LAUSR

The patent foramen ovale (PFO) is not considered a disease as it is present in 20–30% of the population. Yet, it is of significant concern. Silent most of the time, the PFO can act up as a serious problem in its Mister Hyde role as the fundamental reason for paradoxical embolism engendering death, stroke, myocardial infarction, and peripheral ischaemia. More likely than late spontaneous PFO closure during life, selective mortality must be blamed for the well-documented decreasing prevalence of the PFO with age. Observational data on PFO carriers with thrombo-embolic events and long-term followup of random patients with or without PFO closure underpin the PFO’s licence to kill. This issue of the journal includes an insightful meta-analysis encompassing all currently published randomized trials comparing PFO closure (mostly with continued medical therapy) with medical therapy alone in patients with cryptogenic cerebral (or systemic) ischaemic events. The superiority of PFO closure was the writing on the wall from the first randomized trial and, even before, from comparative data. However, it fell short of statistical significance in the initial randomized trials, in part because intention-to-treat analyses were used. Such analyses make little sense when implanted devices are at stake. There is no placebo effect involved and a device that stays on the shelf can hardly protect a patient against anything. Now, predefined P-values have been reached by studies concentrating on highrisk PFOs or following patients up for long enough; it took 3440 patients and more than 4 years of follow-up. Most turned their backs on PFO closure when it missed a superiority trial with a P-value of 0.08 only to embrace it enthusiastically when the P-value has dropped to 0.046 with extended follow-up. How strange is that? The positive result with an occluder not reaching the closure rates of Amplatzer occluders points out that a device may act as a filter even when it is not watertight. This may not be the case after suture closure of the PFO. The meta-analysis in this issue highlights the crucial point that a large PFO is more dangerous than a small PFO; why not throw in a PFO with an atrial septal aneurysm opening the PFO with virtually every heartbeat as well as a PFO with a Eustachian valve or a Chiari network, guiding the clots from the inferior vena cava straight on to the gap? The meta-analysis also expressly calls for rapid modifications of the guidelines. Guidelines so far have not acknowledged the blatant proof of non-inferiority of PFO closure vs. pure medical therapy, because the respective trials were designed as superiority trials. Failed superiority trials cannot be used to establish equipoise according to the modern stringent publication rigour. To the patient, a 15-min painless ‘mechanical vaccination’ must look more attractive than lifelong blood thinners with their ever-accruing, and with increasing age higher and more threatening, risk of bleeding. The guidelines simply do not condone such an offer to the patient. The outspoken plea to correct that adds value to this meta-analysis in comparison with recent similar publications. Focusing on large PFOs, in addition, sounds trivial at first but has important implications. Large PFOs can be found by simple transthoracic contrast echocardiography; small PFOs are evident only by cumbersome and unpleasant transoesophageal echocardiography. This puts a foot in the door of screening for large PFOs with the purpose of primary prevention. People before major surgery, with a high risk for venous thrombo-embolism, or of families with PFO problems might be a good place to start. In real life, the frequency of PFO-mediated ischaemic events is underestimated. Other causes are preferentially blamed, such as atrial fibrillation, ruptured atherosclerotic plaques, or spontaneous cerebral or coronary dissections. The PFO is almost invariably the last on the list to be considered, let alone looked for. With a ratio of coronary to cerebral blood flow of 1:3 (220 versus 750 mL/min), one PFO-mediated myocardial infarction per three PFO-mediated cerebral ischaemic events would make more sense than the 1:100 000 we seem to find in the literature. By the same token, in a large field study, high rates of coincidental strokes and myocardial