The heart is a muscle, and like every muscle it needs a blood supply. It gets one from the coronary arteries, which branch off the aorta and run across the surface of the heart before diving into it. They are not large — a few millimetres across — and everything in this review follows from that.
Coronary artery disease is the slow narrowing of those arteries by atherosclerosis. Cholesterol-carrying particles enter the artery wall and are retained there; the wall responds with inflammation; over decades a plaque builds up. If a plaque narrows the artery enough, the muscle downstream cannot get the extra blood it needs during exertion, and the result is angina — chest pain or tightness on exercise, relieved by rest. That is the chronic disease.
A heart attack is a different event. A plaque ruptures or erodes, its contents meet the blood, and a clot forms on the spot. If the clot occludes the artery, blood flow stops and the heart muscle downstream begins to die. That is myocardial infarction — literally, death of heart tissue. It is not usually the biggest plaque that does it, which is one of the most important and least intuitive facts about this disease.
It is the leading cause of death worldwide. Between 1990 and 2010 the global burden of ischaemic heart disease rose by 29 million disability-adjusted life-years, a 29 percent increase, with 32.4 percent of that growth attributable to population ageing and 22.1 percent to population growth [1]. Age-standardised acute myocardial infarction incidence and angina prevalence fell over the same period — the disease is getting less common per person and more common in total, because there are more people and they are older.
Two things make this disease unusual among the major killers, and they organise everything below.
It is largely preventable, and we know by how much. The INTERHEART case-control study of 15,152 myocardial infarctions and 14,820 controls across 52 countries found that a small number of modifiable factors accounted for most of the risk: smoking (odds ratio 2.87 for current versus never, population attributable risk 35.7 percent for current and former versus never), a raised ApoB/ApoA1 ratio (odds ratio 3.25 for the top versus lowest quintile, attributable risk 49.2 percent for the top four quintiles versus the lowest), hypertension, diabetes, abdominal obesity, psychosocial factors, diet, physical activity and alcohol [2]. These were consistent across every inhabited continent.
And when it strikes acutely, it is a race. From the moment an artery occludes, heart muscle is dying. Everything about the emergency response to a heart attack is built around shortening that interval, and the evidence says the clock runs continuously rather than in steps.
This review is organised around two clocks — the decades-long one that decides whether you get the disease, and the minutes-long one that decides how much muscle you keep once it declares itself.
Centerpiece: how much you lower LDL, and for how long
The initiating event in atherosclerosis is not inflammation and it is not the artery wall. It is the subendothelial retention of apolipoprotein B-containing lipoproteins — LDL particles crossing into the artery wall and being trapped there by the matrix. The inflammatory response, the macrophages, the T cells, the plaque all follow from that retention, and the reason LDL-lowering drugs work is that fewer circulating particles mean a lower probability of entry and retention [3]. That is the mechanistic claim. The rest of this section is about how thoroughly it has been tested.
The left panel is a natural experiment run nine different ways. A meta-regression of 49 randomised trials — 312,175 participants, mean baseline LDL 3.16 mmol/L, 39,645 major vascular events — asked what the relative risk of a major vascular event is per 1 mmol/L (38.7 mg/dL) of LDL lowering, separately by how the lowering was achieved [4]. For statins it was 0.77 (95% CI 0.71–0.84). For established non-statin interventions that work by upregulating LDL-receptor expression — diet, bile acid sequestrants, ileal bypass surgery and ezetimibe — it was 0.75 (0.66–0.86), with a between-group difference of P = 0.72. Pooling all five gave 0.77 (0.75–0.79) [4].
Consider what those five things have in common, which is nothing at the level of mechanism. Statins inhibit an enzyme in the liver. Bile acid sequestrants are resins that bind bile acids in the gut lumen. Ileal bypass is an operation that removes the absorbing bowel. Ezetimibe blocks an intestinal sterol transporter. Diet changes what goes in. They converge on the same number because the only thing they share is the LDL reduction, and the LDL reduction is what the arteries respond to.
And it is falsifiable, because one class failed. Cholesteryl ester transfer protein inhibitors lowered LDL and produced an observed relative risk of 1.01 (0.94–1.09) against an expected 0.90 (0.89–0.91), P = 0.002 — worse than predicted, and not better than nothing [4]. Torcetrapib and dalcetrapib were tested in exactly this population and failed [5], as did niacin added to intensive statin therapy [6]. The claim is not "anything that moves a lipid number helps." It is specifically about LDL, delivered through the LDL receptor.
The genetics says the same thing, with different confounders. A Mendelian randomisation study used inherited variants in PCSK9 and in HMGCR — the gene encoding HMG-CoA reductase, the statin target — as instruments, across 112,772 participants with 14,120 cardiovascular events [7]. Per 10 mg/dL (0.26 mmol/L) lower LDL, the odds ratio for cardiovascular events was 0.81 (0.74–0.89) for PCSK9 and 0.81 (0.72–0.90) for HMGCR [7]. Two different genes, two different proteins, identical effect per unit of LDL. The effects were independent and additive.
The right panel is the part that changes how you think about prevention. The magnitude of benefit is not fixed per unit of LDL — it grows with how long the LDL has been lower. In 58 trials, for a 1.0 mmol/L reduction, ischaemic heart disease events fell by 11 percent in the first year, 24 percent in the second, 33 percent in years three to five, and 36 percent thereafter (P < 0.001 for trend) [8]. The artery does not respond instantly; it responds cumulatively.
The check the arithmetic was never given. The trials and the genetics both report an effect per unit of LDL, on different scales, and no one puts them on the same one. Rescaling the genetic estimate from 0.26 mmol/L to 1 mmol/L gives 0.81^(1/0.26) = 0.44 — a 55.5 percent risk reduction, against the 23 percent the trials deliver for the same 1 mmol/L over a few years. Lifelong exposure to an identical LDL difference is worth about 2.4 times as much as a drug producing it in middle age. Neither paper computes this, and it is the strongest available argument for treating LDL early rather than intensively-but-late.
The same conclusion arrives from an entirely separate direction. Among 3,363 black participants in the Atherosclerosis Risk in Communities study, 2.6 percent carried PCSK9 nonsense mutations, associated with a 28 percent reduction in mean LDL and an 88 percent reduction in coronary heart disease risk (hazard ratio 0.11) over 15 years [9]. A 28 percent LDL reduction from a drug does not produce an 88 percent risk reduction in five years. Carried from birth, it does. A PCSK9 missense variant is likewise associated with reduced risk of early-onset myocardial infarction across multiple European and US cohorts [10].
Two secondary checks the papers also leave undone. Grouping trials by the size of the LDL reduction and excluding the first two years, events fell by 20, 31 and 51 percent for mean reductions of 0.5, 1.0 and 1.6 mmol/L [8]. If risk fell by a constant proportion per unit, then anchoring on the 1.0 mmol/L result (relative risk 0.69) would predict 16.9 percent at 0.5 mmol/L and 44.8 percent at 1.6 mmol/L. The observed values, 20 and 51 percent, both exceed the anchor's own extrapolation. And the same paper's estimate of what statins can achieve — an average 1.8 mmol/L LDL reduction, cutting IHD events by about 60 percent after several years [8] — is a far larger number than the five-year trial results alone would suggest, for the same reason the genetics is larger.
One honest limit about the figure. Every y-value plotted is a reported figure. The x-positions are illustrative: "years three to five" is drawn at 4, "thereafter" at 7.5, the meta-regression at 5, and lifelong genetic exposure at 50. None of those durations is published as a number, and the shape of the curve between the trial points and the genetic point is an interpolation, not a measurement.
The teaching point. Lower LDL, fewer events — in proportion to how much you lower it and how long you keep it lowered. The route does not matter provided it works through the LDL receptor. And the earlier you start, the more each unit is worth.
Pillar 1: measurement and diagnosis
For prevention: estimating risk before anything has happened
Coronary artery disease is silent for decades, so prevention depends on estimating risk from measurable factors rather than waiting for symptoms. The Framingham approach built sex-specific prediction equations from age, diabetes, smoking, blood-pressure category and cholesterol category, and showed that a categorical scheme performs about as well as one using the continuous variables — with roughly 28 percent of coronary events in men and 29 percent in women attributable to blood pressure above high-normal, and 27 percent in men and 34 percent in women attributable to total cholesterol at or above 200 mg/dL [11]. The PROCAM score used eight variables ranked by importance — age, LDL cholesterol, smoking, HDL cholesterol, systolic blood pressure, family history of premature myocardial infarction, diabetes and triglycerides — and achieved an area under the curve of 82.4 percent against 82.9 percent for the full continuous model [12]. Improved algorithms for women followed [13], and guidelines now formalise risk assessment in asymptomatic adults [14] [15] [16].
What a risk score is for is worth stating plainly: it converts a set of individually unremarkable numbers into an absolute probability, so that the decision to treat can be made on expected benefit rather than on whether any single value looks abnormal. It is the reason a person with an entirely normal cholesterol can warrant a statin and a person with a mildly raised one may not.
Familial hypercholesterolaemia is the exception that risk scoring handles badly, because the exposure is lifelong and the scores are calibrated on middle-aged risk. Against a theoretical prevalence of 1 in 500 for the heterozygous form, fewer than 1 percent are diagnosed in most countries; direct screening in a Northern European population found approximately 1 in 200, implying 14 to 34 million people worldwide, with up to a 13-fold increased risk of coronary heart disease [17]. The underlying LDL-receptor variants are catalogued [18], age- and sex-specific LDL cutoffs have been developed for detection [19], and treatment recommendations exist for adults [20] and children [21]. This is the clearest practical application of the centerpiece: the harm is cumulative exposure, so the intervention has to start early.
For symptoms: the acute coronary syndrome spectrum
When a patient presents with chest pain that might be cardiac, the first question is not "what is the diagnosis" but "is an artery closing right now."
The ECG splits the field in two. Persistent ST-segment elevation indicates complete occlusion of an epicardial artery, and defines STEMI — a diagnosis that mandates immediate reperfusion without waiting for anything else. Without ST elevation, the presentation is a non-ST-elevation acute coronary syndrome, which is then divided by the biomarker: raised troponin makes it NSTEMI, normal troponin with ischaemic symptoms makes it unstable angina. Guidelines for each arm are extensive and long-standing [22] [23] [24] [25] [26] [27] [28] [29] [30] [31].
Troponin is what made that division possible. Cardiac troponin T and troponin I are components of the contractile apparatus of heart muscle; they appear in the blood when heart muscle cells die [32]. The founding observation was one of specificity rather than sensitivity: among 215 patients, creatine kinase-MB was raised in 59 percent of those with acute skeletal muscle injury, 78 percent of those with chronic muscle disease and marathon runners, and 3.8 percent of dialysis patients — while troponin I was elevated only in the six patients who had actually sustained myocardial injury [32]. That is why troponin replaced the enzymes.
Assay sensitivity then improved by orders of magnitude. In 718 consecutive emergency-department patients with suspected myocardial infarction (17 percent of whom had one), sensitive troponin assays achieved areas under the curve of 0.95–0.96 at presentation against 0.90 for the standard assay — and among those presenting within 3 hours of chest-pain onset, 0.92–0.94 against 0.76 [33]. The gain is concentrated exactly where it matters. Analytical validation of the high-sensitivity troponin T assay [34], serial-change protocols [35] and standardised educational materials on interpretation [36] followed.
Sensitivity has a price, and it is the central practical problem with troponin. Because the assay detects any myocardial injury, it detects a great deal that is not a heart attack. Elevated troponin occurs after non-cardiac surgery, where it independently predicts 30-day mortality [37]; in heart failure [38]; in myocarditis [39]; in critical illness [40]; in dialysis patients [41]; after prolonged exercise [42]; in neuromuscular disease [43]; and after percutaneous coronary intervention, where it directly reflects procedural myocardial injury [44]. A highly sensitive assay detects troponin T in a substantial share of the general population [45], where it predicts coronary heart disease, heart failure and mortality [46] and carries prognostic information even in stable coronary disease [47]. The differential diagnosis of a raised troponin is therefore a clinical skill in its own right [48] [49] [50] [51] [52] [53]. This is exactly why myocardial infarction has a formal, repeatedly revised definition requiring a rise or fall of troponin together with clinical evidence of ischaemia — not a threshold alone [54] [55] [56] [57] [58].
Risk scores triage the non-ST-elevation group. The TIMI risk score uses seven variables — age 65 or over, at least three coronary risk factors, prior stenosis of 50 percent or more, ST-segment deviation, at least two anginal episodes in 24 hours, aspirin use in the prior week, and elevated cardiac markers — and event rates rose from 4.7 percent at a score of 0–1 to progressively higher strata [59], validated in unselected populations [60]. GRACE registry data provided the observational counterpart [61] [62], and bleeding risk has its own score [63], which matters because the treatments are antithrombotic.
Imaging, and what it is actually for
Coronary CT angiography visualises the arteries non-invasively; its diagnostic performance with 64-detector systems was established early [64] [65] [66] [67]. The SCOT-HEART trial randomised 4,146 patients with suspected stable angina to standard care with or without CT angiography [68]. At 6 weeks it reclassified the diagnosis of coronary heart disease in 27 percent of patients and of angina due to coronary heart disease in 23 percent, against 1 percent in standard care, changing planned investigations (15 percent versus 1 percent) and treatments (23 percent versus 5 percent) [68]. At 5 years the primary endpoint of coronary death or non-fatal myocardial infarction was 2.3 percent versus 3.9 percent (hazard ratio 0.59, 0.41–0.84, P = 0.004) — achieved with no significant increase in invasive angiography (hazard ratio 1.00) or revascularisation (1.07), but with more preventive therapy started (odds ratio 1.40) [69].
That result deserves emphasis because it is easily misread. The benefit of the scan came not from finding lesions to fix but from identifying who should be on prevention. Related work has used CT to guide management directly [70], to characterise plaque features that predict events [71] [72] [73] [74], and in the emergency department [75] [76] [77]. Coronary calcium scoring improves risk classification, particularly in the elderly [78] [79] [80] [81] [82] [83].
Invasive angiography with physiological measurement remains the reference for deciding on revascularisation. Fractional flow reserve measures whether a stenosis actually limits flow, and the discordance between angiographic and functional severity is large [84] [85]. In FAME 2, patients with stable disease and at least one functionally significant stenosis (FFR ≤ 0.80) randomised to FFR-guided PCI had a primary endpoint rate of 4.3 percent versus 12.7 percent on medical therapy alone — but the difference was driven by urgent revascularisation (1.6 percent versus 11.1 percent), and patients whose stenoses were not functionally significant did well on medical therapy alone [86]. Read carefully, that trial is as much about which lesions to leave alone as about which to treat.
Ischaemia without obstructive coronary disease is a real and under-recognised entity, particularly in women: microvascular dysfunction, impaired coronary flow reserve and abnormal perfusion reserve occur with unobstructed epicardial arteries [87] [88] [89] [90] [91] [92]. A normal angiogram is not the same as a normal coronary circulation.
Pillar 2: management — the two arms
Arm one, the acute heart attack: time is muscle
When an artery is occluded, the only treatment that matters is opening it, and the only question is how fast.
The relationship between delay and death is continuous, not stepwise. In 1,791 STEMI patients treated with primary angioplasty, the association between ischaemic time and one-year mortality was modelled as a continuous function; after adjustment for age, sex, diabetes and previous revascularisation, each 30 minutes of delay carried a relative risk for one-year mortality of 1.075 (95% CI 1.008–1.15, P = 0.041) [93]. The authors' conclusion was that every minute counts. This is the direct counterpart of "time is brain" in stroke, and the delay literature is comparably developed [94] [95].
Primary PCI is the preferred reperfusion strategy where it is available, established against thrombolysis in a quantitative review of 23 randomised trials [96]. Where it is not immediately available, thrombolysis is given rather than waiting — and given as early as possible, with prehospital thrombolysis studied in a meta-analysis of randomised trials against in-hospital administration [97]. Fibrinolytic agents and their pharmacology are characterised [98], with defined pathways for what follows: rescue angioplasty after failed fibrinolysis [99] [100], and routine early PCI after successful fibrinolysis [101] [102] [103].
The systems-of-care work is where most of the recent gain has come from. Prehospital ECG diagnosis with direct referral to an interventional centre substantially shortens time from ambulance call to first balloon inflation [104] [105]. Bypassing the emergency department entirely for patients already identified by prehospital ECG shortened first-medical-contact-to-device time from a median of 88 to 68 minutes in 12,581 patients across 371 US hospitals — while revealing that ED bypass occurred in only 10.5 percent of eligible patients, 18.3 percent during working hours against 4.3 percent off-hours, with hospital rates ranging from 0 to 71 percent [106]. The variation is the finding. Radial rather than femoral arterial access has become standard [107] [108].
Adjunctive drug therapy surrounds the procedure. Dual antiplatelet therapy is the backbone: clopidogrel added to aspirin in 12,562 patients with non-ST-elevation acute coronary syndromes reduced the composite of cardiovascular death, myocardial infarction or stroke from 11.4 to 9.3 percent (relative risk 0.80, 0.72–0.90) [109] [110]. More potent P2Y12 inhibitors improved on it: prasugrel versus clopidogrel in 13,608 patients reduced the primary endpoint from 12.1 to 9.9 percent (hazard ratio 0.81) with a bleeding cost [111] [112] [113], and ticagrelor versus clopidogrel in 18,624 patients reduced it from 11.7 to 9.8 percent (hazard ratio 0.84) with reductions in myocardial infarction and vascular death [114] [115] [116] [117]. A practical detail with real consequences: morphine delays and attenuates ticagrelor absorption and action in myocardial infarction [118].
Anticoagulation during the acute phase has its own long evidence base — low-molecular-weight heparin [119] [120] [121] [122] [123], fondaparinux [124] [125] [126], bivalirudin [127] [128] [129] and direct thrombin inhibition [130] — as do glycoprotein IIb/IIIa inhibitors [131] [132] [133]. An early invasive strategy benefits higher-risk non-ST-elevation patients [134] [122], and even minor troponin elevations identify who gains from it [135]. Cardiogenic shock complicating infarction is the extreme case, where early revascularisation improves survival [136] [137], and out-of-hospital cardiac arrest has its own pathway [138]. In STEMI with multivessel disease, complete revascularisation has been compared against treating only the culprit lesion [139].
Not every adjunct worked, and the failures are informative. Thrombus aspiration during primary PCI improved myocardial reperfusion markers [140]; remote ischaemic conditioning reduced injury in some trials [141] [142]; complement inhibition with pexelizumab was tested and did not deliver [143] [144]. Reperfusion itself causes injury — the paradox that restoring flow damages the muscle it rescues [145] [146] [147] [148] — which is why so much effort has gone into adjuncts that have mostly failed to add to simply being fast.
Arm two, chronic disease and prevention: the pillars
Statins are the foundation, and the trial record is unusually complete. Secondary prevention was established in 4,444 patients with coronary heart disease in 4S [149] [150], primary prevention in men with hypercholesterolaemia in WOSCOPS [151] and in men and women with average cholesterol in AFCAPS/TexCAPS [152], and breadth across 20,536 high-risk individuals in the Heart Protection Study [153]. More intensive lowering then beat less intensive lowering — in stable coronary disease, atorvastatin 80 mg achieved a mean LDL of 77 mg/dL against 101 mg/dL with 10 mg and reduced major cardiovascular events [154]; after acute coronary syndromes, atorvastatin 80 mg achieved a median 62 mg/dL against 95 mg/dL with pravastatin 40 mg [155] [156], and starting early after the event was tested directly [157] [158]. JUPITER extended treatment to people with normal LDL but raised C-reactive protein, where rosuvastatin lowered LDL by 50 percent and hs-CRP by 37 percent and reduced events [159] [160]. Statins work in hypertensive patients with average cholesterol [161], in type 2 diabetes [162] [163] [164] [165], in the elderly [166], in Japan [167] and in women [168] [169] [170] — though not in every population: they did not benefit patients on haemodialysis [171] or chronic heart failure [172], and the pattern of where they fail is itself instructive.
Statin safety deserves accuracy rather than reassurance. Serious myopathy with significant creatine kinase elevation is rare — 1 per 1,000 to 1 per 10,000 people on standard doses — while the broader category of statin-associated muscle symptoms, usually with normal or minimally raised CK, has a reported prevalence of 7 to 29 percent in registries and observational studies [173]. The gap between those two numbers is the clinical problem. A genome-wide study found a single strong association with myopathy at a variant in SLCO1B1, which encodes the transporter regulating hepatic statin uptake [174] — a real, mechanistically coherent susceptibility. Intensive-dose statin therapy also carries a modest excess of new-onset diabetes compared with moderate doses [175], consistent with the genetic finding that HMGCR variants raise diabetes risk while lowering cardiovascular risk, with the diabetes effect smaller in magnitude and confined to those with impaired fasting glucose [7]. Broader assessments of statin safety and its evidence base are available [176] [177] [178] [179] [180] [181] [182].
Beyond statins, the LDL story continues to hold. Adding ezetimibe to simvastatin after acute coronary syndromes in 18,144 patients lowered median LDL from 69.5 to 53.7 mg/dL and reduced the 7-year event rate from 34.7 to 32.7 percent (hazard ratio 0.936, P = 0.016) — a modest effect, and exactly the size the LDL difference predicts [183] [184] [185]. PCSK9 inhibitors go much further: evolocumab lowered LDL by 59 percent to a median of 30 mg/dL in 27,564 patients on statins and reduced the primary endpoint from 11.3 to 9.8 percent (hazard ratio 0.85), with benefit consistent even in the lowest baseline-LDL quartile [186] [187]; alirocumab after acute coronary syndrome in 18,924 patients reduced events from 11.1 to 9.5 percent (hazard ratio 0.85) and all-cause death from 4.1 to 3.5 percent [188] [189]. The development path from the discovery that PCSK9 mutations cause autosomal dominant hypercholesterolaemia [190], through the mouse knockout [191] and the observation that statins upregulate PCSK9 [192], to monoclonal antibodies [193] [194] [195] [196] [197] [198] [199] [200] [201] [202] and coronary plaque regression [203], is one of the cleanest genetics-to-drug sequences in medicine.
Blood pressure control is the second pillar, and its own meta-analysis of 147 randomised trials in 464,000 people quantified how much of the benefit is attributable to the pressure reduction itself rather than to drug class — with beta blockers showing a specific extra effect in preventing recurrent coronary events in people with existing coronary disease [204]. ACE inhibition reduces events in stable coronary disease [205], though not uniformly across trials [206]. Guidelines integrate the targets [207].
Antiplatelet therapy for secondary prevention, smoking cessation, diabetes and weight management complete the set, and the guideline statements bundle them explicitly [208] [209] [210] [211]. Smoking is the single largest modifiable attributable risk in INTERHEART [2], its cardiovascular effects are characterised for both active and passive exposure [212] [213], and psychosocial cessation interventions have been assessed specifically in coronary heart disease [214]. Aspirin's role differs sharply by setting: it is established in secondary prevention, while in primary prevention the balance against bleeding is much closer and has been examined in dedicated trials and recommendations [215] [216] [217] [218] [219]. Stopping aspirin is itself a hazard — coronary syndromes cluster after withdrawal [220], and premature discontinuation of dual antiplatelet therapy after stenting is a recognised danger [221] [222]. How long to continue dual therapy is a genuine trade-off between ischaemic and bleeding risk, addressed in dedicated trials and focused guideline updates [223] [224] [225] [226] [227] [228] [229].
Cardiac rehabilitation is the pillar most often left out, and it should not be. Exercise-based cardiac rehabilitation for coronary heart disease is supported by a Cochrane review [230] [231], home-based and centre-based delivery have been compared [232], and it retains prognostic benefit in the era of routine revascularisation and statins [233] [234]. It also shows a dose-response: among 30,161 elderly Medicare beneficiaries, patients attending 36 sessions had a 14 percent lower risk of death and 12 percent lower risk of myocardial infarction than those attending 24, and 22 and 23 percent lower than those attending fewer still [235]. More sessions, fewer events — the same shape as the LDL curve, in a different currency.
Revascularisation in stable disease: the part that is most often misunderstood
For symptoms, revascularisation works. For preventing infarction and death in stable disease, the evidence is much weaker than intuition suggests, and the reason connects directly to the biology.
In the COURAGE nuclear substudy, adding PCI to optimal medical therapy produced greater reduction in ischaemic myocardium (−2.7 percent versus −0.5 percent, P < 0.0001) [236] — a real physiological effect. But the events that kill people in this disease frequently do not arise from the lesion that was fixed. In a prospective natural-history study, 697 patients with acute coronary syndromes underwent three-vessel angiography and intravascular ultrasound after PCI; over a median 3.4 years the cumulative major adverse cardiovascular event rate was 20.4 percent, of which 12.9 percent were related to the treated culprit lesion and 11.6 percent to untreated non-culprit lesions — and most of those non-culprit lesions had been angiographically mild at baseline, with a mean diameter stenosis of 32.3 percent [237].
That is the fact that reconciles everything. Stenting relieves the obstruction that causes angina. It does not treat the plaque biology that causes infarction, and the plaque that ruptures next is often one that looked unimportant. It is why systemic therapy — LDL lowering, blood pressure, antiplatelet, smoking cessation — carries the prognostic burden, and why an anatomically impressive procedure can leave prognosis roughly unchanged.
The pathology backs this up. In 20 patients who died of acute myocardial infarction, the site of plaque rupture or superficial erosion was dominated by macrophages and T lymphocytes irrespective of the dominant plaque morphology — inflammation at the rupture site, not simply bulk [238]. And rupture is not the only mechanism: in 50 consecutive sudden coronary deaths, 28 had fibrous cap rupture into a lipid pool while 22 had superficial erosion of a proteoglycan-rich plaque with no rupture at all — the eroded group being younger (mean 44 versus 53 years) and much more often women (50 percent versus 18 percent) [239]. Plaque erosion has since been identified in living patients by intracoronary imaging [240].
Choice of revascularisation modality — PCI versus coronary artery bypass grafting — depends on anatomy, diabetes and completeness of revascularisation, and is the subject of its own guideline literature [241] [242] [243] [244] [245] [246] [247], with long-term comparisons in multivessel [248] and left main disease [249] [250] [251]. Aspirin improves graft patency after bypass [252] [253].
The biology, briefly, because it explains the drugs
Retention of apolipoprotein B lipoproteins in the subendothelial space initiates the lesion [3]. Endothelial dysfunction is both an early marker and a mechanism, detectable non-invasively in children and adults at risk [254], concordant between coronary and peripheral circulations [255], and improved by cholesterol lowering [256] [257] [258]. Inflammation is not a bystander: it is central to lesion development and to the acute event [259] [260] [261] [262]. Monocytes and macrophages accumulate in plaque through defined chemokine receptors [263] [264] [265] [266]; macrophage death and failure to resolve inflammation drive necrotic core formation [267]; macrophages break down collagen in fibrous caps [268] [269]; mast cells [270] and neutrophil extracellular traps [271] contribute; smooth muscle cells are central to cap stability [272] [273]; CD40 ligand is expressed across the vascular cell types [274] [275]; and endothelial shear stress patterns govern where lesions form and how arteries remodel [276]. Statins have effects beyond LDL — on endothelial progenitor cells [277] [278] [279] [280], reendothelialisation and inflammation — collected under the heading of pleiotropy [281] [282] [283], and C-reactive protein falls with treatment and tracks outcome [284] [285] [286] [287]. Inflammatory biomarkers predict events in men and women [288] [289] [290] [291] [292] [293] [294] [295] [296] [297] [298].
Pillar 3: progress
Lowering LDL further, and more easily. The PCSK9 antibodies proved that LDL well below previous targets still yields benefit [186] [188]. Inclisiran is the delivery innovation: a small interfering RNA that inhibits hepatic PCSK9 synthesis, given subcutaneously on day 1, day 90 and every six months thereafter, producing LDL reductions of about 50 percent (52.3 percent in ORION-10, 49.9 percent in ORION-11) in patients already on maximally tolerated statins [299]. Twice-yearly dosing addresses the failure mode that limits every chronic preventive therapy, which is that people stop taking them [300]. Antisense inhibition of apolipoprotein B synthesis has also been developed for severe hypercholesterolaemia [301].
Targeting inflammation independently of lipids. This was a genuine open question and it has now been answered in both directions. CANTOS tested canakinumab, an interleukin-1β antibody, in 10,061 patients with previous myocardial infarction and hs-CRP of at least 2 mg/L [302]. It lowered hs-CRP substantially without changing lipid levels, and the 150 mg dose reduced the primary endpoint (hazard ratio 0.85, P = 0.021) — proving the inflammatory hypothesis — but was associated with more fatal infection and no reduction in all-cause mortality [302]. Colchicine then showed benefit in chronic coronary disease: in 5,522 patients over a median 28.6 months, the primary endpoint occurred in 6.8 percent versus 9.6 percent (hazard ratio 0.69, P < 0.001), with a non-significant excess of non-cardiovascular death [303] [304]. But in 7,062 patients randomised soon after myocardial infarction and followed a median 3 years, colchicine did not reduce the composite primary outcome — 9.1 percent versus 9.3 percent, hazard ratio 0.99, P = 0.93 — despite lowering C-reactive protein by 1.28 mg/L [305]. Earlier work had shown colchicine reduces infarct size after STEMI [306], and interleukin-6 receptor blockade attenuates troponin release [307]. The honest summary is that inflammation is causal, that anti-inflammatory therapy helps in stable disease, that it did not help when started acutely in the largest trial to test it, and that infection risk is the constraint on the potent agents.
Better systems of care. Most of the mortality gain of the last three decades came from organisation rather than from new molecules. Over 1987–2008 in four US communities, coronary heart disease death rates among people without prior infarction fell an average of 4.7 percent per year in men and 4.3 percent in women, with incident myocardial infarction falling 4.3 percent annually among white men and 1.5 percent among black men [308] — a disparity that is itself the finding. Trends in STEMI and NSTEMI incidence, treatment and outcome have been tracked [309] [310], along with how quickly guidelines reach practice [311].
Risk-based and imaging-guided prevention. SCOT-HEART's 5-year result — fewer infarctions with no more revascularisation, achieved by starting more preventive therapy [69] — is the template: use imaging to decide who needs the pills, not who needs the stent.
Sex differences remain unclosed. Women present later and are treated more slowly: in 13,451 patients undergoing PCI, women with STEMI had longer symptom-to-door and door-to-balloon times and higher 30-day mortality (9.3 percent versus 6.5 percent, P = 0.005) [312]. Sex-specific differences in presentation, mechanism and outcome are the subject of a dedicated scientific statement [313], and women are over-represented in plaque erosion [239] and in ischaemia without obstructive coronary disease [87] [88]. Trends and treatment gaps have been documented repeatedly [314] [315] [316].
What is still missing. No therapy reverses established plaque to normal artery. Hormone therapy failed decisively in secondary prevention [317] [318]. Cell therapy for ischaemic myocardium remains exploratory [319]. And a large share of the world's coronary burden sits in countries where the constraint is not knowledge but delivery — of aspirin, of a statin, of a functioning ambulance system with a prehospital ECG.
Dig deeper in lmmol
Heart failure is where this disease most often ends: coronary disease is its leading cause, and infarction destroys muscle that the remaining heart must compensate for — the mechanisms linking infarction to subsequent failure in the primary-PCI era are set out directly [320], and reading the two reviews together shows why the acute clock matters long after the acute event. Hypertension is the second pillar of prevention and shares the meta-analytic logic of the centerpiece almost exactly: the benefit tracks the size of the pressure reduction, largely irrespective of drug class [204]. Diabetes and glycemic control is both a major risk factor in INTERHEART [2] and a place where the statin evidence is unusually strong [321] [163] — and where the HMGCR genetics shows the trade-off honestly, with lower LDL and slightly higher diabetes risk arising from the same variants [7]. Obesity supplies the abdominal-adiposity risk factor and the metabolic mediators through which body-mass index acts on coronary risk [322]. Stroke is the closest structural parallel: the same atherosclerotic and thrombotic biology in a different organ, the same time-critical reperfusion logic, and LDL lowering reduces both — by about 10 percent per 1 mmol/L for all stroke, against far more for coronary events, a difference the same meta-analysis quantifies [8]. On the molecular side, cardiac troponin T is the protein whose release defines myocardial infarction [32] [54], and HMG-CoA reductase is the statin target that Mendelian randomisation used as its genetic instrument [7]. The full collection is at health.