why it produces strictures, fistulas and abscesses [1].
Ulcerative colitis affects only the colon, in a continuous stretch extending back from the rectum, and involves only the inner lining. That anatomical confinement is why removing the colon removes the disease — the one situation in this review where the word "cure" is accurate [2].
Both run a relapsing-remitting course over decades, both typically begin in adolescence or young adulthood, and both have been rising in incidence worldwide, with the largest increases in newly industrialising regions [3]. Neither is usually curable. Both are now, in a way that was not true twenty-five years ago, controllable.
The most important change in how they are managed is not any single drug. It is a shift in what counts as success: from "the patient feels better" to "the inflammation is gone from the mucosa." That shift is the subject of the centerpiece.
Start here: what is going wrong
The current understanding is that IBD arises when a genetically susceptible immune system responds inappropriately to the intestinal microbiota across a barrier that is not doing its job. None of those three components alone is sufficient.
The genetic contribution is real, polygenic and shared: association analyses identified 38 loci across both diseases [4] and genome-wide meta-analysis raised the confirmed Crohn's disease loci to 71 [5]. Many implicate innate immune sensing of bacteria and autophagy, which is a strong hint about mechanism rather than a diagnostic test — the heritability is substantial and the predictive value of any individual variant is small.
The immunology has been productive precisely because it generated drugs. The interleukin-23 pathway in particular was shown to drive a pathogenic T-cell population causing autoimmune inflammation [6], to be the critical cytokine rather than interleukin-12 in that setting [7], and to be essential for T-cell-mediated colitis specifically [8]. That chain of laboratory work is why an anti-IL-12/23 antibody is now a standard IBD therapy.
The disease is not confined to the gut. Extraintestinal manifestations affect joints, eyes, skin and the biliary tree; fatigue and psychological burden are substantial and measurable [9]; and depressed mood has been shown prospectively to predict relapse [10].
Pillar 1: measurement and diagnosis
Endoscopy makes the diagnosis, and keeps making it
Diagnosis rests on ileocolonoscopy with biopsies, interpreted alongside the clinical picture, laboratory tests and imaging [11]. What the endoscopist sees distinguishes the two diseases: continuous inflammation extending proximally from the rectum with a clear upper margin suggests ulcerative colitis; patchy inflammation with intervening normal mucosa, deep ulceration, ileal involvement or perianal disease suggests Crohn's. Histology adds granulomas when present.
For Crohn's disease, endoscopy cannot see what lies beyond the lumen, and cross-sectional imaging — magnetic resonance enterography in particular — is what identifies strictures, fistulas and abscesses [11] [12].
Monitoring without a camera
Repeating colonoscopy often enough to guide treatment is invasive, expensive and unpopular. The practical solution is non-invasive markers, and their performance has been quantified.
A meta-analysis of nineteen studies in 2,499 symptomatic patients, using endoscopy as the gold standard, found fecal calprotectin at sensitivity 0.88 and specificity 0.73, against C-reactive protein at sensitivity 0.49 and specificity 0.92 [13]. That contrast is worth reading carefully, because it tells you what each test is for. CRP is specific and insensitive: a raised CRP means real inflammation, but a normal CRP does not mean a quiet gut — about half of endoscopically active patients have a normal one. Calprotectin is sensitive and less specific: it is a neutrophil protein shed into stool, it catches most active disease, and it will occasionally be raised for other reasons.
Calprotectin's correlation with endoscopic scores has been established repeatedly [14] [15] [16], it performs well for distinguishing inflammatory from functional bowel symptoms [17] [18], and surrogate markers of intestinal inflammation were shown to predict relapse as far back as 2000 [19].
The change in what is measured
Historically, IBD was managed to symptoms. The problem is that symptoms and inflammation come apart in both directions: patients feel well with an ulcerated colon, and patients feel dreadful with a healed one.
The STRIDE initiative formalised the alternative, selecting objective therapeutic targets to aim at rather than merely monitoring [20], and STRIDE-II updated them, separating immediate targets from intermediate and long-term ones [21]. Endoscopic healing is the central long-term target, with symptom control and normalisation of calprotectin and CRP as earlier waypoints.
Why that is the right target rather than an academic refinement is the centerpiece.
Centerpiece: a simple simulatable model of treating to target
The argument for treat-to-target is usually made qualitatively. It can be made arithmetically, from three proportions reported in a single cohort.
A study of 214 Crohn's disease patients on maintenance infliximab, of whom 183 responded to induction, reported: that scheduled dosing achieved mucosal healing in 76.9 percent of patients against 61.0 percent for episodic dosing (odds ratio 2.14); that concomitant corticosteroids reduced healing (37.9 versus 63.2 percent, odds ratio 0.36); and that during long-term follow-up major abdominal surgery occurred in 14.1 percent of patients who healed against 38.4 percent of those who did not [22].
Those numbers chain. A strategy determines how often patients heal; healing status determines how often they come to surgery; and the law of total probability combines them into the surgery risk a strategy implies:
P(surgery | strategy) = p_heal × P(surgery | healed) + (1 − p_heal) × P(surgery | not healed)
There are no free parameters. Every quantity is measured.
The model is checked twice, against numbers it was not given. The source reports odds ratios alongside its proportions, so recomputing those odds ratios from the proportions alone tests whether the arithmetic is reading the study correctly: scheduled versus episodic dosing gives 2.128 against a published 2.14, and corticosteroids give 0.355 against a published 0.36. Both reproduce.
Running the chain forward: scheduled dosing implies a surgery risk of 19.7 percent against 23.6 percent for episodic — a difference of 3.9 percentage points, or a number needed to treat of about 26 to prevent one major abdominal operation by changing dosing strategy alone.
The right-hand panel generalises it. Because the combination is linear in the healing rate, surgery risk falls along a straight line from 38.4 percent when nobody heals to 14.1 percent when everybody does. The slope is 24.3 points of surgery per 100 points of healing — so every additional ten percent of patients reaching mucosal healing buys about 2.4 fewer operations per hundred patients.
That is the teaching point, and it is why the target is the target. You cannot act on "surgery in five years." You can act on whether this patient's mucosa is healed at the next assessment. The intermediate target is the only thing a clinic can steer, and the slope is the exchange rate between steering it and the outcome anyone actually cares about.
The supporting evidence is broader than one cohort: mucosal healing predicted long-term outcome in a population-based cohort [23], and early mucosal healing on infliximab was associated with better long-term clinical outcomes [24].
Two honest limits, and the first is serious. The healing-to-surgery association comes from an observational cohort, and the model treats it as though intervening to produce healing transfers that risk. Patients who heal on treatment very likely differ in prognosis from those who do not, so some of the 24-point gap is selection rather than benefit. The direction is well supported; the magnitude should be read as an upper bound on what targeting healing can buy. Second, the cohort was Crohn's patients on infliximab in a particular era, and the healing rates achievable with current agents differ.
Pillar 2: treatment
The conventional ladder
5-aminosalicylates (mesalamine) are topical anti-inflammatories acting on the colonic mucosa. They are a mainstay of mild-to-moderate ulcerative colitis, for both induction and maintenance [2], and have little role in Crohn's disease.
Corticosteroids work quickly for flares and are among the most effective induction agents available. They are also unsuitable for maintenance — the toxicity is cumulative and unavoidable, and, as the centerpiece data show incidentally, patients on concomitant steroids healed less often, not more [22]. Getting patients off steroids is itself a treatment goal, and "steroid-free remission" is the endpoint most guidelines use for that reason [12].
Immunomodulators — thiopurines (azathioprine, mercaptopurine) and methotrexate — are slow-acting maintenance drugs. Their other role is to reduce immunogenicity against biologics, which matters more than it sounds: antibodies to infliximab predict loss of response and shorter duration of benefit [25].
Biologics, sorted by what they block
Anti-TNF agents came first and changed the disease. A single infusion of the chimeric anti-TNF antibody cA2 — infliximab — produced clinical response at four weeks in 81 percent of patients at 5 mg/kg against 17 percent on placebo, with remission in 33 versus 4 percent, in Crohn's disease resistant to everything else [26]. It also closed fistulas, which nothing previously had [27], and adalimumab established subcutaneous maintenance [28]. The class carries a real infection risk, most notably reactivation of latent tuberculosis, which is why screening before starting is mandatory [29].
Anti-integrin therapy is the gut-selective option. Vedolizumab blocks lymphocyte trafficking into intestinal tissue specifically rather than suppressing immunity systemically. In GEMINI 1, response at six weeks was 47.1 percent against 25.5 percent on placebo, with maintenance remission clearly superior to the 15.9 percent on placebo and an adverse-event frequency similar to placebo [30]; a parallel trial established efficacy in Crohn's disease [31]. That safety profile is the point of the mechanism.
Anti-IL-12/23 and anti-IL-23 agents target the pathway the laboratory work implicated. Ustekinumab, a monoclonal antibody against the p40 subunit shared by interleukin-12 and interleukin-23, was effective for induction and maintenance in moderately-to-severely active Crohn's disease [32] — a direct translation of the mechanistic studies above [6] [8].
JAK inhibitors are oral small molecules that block intracellular cytokine signalling rather than a single cytokine. In the OCTAVE programme, tofacitinib produced remission at eight weeks in 18.5 percent versus 8.2 percent and 16.6 versus 3.6 percent in two induction trials, and remission at 52 weeks in 34.3 percent (5 mg) and 40.6 percent (10 mg) against 11.1 percent on placebo [33] [34]. The trials also reported more overall and herpes zoster infection, five non-melanoma skin cancers versus one, five cardiovascular events versus none, and raised lipids [33] — and a dedicated safety trial in rheumatoid arthritis subsequently confirmed cardiovascular and cancer signals for the class [35]. Efficacy is not the only axis on which these drugs differ.
Surgery
Surgery is not failure; it is part of the disease's management. In ulcerative colitis, colectomy removes the diseased organ and, with it, the disease. In Crohn's disease it does not, because the disease can recur anywhere — so the surgical principle is bowel preservation: resect as little as possible, as late as reasonable, because a patient has a finite length of intestine and a lifetime of disease [12] [1].
Pillar 3: what is unresolved
Positioning. With four or five mechanistic classes and no reliable way to predict which will work for a given patient, drug selection is largely sequential trial and error. The classes differ in speed of onset, in safety profile [30] [33] [35], and in efficacy for particular manifestations such as fistulising disease [27], but which patient needs which mechanism first is the field's central unanswered question — and each additional mechanism that arrives widens it rather than closing it.
How hard to treat to target. STRIDE-II sets endoscopic healing as the long-term target [21], and the arithmetic above shows what reaching it is worth. What it does not settle is how aggressively to escalate in a patient who feels entirely well with mild residual inflammation, where the benefit is statistical and the drug toxicity is immediate and personal.
Colorectal cancer risk and surveillance. Long-standing colitis raises colorectal cancer risk, and population-based cohort data have refined the estimate downward from older referral-centre figures [36]. Chronic inflammation driving tumorigenesis has a defined mechanistic basis [37]. The practical response is surveillance colonoscopy with dysplasia detection, for which international consensus now favours high-definition chromoendoscopy and targeted biopsy [38] — though how much surveillance actually changes outcomes has been questioned for a long time [39] [40], and better disease control may itself be the more effective cancer prevention.
Dig deeper in lmmol
Rheumatoid arthritis is IBD's closest relative in this collection and the comparison is unusually direct: the same anti-TNF agents, the same JAK inhibitors with the same cardiovascular and malignancy signals [35], the same treat-to-target philosophy, and the same problem of choosing among mechanisms without a predictive test. Long-standing colitis raises colorectal cancer risk, which links to colorectal cancer and to the surveillance question there [36] [38]. Corticosteroids and chronic inflammation both damage bone, which connects to osteoporosis. Anti-TNF therapy requires screening for latent infection before starting, which connects directly to tuberculosis [29] and, for the same immunosuppression logic, to HIV. Depression is both a consequence of IBD and a prospective predictor of relapse [10] — see depression. The full collection is at health.