Complications in pediatric interventional neuroradiology management. Reflections on a personal experience

Abstract

Pediatric interventional neuroradiology is a rewarding but demanding specialty. It cannot be classified as a miniaturized technical procedure simply because it deals with small vessels in small patients but remains a field requiring specific knowledge combining clinical, anatomical, physiologic, and technical features. Complications that occur remain terrible experiences in this population whose patients have not personally requested to be treated for their vascular malformations. The aim of this article is to evoke the various possible groups of complications that may occur during a pediatric endovascular procedure in order to make physicians aware of these problems and help them to avoid them.

Introduction

The famous Norwegian artist Edvard Munch painted five different versions of his well-known picture “Skrik” (“The scream”) representing a man suffering from existential anxiety and screaming for help.¹ A similar kind of feeling is close to what I felt when I had to face either a severe complication or the death of a patient related to an endovascular procedure.

It is indeed always difficult to experience a complication, especially in the pediatric population where it may occur in neonates and infants who have not personally asked to be treated. The interventional neuroradiologist responsible for the morbidity feels guilty because he has created the problem. The parents feel guilty for having made the decision to allow the treatment of their child. The team that takes the child in charge in the angiosuite is often devastated because of the occurrence of the complication. The psychological shock that such a complication represents in children has sometimes brought doctors, radiographers, and nurses to abandon interventional neuroradiology (INR) and turn toward less stressful specialties.

After having lived through the complication, it is even more difficult to talk about it to the parents, and it will remain painful and scary for the physician to confront it and to describe it to colleagues. As for all kinds of morbidities created by endovascular procedures, the recognition of the moment at which the intervention has got into a skid is the first necessary step in the analysis of the complication. It remains thus, even for well-trained specialists, an indispensable personal exercise if one wants to avoid further problems or repeat the mistake that might have been at origin of it.

The aim of this paper is not to review extensively the various types of complication and rates reported in the literature nor to make a meta-analysis of the papers witnessing the problems seen in each center practicing INR. Here, I wanted here to share my reflection about a global “classification” of the complications we have been confronted with or have faced during neuro-endovascular procedures in the pediatric population.

Methods

The local institutional ethics committees approved this retrospective review and waived informed consent, given the study design. The current study was performed under the ethical standards of the 1964 Declaration of Helsinki and its later amendments.

From our database comprising 2167 patients managed between March 2002 and March 2020, we extracted 232 children that I was asked to take in charge by embolization (4 before 2008 and 228 since then); 549 endovascular sessions have been performed (2.4 sessions/patient as a mean number). We have reviewed all types of vascular malformations encountered in this population at various age groups (from neonates to children up to 15 years): vein of Galen malformations (both mural and choroidal), brain AVMs, dural shunts (in dural sinus malformations, juvenile-type or adult-type lesions), spinal cord AVMs, maxillofacial vascular lesions, or tumors.

Results

The global rate of complications encountered was 5.3% among the number of procedures undertaken (both neurological transitory and permanent), of which fatal outcomes as a direct consequence of the procedure or related to decisions taken occurred in 3% (mostly in in neonates or infants with high flow intracranial shunts).

After having thought about all types of complications that I have had to face in my pediatric INR practice since 1990 (first with late Professor Lasjaunias, and then on my own since 2008) and having reviewed their types, I have been able to separate them into five different categories illustrating mistakes made during endovascular procedures which have all have led to more or less severe impairment in the clinical condition of the child.

Technical complications (examples in Figures 1 to 3)

Figure 1.

Technical complication. A neonate with vein of galen AVM. Heart failure not responding to medical treatment in ICU. Decision to embolize in order to stabilize the hemodynamic situation. Angiography shows multiple mural fistulas vascularized by right and left choroidal arteries ((a) left vertebral artery, AP view), draining in the medial vein of the prosencephalon, the falcine sinus, torcular, transverse, sigmoid and occipital sinuses ((b) venous phase of the left vertebral artery injection, AP view). Two superselective catheterizations of the main feeders were achieved ((c and d): AP views), and pure glue was injected rapidly into the shunt after decrease of the systolic blood pressure by the anesthesiologist. Despite positions of the catheter tip ((c and d): arrow) that were considered appropriate at that time (but retrospectively too close to the shunt considering the flow velocity), the glue flowed away too far into the lesion, reached the lungs, and polymerized in the venous outlets of the malformation ((e,f) AP views). The child decompensated hemodynamically on the angio table and passed away.

Figure 2.

Technical complication. A thirteen-month-old female infant. HHT. Cerebral hemorrhage with right hemiparesis. Angiography of the left internal carotid ((a) lateral view) confirmed the diagnosis of MAVF vascularized by frontal branches of both MCA ((a) arrow) and ACA ((a) double arrow) and draining into a giant venous ectasia ((a) asterisk). Superselective catheterization of the main feeders of the MAVF was achieved (b), and embolization was performed with pure glue after decrease of the systolic blood pressure. The glue polymerized too proximally and failed to reach the AV communication (c). Immediate control angiography ((d) left internal carotid artery, lateral view) showed what seemed to be a correct occlusion of the shunt, but a further control in the same vessel at three months ((e), lateral view) diagnosed angiogenesis in the pathological area related to sprouting phenomena due to the proximal vascular occlusion in the pathological area.

Figure 3.

Technical complication. An 11-year-old boy. Progressive left hemiparesis related to a right thalamo lenticulo capsular nidus-type AVM well seen on axial MR (a) and confirmed by left vertebral artery (not shown) and right internal carotid artery angiography (b, lateral view). Several sessions of embolization of the main vascular compartments of the AVM allowed improvement of the child’s clinical situation. During one of the sessions, distal catheterization of the feeder was difficult to achieve because of the vessel tortuosity and size. While a micro guide was used to make the catheter progress, perforation of the vessel occurred with extravasation of contrast (c). The vascular rupture point was immediately sealed with glue to stop the hemorrhage (d). The accident remained clinically silent in this case.

These problems are perhaps the most frequent and easily recognizable ones. They happen principally when the endovascular procedure material (introducers, catheters, guides, microcatheters, embolic materials) has been wrongly used, but also as consequences of contrast overdoses or heavy radiation exposures.

Strategic complications (examples in Figures 4 and 5)

Figure 4.

Strategic complication. Male infant with macrocrania that progressed over seven months before the diagnosis of dural sinus malformation was made on MR (a) and angioMR (b) in the child’s home country. The referring physicians decided to wait for treatment, although the parents regularly alerted about their child’s clinical worsening. At 18 months of age, the boy became lethargic and tetraparetic and was sent to us. A bruit was heard at auscultation of the skull on admission. Occipital artery angiography ((c) arrow; the two upper pictures are AP views, the two lower ones are lateral views, arterial and venous phases) confirmed a dural sinus malformation affecting the right sigmoid and transverse sinus with a single hole shunt vascularized by the mastoid artery ((c) double arrows). Because of non-patency of the right sigmoid sinus, and tight stenosis of the left sigmoid sinus ((c) asterisk), the high flow of the fistula refluxed into the posterior fossa and supratentorial veins. The shunt was cured by embolization with pure glue (d), which was confirmed by control angiography performed five months later (e). Because the infant was treated too late with no respect of the optimal therapeutic window, the hydrovenous disorder induced by the reflux into cortical veins was responsible for severe chronic venous damage well diagnosed on control MR and CT (f) that showed cortical atrophy and subcortical calcifications. The child remained with a severe neurocognitive delay that could have been avoided if early diagnosis and management had been made. We call this a strategic complication because the disease and its consequences in an infant had been ignored, which has led to error in the management strategy with dramatic consequences. Failure to obtain normal neurocognitive development in a treatable lesion but poorly taken in charge should thus be included in such a complication group.

Figure 5.

Strategic complication. A five-month-old male infant is presented with grand mal seizures and progressive left hemiparesis. MR ((a) sagittal view) and CT ((b) axial view) was diagnosed with a dural sinus malformation of the superior sagittal sinus ((a and b) asterisk) with multiple shunts vascularized by middle meningeal artery branches ((c) lateral view, small arrows) and that refluxed ((d) venous phase of the internal maxillary injection, lateral view) in the straight sinus ((d) arrow) and the right basal vein ((d) triple arrows) because of tight stenosis of both sigmoid sinuses. Embolization was performed during three endovascular sessions with correct penetration of glue in the shunt (e) but failed to suppress the reflux into the deep venous system ((f) control angiogram of the internal maxillary artery, lateral view, arrows). The infant bled in the right internal temporal area between two sessions of embolization. In order to suppress the deep reflux permanently, the straight sinus was then occluded with coils ((g) small arrows) and glue ((g) asterisk). The malformation was finally cured (h) but the too long-lasting reflux had created severe brain damage with calcifications ((i) non-enhanced CT, axial view) related to chronic venous ischemia. MR confirmed the damages with cortico-subcortical atrophies and hemorrhagic scars ((j) axial view). The child remained severely encephalopathic. Based on this bad experience, we nowadays occlude the straight sinus reflux associated to dural sinus malformation rapidly with coils to control that venous ischemia: all children managed that way have since had a favorable evolution.

In these situations, the physician has misdiagnosed the disease and has misevaluated the risks involved in the natural history of the malformation. He has thus missed the precise moment at which the child (usually a neonate or an infant) should have been treated (what Pierre Lasjaunias called the “optimal therapeutic window”) and has facilitated the progression of the hydrovenous consequences related to the malformation with brain damage and failure to obtain a normal maturation process.

Analytical complications (example in Figure 6)

Figure 6.

Analytical complication. Choroidal type of vein of galen aneurysmal malformation diagnosed by MRI ((a) (sagittal view), (b) (axial view)) in a 13-month-old female infant with macrocrania, facial venous collateral circulation, and strabismus. Right internal carotid ((c) AP view) and left vertebral artery ((d) AP view) angiographies confirmed that the lesion was vascularized by multiple choroidal and circumferential arteries. Two compartments were first embolized with glue (e,f); the child tolerated the procedure well and was scheduled for another session three months later in order to treat what was at that time considered to be a high flow mural shunt ((b,d) large arrow) vascularized by a large left choroidal artery ((d) arrowhead). A left vertebral artery injection ((g) AP view) at that time confirmed stable disconnection of the previously treated compartments and the supposed main remaining mural shunt ((g) asterisk) vascularized by an enlarged left choroidal artery ((g) low arrow). The vessel was catheterized, and the presumed shunt was embolized with pure glue (i). Immediate control angiograms in the right internal carotid ((j) AP view) and left vertebral arteries ((k) AP view; (l) venous phase of the latter) showed what was considered to be cure of the lesion with stagnation in the stumps of the arterial feeders ((l) arrows). The child bled severely in the brain stem some hours after the procedure. Extension of the bleed into the mesencephalon, the cerebellum, the ambient cistern, and the subarachnoid spaces was secondarily confirmed by MR in sagittal (m) and axial planes (n). Retrospective analysis of the procedure showed that in fact the last embolization had occluded the venous choroidal outlet of a fistula more proximally located on the vessel at the point where the change of caliber between artery and vein was present ((b, h) (magnified view of the vertebral artery injection on (g)) long arrow). Embolization had thus occluded only the draining vein of a shunt vascularized rather by a left short circumferential artery than by a choroidal one and also by right homologous long ones crossing the midline above the quadrigeminal plate and ending into the left choroidal fistula. The stagnations seen on the venous phase corresponded thus to arterial stumps of feeders ((l) arrows) and to the stump of the left choroidal vein occluded distally ((k, l) asterisk). This complication lets us confirm that the choroidal vein of Galen malformations can either be constituted by arteries either ending directly or through arterio-arterial networks in the venous ectasia or draining first into choroidal veins that secondarily reach the medial vein of the prosencephalon. Distinction between these two architectures is mandatory in order to avoid such complications.

The interventional neuroradiologist has made a mistake in the analysis of the angioarchitecture of the malformation. The therapeutic decisions taken during the endovascular management were therefore erroneous.

Proper training in INR including neurovascular anatomy, pathophysiology, natural history of diseases treated, and overall neuroscientific knowledge is necessary to avoid both strategical and analytical complications.

Thrombotic complications (example in Figure 7)

Figure 7.

Thrombotic complication. A six-year-old boy with evolving macrocrania, facial venous collateral circulation, and neurocognitive delay. At that age, late diagnosis by MRI (a) of a vein of Galen arteriovenous malformation. At clinical examination of the skin (b), presence of multiple capillary malformations is related to RASA 1 mutation. Right internal carotid angiography (c, lateral view) confirmed a mural type of vein of Galen aneurysm malformation (arrowhead) draining into falcine sinuses ((c) arrows). (d) Venous phase of right internal carotid injection: because of tight stenosis of both sigmoid sinuses (asterisk) and high flow of the shunt, reflux into the superior sagittal sinus (large arrow), cortical veins (double arrows), cavernous sinus (arrowhead), and superior ophthalmic vein (open arrow) was seen. Because of the size of the choroidal feeder and the flow velocity, it was decided to occlude the shunt with coils; the procedure cured the malformation (e). Control of the late phase of the right internal carotid angiogram showed restauration of a normal venous drainage of the hemisphere (f). The child tolerated the procedure well but presented headaches, vomiting, and apathy 48 h later. Non-enhanced CT diagnosed thrombosis of the medial vein of the prosencephalon extended to its venous outlets and the superior sagittal sinus (g). The child was treated with low molecular heparin for two months and fully recovered from his symptoms induced by the procedure. He is expected to be submitted to MR and angiographic controls.

The clinical problems related to post embolization thrombosis of the vascular malformation in total or in part, and of its outlets, should be anticipated or recognized rapidly after they occur, as a well-conducted anticoagulation treatment can either avoid them or cure them.

Emotional complications (example in Figure 8)

Figure 8.

Emotional complication. A four-year-old girl with macrocrania, progressive loss of vision, headaches, and evolving forehead AVM (a, b). She had been diagnosed and partially treated with coils in her home country for a juvenile dural arteriovenous fistula with multiple shunts on the skull base ((c) right internal carotid artery, AP view, arrow) and the superior sagittal sinus ((g) left vertebral artery, lateral view, arrow). She was secondarily referred to us for further management: angiograms confirmed that multiple lesions had developed over time at the skull base ((d) right internal carotid artery, lateral view; (f) left internal maxillary artery, AP view), the right transverse sinus (e), the superior sagittal sinus and the scalp (f). The child had also developed parietal pial arteriovenous shunts ((d) right internal carotid artery, AP view, asterisk; (h) left vertebral artery, lateral view, asterisk; (l) right internal carotid artery, lateral view) vascularized by branches of the middle, anterior and posterior cerebral arteries. Embolization of the main dural skull base shunts was performed with glue with good penetration of the acrylic material into the venous drainage of the fistula (j); the child tolerated each procedure well and remained clinically stable. However, between two sessions, the parietal shunt bled ((i’) coronal MR) was also embolized (k). Immediate control internal carotid angiogram (l) showed good disconnection of the dural and pial shunts, and the procedure should have been reasonably stopped at that moment. However, as the maxillofacial AVM had always been a huge concern to the child’s relatives, and as I was still shaken by the child’s condition, I continued the procedure against all odds and treated this latter lesion by direct puncture and injection of glue into the scalp vein draining the whole malformation. Despite manual compression of the venous outlet, some glue was washed out by the flow; a firmer compression stopped the flow but created a rotation of the head and my fingers were suddenly in the X-ray field (m). As a reflex, I withdrew my hand. The glue flowed massively into the venous system, reached the lungs (n), and the child decompensated hemodynamically on the angio table. She passed away despite all possible efforts of resuscitation.

These constitute certainly the most abstract type of complication that can occur. They are originally exclusively related to the behavior of the interventional neuroradiologist who, because of erroneous personal psychological involvement in the pediatric case he is managing, loses objectivity and critical thinking, thus making bad therapeutic decisions leading to problems or catastrophic consequences.

Half of all these complications were related to technical mistakes (mainly arterial or venous erratic emboli and rupture of small distal vessels during catheterization), while thrombotic problems counted for about a quarter of complications.

Discussion

It can be tempting to consider that a complication related to endovascular INR treatment is just the consequence of a combination of the aggressiveness of the disease and the weakness of the pediatric host. The physician may then consider that he has done everything possible to save the patient’s life and may not feel directly responsible, and certainly not guilty, for what has occurred.

Precise review of the various steps of the procedure identifies however a moment where either a wrong decision was made, an inappropriate endovascular gesture has been used, or a lack of attention has occurred. In most cases, the complication is therefore related to a human error and rare are the cases where no explanation can be found.

Technical complications

An endovascular procedure begins with the puncture of the femoral artery at the groin and ends with the compression of that puncture point. Technical complications can occur at each step of the intervention, and the interventionist’s concentration should thus be kept maximized during each of these steps in order to avoid them. Technical complications comprise indeed various shades of problems from incorrect use of the materials needed for embolization^2–13 to those related to stronger challenges represented by proper in situ deposition of liquid embolic materials.

Mistakes in punctures or compressions can lead to groin hematomas, but also to femoral artery thrombosis in babies with loss of access for future interventions and limb ischemia. The latter has not occurred so far in our practice. It is also recommended in neonates and infants who need multiple sessions to puncture alternatively both femoral arteries with small size short introducer sheaths in order to avoid such problems.

In high flow fistulas like pial macrofistulas seen in HHT or vein of Galen arteriovenous malformations, if glue is to be used to occlude the shunt, the use of pure glue is necessary, often helped by lowering the systolic blood pressure. Additionally, it is useful to choose a large lumen microcatheter (1.8 F if possible) allowing delivery of a large amount of glue and also to place the catheter tip in a position appropriate for fast injection of glue: at a certain distance from the shunt, close to the vessel wall and in front of a vascular loop, and never facing the arteriovenous shunt frontally in order to avoid distal migration of glue in the shunt and in the draining vein with impairment of the venous drainage of the brain (Figure 1), pulmonary emboli, or hemorrhage. A proper positioning of the catheter allows indeed the glue to first stick to the arterial walls before penetrating the fistula with a “mushroom” aspect. Using pure glue to occlude such high flow shunts carries a theoretical risk of gluing the catheter: such a problem occurred in my practice for the last time nine years ago. This was related to my misinterpretation of the flow velocity in a dural sinus malformation in an infant. It has not occurred since. Careful attention to correct use of embolic materials with ongoing attention toward each step of the procedure following strict rules of delivery of embolic agent is mandatory to avoid this type of problem.

I have mainly used glue in endovascular treatments of pediatric vascular malformations, and mistakes in the appreciation of the correct mixture of glue and lipiodol have led to incorrect deposition of the embolic material in the malformation. Too diluted glue, or distal malposition of the microcatheter, can create the above-mentioned problems, but too concentrated glue or proximal malposition of the microcatheter can be responsible for proximal vessel occlusion and secondary angiogenesis. Enlargement or development of collaterals reconstructing the shunt is very often seen then; sprouting angiogenesis with de novo vessels are rarer and might depend on the genetic background of the patient and the biological activity of the vascular malformation (Figure 2). Both make the angioarchitecture more complex and may impede proper transarterial access to the shunt itself.

Catheters and guides can create vascular dissections or ruptures (Figure 3). Erratic emboli in normal arteries in the vicinity may also occur as they have also been described in some reports.⁹ Any liquid embolus can create such problems, and onyx has been shown to be responsible for similar complications.^{4–6,9,10,12}

Other complications that can be linked to this technical group include those related to the use of excessive amounts of contrast or high doses of radiation. We use a dose of 6 cc/kg of contrast as a limit in neonates, infants, or young children; if this is not as strict in adolescents, one should always keep in mind not only the potential toxicity of the contrast on a suffering brain but also on other organs as kidneys or lungs. Medical managements (sometimes in the ICU if babies were initially intubated and ventilated) may be necessary to compensate these contrast overdoses (less than 1% of our cases).

We strive to keep the procedural fluoroscopy time in babies at a maximum of 1 h, and fluoroscopy and runs should be restricted to the minimum effective ones to obtain all information needed. Radiation-induced complications have not been seen so far in this experience.

Strategic complications are often related to misdiagnosis, misunderstanding, or ignorance of the natural history of the disease:¹⁴ the therapeutic strategy is thus erroneous, the disease worsens, and the right treatment occurs too late (Figure 4). It is therefore important to diagnose the lesion early even when faced with a “minor” symptom such as macrocrania. Enlargement of the head circumference in an infant may indeed be a sign of the onset of hydrovenous disorders and should always raise the question of an underlying vascular disease. The venous hyper-pressure induced by the shunt influences both venous drainages and CSF resorption and accumulation of fluids in the white matter is responsible for the macrocrania, and later for subependymal atrophies and ventriculomegaly, or hydrocephalus. These problems illustrate the role played by subpial and medullary veins in the maintenance and development of the cerebral white matter: the subpial veins communicate with the subependymal veins through medullary veins and may therefore interfere with the CSF equilibrium.^14,15 A systolo-diastolic bruit heard at auscultation of the head in a macrocephalic child testifies to the existence of an intracranial vascular problem; CT or better MR should then be rapidly performed in order to confirm the diagnosis. According to the type of lesion, early or delayed treatment will be proposed. Not recognizing the lesion properly and early and not managing it in due time may lead to severe neurocognitive problems. The failure to obtain a normal maturation process can also be considered as a therapeutic failure or complication if the optimum moment for intervention is missed or overlooked.¹⁴ Furthermore, if during staged embolization of a vascular malformation in the early pediatric population (neonates or infants) occlusion of certain compartments reroutes the flow toward superficial or deep veins, one should definitely control this abnormal flow as much as possible to avoid focal or general hydrovenous disorders that can lead to catastrophes such as melting brain syndrome.¹⁶ The case illustrated in Figure 5 shows this situation.

Analytical complications are related to misinterpretation of anatomy,¹⁷ vascular architectonics, or architecture of the lesion.¹⁸ One of the worst examples of this occurred when I had to deal with choroidal types of vein of Galen arteriovenous malformations. It is not always easy to distinguish in the choroid fissure what portions of the, sometimes extensive, network correspond to arteries and which to veins. The medial vein of the prosencephalon initially does not drain the classical deep venous drainage of the brain but only choroidal veins.¹⁴ During its maturation, the deep venous drainage will be captured, and the well-known distribution of the vein of Galen and its tributaries will be constituted. Malformations of the vein of Galen are created during embryonic life up to the eighth week of intrauterine development,¹⁴ and the choroidal networks are constituted anatomically either by choroidal and subependymal arteries draining directly or via arterio-arterial connections into the venous ectasia, or by bunches of same type of arteries draining first into choroidal venous collectors that end secondarily in the medial vein of the prosencephalon. If such a choroidal vein drains several different arterial compartments, early occlusion of the vein may, as in any AVM, carry a high risk of rupture and bleeding. The case described in Figure 6 emphasizes such an architectonic distribution and explains the complication that has occurred linked to misunderstanding and incorrect analysis of the architecture of the malformation. Recognizing the various architectures can be extremely difficult: cautious analysis of pre-embolization selective runs with 3D (or possibly 4D if available?) and cone beam CT reconstructions, correlated to MR images may help to understand the shunt and to occlude it in good conditions.

Thrombotic complications are perhaps the most frequent type of complications that can occur and those that can be best managed medically by anticoagulation. Even if no erratic embolus of glue has occurred, venous or sinus thrombosis may happen through flow modifications after embolization.^18–20 When draining AV shunts, these structures are under high pressure and flow: changes in these hemodynamic conditions can create secondary sludge phenomena, stagnation, and thromboses. A difficult question is to know if patients should be treated by anticoagulation, and when. Anticoagulation has been used in 12, 5% of children in our series most often in a preventive way but also in a curative way. Spinal cord AV shunts, dural AV shunts, vein of Galen malformations, and more rarely brain AVMs have been managed like this. We have preferred to treat these patients on a case-by-case basis and not systematically, basing our arguments on the type of lesion (even if not cured in one session, spinal cord, and dural shunts needed anticoagulation more frequently) associated to a worrying slow flow in the draining veins or sinuses after treatment that could compromise the normal venous drainage. Other teams might prefer to routinely anticoagulate their patients: this might however carry hemorrhagic risks. We have favored careful attention with the neuropediatricians to the patient’s clinical status, especially in neonates and infants, as such, thromboses may first be clinically revealed by a grumpy and painful child difficult to examine neurologically.¹⁴ Not thinking about such potential complications could lead to extension of the thrombotic phenomena and worsening of the child’s status with sometimes a fatal outcome in severe conditions. We consider that as soon as thrombosis is suspected and if the child is clinically impaired, anticoagulation with heparin or low molecular heparin should be begun after CT or MR has proven the thrombosis. It is difficult to say for how long the treatment should last: in our experience, the child is usually treated for two to four weeks before reaching a therapeutic window and control of the thrombosis by MR or CT. According to the findings and the child’s clinical examination, treatment is then either stopped or continued till the situation is properly mastered. The case described in Figure 7 shows such a complication and its management.

What I have called emotional complications are certainly the most difficult ones to understand, describe, or handle. It regroups many factors that include technical, tactical, and analytical data taken together, but, among others, these complications are related to an abnormal personal involvement of the physician in the case he has to treat: it is no longer empathy but loss of objectivity and critical thinking. The poet Walt Whitman has expressed this feeling well by saying “I don’t ask the injured person how she feels, I become myself the injured person”. Carl Rodgers, one of the most important 20th century psychologists, has confirmed this notion by affirming “In empathy with a patient, one can feel his pain or his happiness, without ever forgetting that it is the patient’s experience or perception. If that last condition is absent, then it is no more empathy, but identification”. A clear objective mind that should lead to appropriate therapeutic judgment is therefore lost to the benefit of subjective erroneous choices dictated by the physician’s anxiety: incorrect decisions are taken, which then open the doors to all other types of mistakes (technical, strategical, or analytical). I have faced this situation once, but it has been one of my most terrible experiences as I was the main actor for the disaster that occurred. Details of the error made are described in the case described in Figure 8.

Conclusions

The lessons learned from complications after pediatric INR procedures, whatever the results, are some of the most important ones a physician can experience from personal, psychological, philosophical, medical, and technical points of view. Good results are obtained if careful objective decisions are properly taken at each step of the interventional procedure.^21–23 Beyond visible deficits such as hemiplegia, for example, a failure to obtain normal neurocognitive evolution in neonates or infants can also be considered as a complication of treatment. Analysis of the complications is mandatory, and their understanding avoids reproducing them. It also allows future-improved procedures in both children and adults. Even if several profiles of complications exist, almost all of them have a human origin. Avoiding them requires permanent attention and balance in the decisions taken. This needs ongoing personal improvement in technical and scientific knowledge and requests collaboration between physicians interested in pediatric neurosciences (neuropediatricians, pediatric neurosurgeons, ICU, etc.) in order to allow the interventionalist to gain and improve skills, each actor bringing his special contribution to the understanding of the pathological condition and its management. The use of artificial intelligence²⁴ may be helpful in the future but should not hide the need to familiarize oneself with basic and clinical neurosciences. Pediatric INR as a clinical specialty is thus not only a fight against the disease but also a fight of the interventional neuroradiologist against himself as he must strive to surpass himself constantly. Pediatric INR is therefore also a school for humility.