Paradoxical embolism,stroke and sclerotherapy

Diagnostic criteria

Paradoxical embolism, first described by Connheim in 1877, is the transmission of a venous embolus to the arterial circulation via a defect in the cardiac septum or a pulmonary arteriovenous malformation (AVM). ^34,35

The traditional diagnostic criteria for paradoxical embolism include: ^36–38

Arterial embolism with no evidence of a source in the left heart or arterial circulation;

The foramen ovale

The pulmonary circulation is virtually bypassed in fetal circulation. The fetal lungs are collapsed and the placental blood enters the right atrium from the inferior vena cava and crosses into the left atrium via the foramen ovale. The foramen ovale is an opening in the interatrial septum that functions as a one-way valve to allow RLS of oxygenated blood in utero. Ductus arteriosus, a shunt connecting the pulmonary artery to the aortic arch, is the other fetal cardiac shunt that allows blood entering the right ventricle to bypass the pulmonary circulation. When the lungs expand at birth, the pulmonary pressure is reduced, which draws the blood from the right atrium into the right ventricle and onto the pulmonary circulation. This results in an increase in the left atrial pressure causing a functional closure of the foramen ovale. In the first two years of life, the subsequent fusion and fibrosis of the foramen ovale causes a permanent closure of this shunt forming the fossa ovalis in the adult heart. ⁴⁰

Patent foramen ovale

PFO is a persistent inter-atrial opening resulting from incomplete fusion of the foramen ovale after birth. This abnormality may be hereditary in some families and is associated with an autosomal dominant pattern of inheritance. ⁴¹ PFO is found in up to 30% of the population ⁴² and in up to 60% of patients with cryptogenic strokes (Table 1). ⁴³ Autopsy findings report the frequency of PFO to be even higher at 40% in adult hearts and the mean PFO diameter to be 4.9 mm. ⁴⁴ Even a higher prevalence of a RLS (38.5% at rest, 51.5% after Valsalva) was found in patients with symptomatic great saphenous incompetence and varicose veins. ⁴⁵ The prevalence of PFO diminishes to 20% while the average diameter increases to 5.8 mm at 80 years of age. ⁴⁶ This is presumably due to spontaneous closure of smaller shunts with age. ⁴⁴ The incidence of PFO is not affected by gender.

OPEN IN VIEWER

Table 1

Reported incidences of patent foramen ovale (PFO) in various populations

	PFO incidence (%)
General population 42	30
Autopsy findings 44	Up to 40
Varicose vein patients 45	Up to 51
Migraine patients 39,60,61	Up to 60
Cryptogenic stroke patients 43	Up to 60

Right- to-left shunt

A RLS is a communication that allows the venous blood to enter the arterial circulation by flowing through the heart. PFO is the most common cause of a RLS found in 70% of cases but other cardiac defects such as ventricular septal defect or atrial septal defect (ASD) can also result in a RLS. ⁴⁷ Extracardiac shunts with a reported prevalence of 10% can also produce a RLS. The most common extra-cardiac shunts are pulmonary AVMs through which paradoxical emboli reach the left atrium via the pulmonary vein. ⁴⁴

Normally, the left atrial pressure is higher than the right side which keeps a PFO closed. The right atrial pressure is elevated during Valsalva manoeuvres, coughing and early systole which can open the PFO and allow a transient RLS. ^48,49 Pulmonary arterial hypertension (that can follow PE) can increase the right atrial pressure inducing a RLS via a PFO. A PFO has a valve-like nature which would only permit a RLS, whereas a true ASD allows bidirectional flow. ⁵⁰

Sources of paradoxical embolism

The most presumed source of paradoxical embolism is lower limb DVT reported in approximately 90% of all cases while pelvic vein thrombosis may act as a rare source. ⁵¹ Thrombi could also form in situ within a PFO and enter the arterial circulation during an increase in the right atrial pressure. ⁵² Rarely, a PFO may ‘trap’ larger thrombi that would otherwise embolize to the brain. ⁵³ Clots arising from the left heart, for example during transient atrial arrhythmias, may also embolize to the brain but would no longer be defined as ‘paradoxical’. ⁵⁴

PFO, stroke and migraine

PFO has been implicated in a range of clinical conditions including stroke, migraine, decompression illness in scuba divers and cerebral fat embolism. ^55,56 Up to 40% of ischaemic strokes reported in young people are cryptogenic. ^39,57 The prevalence of PFO is higher in patients with cryptogenic stroke (60%) compared with the 30% prevalence in the general population (Table 1). ⁴³ Furthermore, the presence of a PFO significantly increases the risk of recurrent cerebrovascular events. ^39,58 Patients with cryptogenic stroke and a PFO are twice as likely to suffer from migraines compared with those without PFO. ⁵⁹ The prevalence of PFO is higher in patients with migraine headaches (up to 60%; Table 1), ^39,60,61 but this seems to be influenced by the migraine subtype. Patients with classic migraines (with aura) are more likely to have a PFO than those with common migraines (without aura) or those without migraines altogether. ⁶² Similarly, although the overall risk of stroke in migraine patients remains low, ⁶³ the relative risk is higher for classic migraine compared with common migraine. ⁶⁴ In addition, patients with classic migraine are reported to have an increased prevalence of sub-clinical posterior circulation infarcts compared with those with common migraines. ⁶⁵ An entity that needs to be differentiated is the hemiplegic migraine which can closely mimic a stroke. ⁶⁶

Detection of PFO

Treatment of PFO

Paradoxical emboli pose a potential risk to patients with PFO and closure procedures are performed to reduce this risk. ⁷³ Treatment options have included surgical closure as well as interventional procedures including transcatheter techniques. ³⁹ The surgical approach, mostly used to prevent recurrent cryptogenic strokes, requires cardiopulmonary bypass and is associated with complications such as cardiac tamponade and ventricular fibrillation. The transcatheter method is minimally invasive and involves the percutaneous implantation of a PFO closure device to occlude the inter-atrial septal defect. A recurrence rate of up to 6% is associated with various closure devices and the reported complications are infrequent. ³⁹ PFO and other septal defects provide an escape route for arterial hypertension and in the presence of pulmonary arterial hypertension, closure of PFO is contraindicated.

Deep vein thrombosis

The true incidence of DVT following sclerotherapy is unknown and most publications include clinically (and not ultrasonically) detected DVTs. In the French registry of 12,173 procedures, only one proximal vein and five cases of calf vein thrombosis were reported. ²⁵ In the Australian Polidocanol Study, three cases of DVT (0.02%) were detected following 16,804 sclerotherapy procedures with polidocanol. ⁷⁴ In a clinical audit data of foam UGS in 7027 patients (11,537 procedures) from nine UK centres, 36 patients (0.5%) were diagnosed with DVT. ²⁹

The incidence of venous thromboembolic events following sclerotherapy is possibly higher as a significant number of procedural DVTs may be silent. ⁷⁵ In a study by Gillet et al., ²⁴ 1025 patients underwent foam UGS. Duplex ultrasound screening was performed between the eighth and 30th day in the majority of patients. Eleven thromboembolic events (1.07%) were reported which included 10 DVTs and one PE. Five DVTs were asymptomatic. Similarly, Bergan ⁷⁶ has reported an incidence of 1.8%. In a study by Myers et al., ⁷⁷ 489 patients underwent 1189 foam UGS procedures. Duplex ultrasound screening revealed 16 cases of DVT (3.2%) all of which were asymptomatic. ⁷⁷ The reported frequency of 1–3% is similar to the author's experience. ⁷⁵ Treatment factors that might influence the risk of deep vein occlusion (thrombosis or sclerosis) following foam sclerotherapy have been reviewed. ⁷⁸

Pulmonary embolism

In earlier publications, PE followed sclerotherapy of small superficial vessels with a reported incidence of one in 250 to one in 1000. ^79–81 In the study by Gillet et al., ²⁴ one case of PE was reported in 1025 patients. In the French registry of 12,173 procedures and the Australian Polidocanol Study involving 16,804 legs injected with polidocanol, no cases of PE were reported. ^25,74 There are no data regarding the incidence of silent PE following sclerotherapy.

Concurrent PE is found in 5% of the diagnosed cases of paradoxical embolism. ⁸² Patients with PE and a PFO > 4 mm have a 10-fold increased risk of death and five-fold increased risk of systemic embolism. ⁵³ Concurrent PE has been reported in only one case of stroke after sclerotherapy. ²

Superficial venous thrombosis

Clots originating in superficial veins can rarely embolize and PE is reported in up to 3% of all cases of superficial thrombophlebitis (STP). ⁸³ In the French registry, only three cases of STP were reported following 12,173 procedures. ²⁵ Although histologically, venous sclerosis (collagen deposition resulting in scar formation), venous thrombosis (intravascular fibrin clot formation) and venous thrombophlebitis (clot formation accompanied by an inflammatory infiltrate) are separate entities, these conditions are not always clinically or sonographically differentiated. ⁸⁴ Clinically, the diagnosis of venous thrombosis in a superficial vein has relied heavily on the co-presence of an inflammatory (phlebitic) component and hence superficial venous thrombosis (SVT) without an inflammatory component is possibly under-diagnosed in sclerotherapy patients. Even sonographically, SVT may be misdiagnosed as the thrombotic occlusion of a superficial vein and dismissed as an expected outcome of sclerotherapy (see below).

Failure to detect venous thrombosis on ultrasound in sclerotherapy patients

An acute venous thrombosis may remain undiagnosed and under-reported in sclerotherapy patients due to a variety of reasons:

Transient events

Transient neurological events such as visual disturbances and migraine following both liquid and foam sclerotherapy have been associated with the presence of a RLS and in particular a PFO. ^32,88 Visual disturbances following foam sclerotherapy have been attributed to classic migraine attacks rather than TIAs. ²⁶ Endothelin-1, a vasoconstrictor, has been implicated in the pathogenesis of migraine and has been shown to be released after sclerotherapy in an animal model. ⁸⁹ Serotonin, another vasoconstrictor, is also implicated in the pathogenesis of migraine. ⁹⁰ Serotonin modulates vasoreactivity in normal and atherosclerotic arteries, and contributes to cerebral ischaemia. ^91–93 In in vitro experiments, low concentration sodium tetradecyl sulphate (STS) induces serotonin release from activated platelets. ⁹⁴ Although sclerotherapy is known to induce venoconstriction of the target vessels, ³¹ arterial vasospasm or vasoconstriction at distant sites has not been studied.

There are few published reports of TIA following sclerotherapy (Table 2). ^6,24,95–97 Both gas and clot microemboli can cause transient neurological dysfunction, ⁹⁸ with recovery when the emboli have lysed and the circulation is restored. ^99,100 All reported cases of TIA after sclerotherapy had an immediate onset and a RLS was documented in all these cases (Table 2). Four cases followed the use of air-based foams and one case reported back in 1994 followed sclerotherapy with hypertonic saline in a patient with an atrial septal aneurysm (ASA). In none of these patients a concurrent deep or superficial venous thrombosis was identified.

OPEN IN VIEWER

Table 2

Published cases of stroke following liquid (Table 2a) and foam (Table 2b) sclerotherapy and transient ischaemic attacks (TIA, Table 2c)

AP, ambulatory phlebectomy; ASA, atrial septal aneurysm; ASD, atrial septal defect; CG, chromated glycerin; DSS, double syringe system; DVS, direct vision sclerotherapy; DVT, deep vein thrombosis; EVLA, endovenous laser ablation; F, female; Giac, Giacomini; GSV, great saphenous vein; HS, hypertonic saline; Immed, immediately; M, male; MCA, middle cerebral artery; MGV, medial gastrocnemius vein; NA, not applicable; O₂, oxygen; OCP, oral contraceptive pill; PAV, posterior arch vein; PCE, paradoxical clot embolism; PE, pulmonary embolism; PGE, paradoxical gas embolism; PFO, patent foramen ovale; POL, polidocanol; PTV, posterior tibial vein; Retics, reticular veins; RLS, right-to-left shunt; SCA, superior cerebellar artery; SFJ, saphenofemoral junction; SSV, small saphenous vein; STS, sodium tetradecyl sulphate; STP, superficial thrombophlebitis; Tels, telangiectasias; TIA, transient ischaemic attack; TOE, trans-oesophageal echocardiogram; UGS, ultrasound-guided sclerotherapy; VA, vertebral artery; VTE, venous thromboembolism

*Personal communication with the authors

Stroke

Probable pathogenic mechanisms

All reported PSS cases have been ischaemic and so far no haemorrhagic cases have been reported. In 10 of the 13 published cases, a RLS was identified and hence paradoxical embolism was considered the most probable cause of stroke (Table 2). In the remaining three, one patient was examined with TTE only (and not TOE), ¹ a non-cardiac shunt was suspected in one case ⁷ and further information was not available in the other case. ¹¹

The underlying pathogenic mechanisms in PSS can be categorized into:

Paradoxical gas embolism (PGE);

Gas embolism is a well-known complication of cardiopulmonary bypass, neurosurgery and laproscopic surgery. Venous gas embolism (VGE) is the entry and embolisation of gas in the central venous system. PGE occurs through the direct passage of gas into the arterial system of the cerebral and coronary circulations via a RLS. ^47,103–105 Even in the absence of a RLS, VGE resulting in a retrograde flow of gas via the jugular veins can cause cerebral infarction. ¹⁰⁶

Detergent sclerosants in the liquid form are usually mixed with a gas at a specific ratio such as 1 + 4 (1 unit of liquid, 4 units of gas) to generate a sclerosing foam. ¹⁰⁷ Bubbles are observed on concurrent echocardiography and TCD to enter the right heart, and in the presence of a RLS onto the brain-supplying and intracranial circulation during foam sclerotherapy. ^{5,6,88,97,108,109} Imaging of embolic bubbles can be challenging and even in the presence of severe neurological abnormalities, imaging for gas emboli can be negative as air and CO₂ are absorbed rapidly. ^110–114 Nonetheless, gas bubbles were visualized in the brain-supplying or intracranial arteries of five patients, all with an immediate onset after foam sclerotherapy. In one case, the patient underwent concurrent ambulatory phlebectomy (AP). ¹¹ A rapid onset is the key feature of PGE, also reported in all cases of cerebral ischaemia following the echocardiogram ‘Bubble Study’ (see above Bubble Study and Stroke). ⁷²

The mechanism of infarction in PGE may be due to:

Direct physical occlusion of intracranial arteries by gas bubbles;

The nomenclature referring to the observed bubbles has been varied and often less than accurate. Various authors have used a range of designations including ‘foam particles’, ‘foam bubbles’, ‘foam emboli’, ‘foam microemboli’ and ‘echogenic particles’. Sclerosant ‘foam’ is a substance comprised of gas bubbles trapped in the liquid sclerosant. What is observed on echocardiograms or neck ultrasound studies following foam sclerotherapy is a shower of bubbles. ^5,109 Currently, we have no clinical or imaging evidence that the bubbles observed in the carotid arteries or heart are trapped or float in the original liquid sclerosant, carry a surfactant layer of detergent or have any sclerosing activity. In fact, the limited data available point to the opposite (see below 3. Idiopathic and other causes). Hence, designations that imply the presence of active sclerosant or a sclerosant film on the surface of the embolic bubbles are not justified by the available evidence. It has also been suggested that the observed echogenic particles may indeed be endothelial cell debris. Although the presence of endothelial derived biological products such as microparticles cannot be ruled out, the observed echogenicity and the ring-down artefact is a typical sonographic feature of gas bubbles. ¹²²

2. Paradoxical clot embolism

In general, gas embolism is associated with an immediate/rapid onset of symptoms although in rare cases it can present with a delayed onset. ¹²³ A delayed onset has been the key feature of thrombotic complications of sclerotherapy as reported in the French registry of 12,173 procedures. ²⁵ A concurrent DVT was detected in only two of the PSS patients and in a third case, an extension of thrombus from the saphenofemoral junction into the deep venous system was observed (Table 2). These three cases had a delayed onset of a few days after the procedure. No concurrent venous thrombosis was detected in the remaining three cases with a delayed onset (Table 2). The failure to detect a concurrent venous thrombosis does not rule out such pathology and may be due to a variety of reasons discussed earlier (see above Failure to detect venous thrombosis on ultrasound in sclerotherapy patients). In addition, the contribution of other pathogenic mechanisms in these patients cannot be ruled out (see below 3. Idiopathic and other causes).

Co-presence of PFO and thrombophilic abnormalities including factor V Leiden and prothrombin gene mutations significantly increases the risk of cryptogenic stroke. ¹²⁴ None of the PSS patients tested were diagnosed with thrombophilia but two had coagulation abnormalities including elevated D-dimer levels in the absence of concurrent DVT (Table 2). ^1,4 This is not unique to sclerotherapy patients as increased levels of fibrinopeptide A and D-dimer levels have been documented for up to four weeks in other stroke patients, suggestive of ongoing fibrin formation and fibrinolysis. ¹²⁵ Moreover, sclerotherapy in the absence of concurrent DVT can result in the elevation of D-dimer levels. ¹²⁶ D-dimer levels may also be elevated in a number of chronic conditions ^127,128 and the observed elevations could have antedated the onset of stroke.

3. Idiopathic and other causes

No gas or clot emboli could be demonstrated in five of the 13 PSS patients reviewed here (2 with an immediate onset after liquid sclerotherapy, 3 with a delayed onset; Table 2). Even in cases attributed to PGE based on the visualization of gas emboli, and those attributed to PCE based on the detection of a concurrent DVT, a contribution from other pathogenic mechanisms cannot be ruled out.

The complexity of this topic is illustrated by the neurological ischaemic events with an immediate onset following liquid sclerotherapy. Such events were reported in two PSS cases ^1,3 and in one case of post-sclerotherapy. ⁹⁵ Clearly, PGE in such cases would be a remote possibility and the immediate onset renders the local formation of thrombus and its instant embolization from the calf to the brain a less likely cause. Furthermore, in none of these patients a concurrent deep or superficial venous thrombosis was diagnosed. Liquid sclerotherapy is also associated with transient neurological symptoms such as migraine and visual disturbances with an immediate onset. ⁹⁵ Such complications may be classified as distant adverse events of sclerosants and may be postulated to be due to a direct effect of the sclerosing agents or mediated by secondary biological by-products of sclerosants.

Currently, there is no evidence that detergent sclerosants maintain their sclerosing or surfactant activity at distant sites and although foam generated bubbles are routinely visualized to enter the cardiac chambers, there is no evidence that such bubbles carry a surfactant layer with detergent properties. Detergent sclerosants are strongly neutralized by blood cell membranes, ⁹⁴ albumin and other plasma proteins ¹²⁹ and blood samples collected from the cardiac chambers in an animal study have shown no sclerosing activity. ¹³⁰ It is unlikely for the injected agents to maintain direct sclerosing activity at distant sites and the observed distant adverse events are most likely mediated by the release of cellular and other biological by-products of sclerosants.

Detergent sclerosants are biologically active and interfere with the coagulation, antithrombotic and fibrinolytic mechanisms in vitro. ^129,131,132 Clinically, low-concentration sclerosants have been reported to cause visual disturbances and neurological complications. ¹³³ Low-concentration detergent sclerosants have been shown to activate platelets and release procoagulant platelet-derived microparticles (PMP) in vitro. ^94,129,131 A rise in plasma levels of PMPs is associated with arterial thrombotic events, ¹³⁴ stroke ^135–137 and myocardial infarction. ¹³⁸ In other in vitro studies, low concentration sclerosants demonstrate a prothrombotic antifibrinolytic profile as evident by an increase in plasminogen activator inhibitor-1 (PAI-1), tissue plasminogen activator (t-PA)/PAI-1 complexes, alpha-2 antiplasmin and thrombin activatable fibrinolyisis inhibitor. ¹³⁹ A systemic rise in the inhibitors of fibrinolysis has been associated with stroke, acute coronary occlusion ^140,141 venous, ¹⁴² and arterial thrombosis. ¹⁴³

Apart from thrombosis, platelet and endothelial by-products may result in sustained vasospasm and secondary thrombotic occlusion. One such vasoconstrictor is serotonin released from activated platelets. In in vitro experiments, serotonin has been recently shown to be released by activated platelets when stimulated by low concentration STS. ⁹⁴ Another vasoconstrictor, endothelin-1, ¹⁴⁴ was also shown to be released after sclerotherapy in an animal model. ⁸⁹ Sustained cerebral vasospasm may result in delayed ischaemic deficits. ¹¹⁵

The release of cell-derived by-products of sclerosants possibly plays a crucial role in the pathogenesis of neurological and other complications of sclerotherapy and need to be investigated in detailed clinical studies.

Finally, a coincidental event due to general causes of stroke should be considered in cases where a procedure-related paradoxical gas or clot embolism has been ruled out.

Patient's position during procedure

Patients undergoing sclerotherapy are usually treated in the recumbent position. The risk of PGE is increased in the sitting position. ^120,145 A number of manoeuvres (such as keeping the patient supine for a few minutes after the procedure and leg elevation before, during or after the procedure) were recommended to prevent VGE. ^146–150 Recent studies have shown these manoeuvres to be ineffective in prevention of VGE. ^109,151 Leg elevation was performed in all four PSS cases with an immediate onset of stroke (Table 2). Prolonged supine position does not prevent bubbles from reaching the right heart but may reduce the bolus volume compared with immediate sitting up or standing at the completion of the procedure.

Sclerosing agents

Both STS (6 foam cases) and polidocanol (3 foam cases, 3 liquid) were implicated in the published cases of PSS (Tables 2). The other sclerosant reported was chromated glycerine (a liquid chemical irritant). One case of TIA followed the use of hypertonic saline in a patient with an ASA. ⁹⁵ Based on the limited data, no conclusion can be drawn regarding the risk associated with particular sclerosants and randomised prospective studies are required to examine the relevant risk.

Sclerosing format

Four PSS cases have followed liquid ^1–4 and nine have followed foam sclerotherapy. ^5–11 The apparent rise in the number of cases associated with foam sclerosants may be due to the massive popularity of foam sclerotherapy and given the limited data, no conclusion can be drawn regarding the possible increased risk associated with one particular format. However, PGE would be a very unlikely complication of liquid sclerotherapy.

Sclerosant concentration

A wide range of sclerosant concentrations were reported in the published cases of PSS. These agents were occasionally used at low concentrations and in very small volumes treating sub-dermal vessels. ⁴ In in vitro experiments, low concentration sclerosants exhibit prothrombotic antifibrinolytic activity ¹³⁹ and initiate strong clots. ¹⁵² At mid-range concentrations (0.15% STS, 0.15–0.3% POL), both agents initiate loose clots prone to lysis while at high concentrations (>0.3%) both inhibit clot formation. ¹⁵² The thrombotic activity described in these in vitro studies relates to the ‘free’ final intravascular concentrations. Such ‘free’ concentrations are determined by the degree of mixing with blood and the subsequent neutralization by blood cells and plasma proteins. ¹²⁹ Hence, although the initial concentration injected by the practitioner may be high, the final intravascular concentration can drop substantially to a ‘pro-coagulant window’. ¹²⁹

The foaming gas

Air, carbon dioxide (CO₂) or a combination of oxygen and CO₂ are used to generate sclerosant foam. ^107,153 Room air, which contains 72% nitrogen, was used in all published cases of PSS (Table 2). CO₂ was used in one case of TIA after foam sclerotherapy, ⁹⁷ 100% oxygen in one case ²⁴ and room air in the remaining two cases (Table 2). So far, stroke has not been associated with the use of CO₂ foam and the reported neurological symptoms have been transient. ^130,150,154 CO₂ is approximately 20 times more soluble in blood than oxygen and nitrogen is the least soluble gas. ¹⁵⁵ CO₂ has not been used as much as air in foam sclerotherapy and is reported to produce fewer adverse reactions compared with air. ¹⁰⁸ As a contrast agent, CO₂ in up to 200–300 mL is injected intravenously during venography especially in renal disease or if there is a history of hypersensitivity to contrast agents. ¹⁵⁶ CO₂ is also used in a number of arterial access procedures including aortagraphy, renal and visceral angiography, transcatheter embolization and endovascular abdominal aortic aneurysm repair. CO₂ as an arterial contrast agent can cause PGE to spinal, coronary, and cerebral arteries especially when used in sites above the diaphragm. ^155,156 Injection of 20 mL of CO₂ in an arteriovenous fistula has resulted in cerebral infarction and there are other cases of stroke associated with the use of CO₂ in this setting. ¹¹⁰ Individual CO₂ bubbles are estimated to dissolve within 2–3 minutes. ¹⁵⁵ However, in a recent study by the author, CO₂ bubbles arrived in the right heart within 60 seconds and continued for another 25 minutes following a single injection of 2.5 mL of sclerosant foam in a small saphenous vein. ¹⁰⁹ Similar findings have been reported by Morrison et al. ¹⁵⁰

Total volume of sclerosant

Neurological symptoms are reported with as little as 20 mL of air injected intravenously. ¹⁰⁶ In humans, an intravenous injection of 200–300 mL of air (5 mL/kg) has been fatal. ¹⁵⁷ Much lower volumes can be fatal in the presence of a RLS or in critically ill patients. ¹⁵⁸ Injection of 2–3 mL of air into the cerebral circulation can be fatal. ¹⁵⁹

In the PSS cases following liquid sclerotherapy, relatively small volumes of liquid sclerosants (0.5–4 mL, mean 2.1 mL) were injected (Table 2). In cases following foam sclerotherapy, a wide range of volumes (2–25 mL, mean 12 mL, median 7 mL) were used (Table 2). The second European Consensus on foam sclerotherapy set an upper limit of 10 mL of foam per treatment session. ¹⁶⁰ The Australasian College of Phlebology Consensus recommendations have set this limit at 20 mL per session. ¹⁶¹ In a recent study, Morrison et al. ¹⁰⁸ compared the adverse effects of CO₂-based foam with a historical control using air-based foam. For volumes >15 mL, the incidence of complications were not directly related to the volume of foam injected. At higher volumes, the choice of foaming gas played a role and CO₂-based foam sclerosants were found to cause fewer complications compared with air-based foams.

Sclerotherapy techniques

Although the question of the total foam volume and its association with adverse events remains controversial, a more critical variable may be the volume of the injected gas that enters the deep venous system and subsequently the systemic venous circulation. The observation of embolic bubbles within minutes of injections indicates direct and rapid entry into the deep venous system. Such entry can happen via perforators or junctions with deep veins. Bubbles are observed to enter the right heart following the injection of 1–2.5 mL of the sclerosant foam. ¹⁰⁹ In a recent study, manual compression or ligation of the saphenofemoral junction did not prevent, but reduced the flow of foam into the femoral vein. ¹⁶² In a study by the author, once the junction was occluded by the initial foam injection, subsequent injections in the proximal trunk of the great saphenous vein (GSV) did not result in further systemic embolization of bubbles. ¹⁰⁹

As recommended by the ‘French School’, sclerosants are administered in a proximal to distal sequence of injections targeting the larger veins and the most proximal sources of reflux first. ¹⁶³ By contrast, following the original Fegan approach, sclerosants are injected distally, and often directly into visible superficial tributaries ¹⁶⁴ and when monitored by ultrasound, the infusion is continued until the foam has reached the junction. ⁹⁷ Multiple small injections, as against a continuous infusion, have been shown to reduce the risk of deep vein occlusion post-sclerotherapy. ¹⁶⁵ Treatment techniques possibly play a crucial role in the systemic dissemination of gas emboli and need to be compared in randomized trials.

Peri-venous tumescent anaesthesia ¹⁶⁶ and catheter-directed sclerotherapy ¹⁶⁷ are notable treatment modifications aimed at increasing the safety and efficacy of foam sclerotherapy. Perivenous tumescent anaesthesia may help to contain the injected foam to the intended target site.

Bubble size

The role of the bubble size in foam sclerotherapy is currently under investigation. ¹⁶⁸ Bubble size has been identified as a risk factor for cerebral ischaemia associated with the echocardiogram ‘Bubble Study’ (see above Bubble Study and stroke). ⁷² Cerebral gas embolization can cause acute ischaemia from physical obstruction of vessels by large bubbles. ¹⁶⁸ The average bubble radius produced with the Tessari method using air as the foaming gas is around 20 µm after 10 seconds, and 33 µm for STS and 38 µm for POL after 60 seconds. ¹⁰⁷ CO₂ foam bubbles are 25–30% smaller while the combined CO₂/O₂ bubbles (usually 70% CO₂, 30% O₂) are only 15–20% smaller than air-based foam bubbles. ¹⁰⁷ Although the smaller size may allow for rapid transition through small cerebral arterioles, ⁶ bubbles can cause a progressive decline in the cerebral blood flow even if they do not completely occlude vessels. ^117,121 In addition, bubble contact can affect the vascular endothelium ^116–118 resulting in activation of the coagulation system, platelet aggregation and a thrombo-inflammatory response. ^119,120

Bolus entry of gas emboli

Bolus entry of gas is more likely to cause clinical abnormalities compared with a slow release. ¹⁶⁹ Rapid entry of air (>0.3 mL/kg/minute into the systemic venous circulation can result in significant inflammatory changes, ¹⁷⁰ a rise in pulmonary artery pressures and right ventricular outflow obstruction. ^120,171,172 Manual compression of the saphenous junctions at the time of foam sclerotherapy has been shown to release boluses of foam into the central venous circulation when the pressure is released. ¹⁵¹ This manoeuvre was reported in two published cases of stroke after sclerotherapy. ^5,9 Immediate sitting up or standing following the procedure may also encourage the bolus entry of bubbles into the systemic circulation.

Foam stability

It is unknown whether a more stable foam is safer or whether the increased stability may lead to a higher incidence of complications such as deep vein occlusion or other unwanted distant adverse events. The physical properties of foam may influence the complication profile but this has not been systematically studied. There are many factors that influence the physical properties of foam and in particular its wetting half-life and stability. ^{81,107,173–175} The resorption time of bubbles depends on multiple factors, including the bubble shape, diffusivity, solubility, gas partial pressure gradients and vessel diameter. In turn, these depend on the gas–liquid surface tension, gas pressures in the bubble and gas pressures in the surrounding tissue. ^176–178

It is unknown whether foam-derived bubbles reaching the brain maintain a thin film of detergent at the blood-gas interface (which would reduce the surface tension) or whether such bubbles are devoid of a surfactant surface. The direct effect of surfactant would be to reduce diffusion across the gas-liquid interface. ^179,180 However, bubbles with a surfactant film exhibit greater mechanical compliance and are more likely to separate or elongate under the influence of external forces. ^177,179 Such changes would increase the bubble's surface area resulting in an increase in gas diffusion ^177,179 and a faster resorption time. ¹⁷⁹

Other technical variables

Other variables include the type of syringe, three-way tap and filters used to generate the sclerosant foam. ¹⁷⁵ Silicon in syringes acts as an antifoam and reduces the half-life of the generated foam. Certain three-way taps are designed to separate bubbles from the liquid as against those that increase the mixing. Filters would affect the size of the generated bubbles. All these factors and the liquid:gas ratios affect the physical properties of the generated foam.

Stroke and endovenous laser ablation

Tongues of thrombus have been observed following endovenous laser ablation (EVLA) of the GSV extending into the common femoral vein. ^181,182 A recent case report described stroke following EVLA (980 nm) ¹⁸³ and in another case, EVLA (1470 nm) was performed concurrently with foam sclerotherapy. ¹⁰ Loose thrombi released from the SFJ were thought to be responsible for PCE in these cases. Laser energy from EVLA can result in cavitation and the generated steam bubbles are visualized on ultrasound to enter the cardiac chambers during or soon after the procedure (N Morrison, personal communication via e-mail to K Parsi [kparsi@ozemail.com.au] on 27 July 2010). It is yet to be shown whether such bubbles can act as a source for PGE and a subsequent cerebrovascular event.

Stroke and venous surgery

Varicose vein surgery including stripping and avulsion is associated with a 5% frequency of DVT after the unilateral and up to 15% after the bilateral procedure. ^184,185 Varicose vein surgery has been associated with PCE leading to stroke ^186,187 and TIA. ¹⁸⁷ A rare case of severe paradoxical gas (nitrous oxide) embolism has been reported during varicose vein stripping using a nitrous oxide cryoprobe. ¹⁸⁸ A severe neurological events was recently reported to occur during AP. ¹⁸⁹ In another case of stroke, AP was performed at the same time as sclerotherapy. ¹⁰

Prevention

Neurological ischaemic events following sclerotherapy are possibly multifactorial. It is clear from the evidence reviewed here that variable factors such as treatment parameters and individual's own risk factors at the time of the procedure possibly play an important role in the pathogenesis of the adverse events.

The presence of a RLS and in particular a PFO is the most consistent risk factor in this group of patients. Despite the risk, pre-sclerotherapy screening for PFO has not been considered practical or mandatory. ^148,190,191 Preoperative screening and percutaneous closure of PFOs remain a controversial issue. Despite the 30% prevalence of PFO in the general population, the overall incidence of significant neurological complications of sclerotherapy remains very low. PFO closure procedures may be indicated in high-risk patients and in particular those with a past history of cryptogenic stroke or a long life history of recurrent classic migraine attacks (with aura). However, closure procedures are not risk-free and the long-term benefits of PFO closure in sclerotherapy patients is not known. Furthermore, the number of PFO closure procedures that need to be done to prevent a single major neurological event in sclerotherapy patients in unknown. The role of preoperative screening and percutaneous closure of PFOs needs to be incorporated in future randomized studies.

From a practical point of view, Valsalva and other manoeuvres that would open a PFO should be avoided during and immediately after the procedure. Hence, compression stockings are best applied by the medical staff rather than the patient at the end of the procedure.

The morbidity and mortality of gas embolism is related to the volume of gas, rate of entrainment, the patient's underlying cardiorespiratory status, and the patient's position. To reduce the embolic gas volume, it is essential to limit the entry of the injected sclerosant into the deep venous system and non-target veins. Although it may be argued that the injected foam is tracked and followed by ultrasound in the target vein, which may justify the infusion of large volumes, the entry of the injected foam into the perforators and the deep venous system along the way can be missed on ultrasound. Hence, it is highly prudent to limit the volume of the injected sclerosant, per injection site, to an absolute minimum. ¹⁶⁵ Compression of the SFJ and its subsequent release after the injection may lead to the bolus entry of foam into the systemic venous circulation and is possibly best avoided. ¹⁵¹ Catheterizing the target vein combined with perivenous tumescent anaesthesia may help to reduce the volume required to achieve vessel closure and hence reduce the bubble load entering the deep venous system. ¹⁶⁷ Leg elevation and other reported manoeuvres are ineffective in prevention of VGE ^109,151 and are best avoided. Finally, CO₂ as a physiological foaming gas may be safer than room air and should be further assessed in future studies.

Stroke with an immediate onset

An immediate onset (within minutes) of stroke after foam sclerotherapy should raise the possibility of gas embolism. VGE presents with dyspnoea and continuous cough, hypotension and dizziness, ‘gasp’ reflex and substernal chest pain. A ‘mill wheel’ murmur may be produced by movement of bubbles in the right ventricle.

Coronary artery air embolism can present with chest tightness and pain, coronary artery spasm, ischaemia, arrhythmias and myocardial infarction. Cerebral gas embolism can present with confusion, focal neurological symptoms and stroke. ^{120,170,192–194}

Patients with symptomatic VGE should be placed immediately in the left lateral decubitus position to reduce entry into the pulmonary arteries and a possible subsequent right ventricular outflow obstruction. ¹⁰⁶ In one reported PSS case, the patient was in the left lateral decubitus position during sclerotherapy but developed neurological symptoms when turned over to the supine position with the leg elevated. ¹⁰

In suspected PGE, head-down position is recommended for up to 10 minutes to clear bubbles from the cerebral circulation. ¹⁰⁶ Head-down position for more than 10 minutes can worsen cerebral oedema and the patient should be returned to a supine position. ¹⁹⁵ 100% Oxygen should be administered. ¹⁹⁶

PGE should be confirmed by imaging of bubbles in the intracranial and brain-supplying arterial circulation. Timing of imaging is crucial and a delay in neuroimaging may result in the resorption of embolic bubbles and a subsequent negative finding.

Hyperbaric oxygen has been recommended in the immediate treatment of gas embolism to enhance the diffusion of nitrogen into blood, compression of existing bubbles, improving the oxygenation of ischaemic tissues and lowering the intracranial pressure. ^6,110,197 Some studies, however, have found hyperbaric treatment not to influence the clinical outcome and hence its routine use has not been universally advocated. ^198–200

There are limited data available regarding the aetiology of PSS cases with an immediate onset after liquid sclerotherapy. In such cases, the management should follow the standard stroke guidelines and selected patients may benefit from thrombolytic therapy. ²⁰⁰ The European Cooperative Acute Stroke Study III, the American Heart Association and the American Stroke Association have found thrombolytic therapy to be efficacious in improving neurologic outcomes, suggesting a treatment window of 3–4.5 hours for the administration of thrombolytic agents such as recombinant t-PA. ²⁰¹

Stroke with a delayed onset

In PSS cases with a delayed onset, thorough duplex ultrasound imaging of lower limb veins should be performed to look for a thrombotic source. Particular attention should be paid to avoid failure to detect venous thrombosis on ultrasound (see section Failure to detect venous thrombosis on ultrasound in sclerotherapy patients). The management of stroke should follow the standard stroke guidelines. ²⁰⁰