The history of the operation of surgical anatomical hemispherectomy (aHSPY) is an interesting one. Its initial use consisted of the surgical management of two different clinical conditions: (1) cerebral gliomas and (2) intractable seizures in children with infantile hemiplegia. The rationale for both was the anatomical removal of most of the involved abnormal hemisphere, leaving only the forebrain gray matter. There have been brief reviews of the history of hemispherectomies, the current types, and the reasons for their variations (e.g., Almeida et al. 2006; Binder and Schramm 2006).

The treatment of cerebral gliomas can be found in the literature in a plethora of papers. The use of aHSPY began in the early part of the twentieth century when the then current surgery for glioblastoma multiforme (GBM) was so very disappointing. The notion developed that perhaps the removal of the whole of the diseased hemisphere would be likely to remove all of the associated peripheral parenchyma, which was considered to be the source of the continued growth of the tumor postoperatively. On the basis of the then understanding of the natural history of the GBM, it was initially considered that an aHSPY would cure the tumor. This was attempted by many of the well-known surgeons of the day (Dandy 1928, 1933; L’Hermitte 1928; Zollinger 1935; Bell and Karnosh 1949; Hillier 1954; Gardner 1953; Gardner et al. 1955; Austin and Grant 1955). The view that an anatomical hemispherectomy (aHSPY) might be a curative surgical management of glioblastoma multiforme was settled by the middle part of the century. The postoperative quality of life in these patients was very, very poor, by nearly any standard, but perhaps more importantly, the gliomas were not cured. Thus, the use of the aHSPY simply prolonged very low-quality lives of the patients and hence was discarded. However, the one aspect of the operations, which was of some importance, was that it was clearly evident that patients could survive this radical procedure, certainly more radical than any previous neurological surgery. In any hemispherectomy, it is true that “. . . the more pronounced the neurological defect, the more advantageous will be the operation” (Goodall 1957). There is no argument with the fact that those with an already poor quality of life because of neurological deficits from the GBM, per se, would not suffer as much as those who were neurologically intact preoperatively, but also the fact remained that the GBM was not cured and the quality of life, irrespective of the preoperative state, was far worse postoperatively.

The treatment of intractable seizures of children with infantile hemiplegia and epilepsy was a different story and stood in marked contrast to that for gliomas. The initial attempt at its use in this condition was by Dr. Kenneth McKenzie in Toronto (1938; see also Williams and Scott 1939). Its success led to its widespread use in the ensuing years (Krynauw 1950; Cairns 1951; Polani 1952; McKissock 1953, 1954; Meyers 1958). The most exciting report was that of Krynauw in South Africa. He carried out aHSPYs on 12 patients, between the ages of 8 months and 21 years. They all had the syndrome of infantile hemiplegia, intractable hemispheric seizures, and mental changes. His operation consisted of the anatomical removal of the hemisphere with the exception of “the thalamus and the caudate nucleus and its tail.” There was one immediate postoperative death. The operation, beginning with McKenzie, was often called “hemidecortication” (Meyers 1958; Carmichael 1966), which in some ways is a more accurate anatomical description of the procedure. The pathology of the removed hemispheric components was a potpourri of abnormalities with no single pervasive morphological identified abnormality. Krynauw’s patients had transient increased motor impairments. However, eventually the motor abilities were improved, if anything. There was a profound cortical sensory impairment, which never completely resolved. There was a contralateral hemianopia, if it had not existed preoperatively, with macular sparing, which was of no clinical consequence.

The popularity of the operation for epilepsy was really attributed to the outcomes of the patients. In all the patients, the seizures were abolished, without the use of anticonvulsant medication. In all, there were profound improvements in behavioral abnormalities, e.g., temper tantrums, improved cognitive function, improved motor function, and improvement in what were referred to as “personality” traits. The latter were so striking that there were some who considered that the aHSPY should be conducted for this abnormality alone! Thus, by the middle of the twentieth century, there was general consensus that the operation of aHSPY was the most beneficial operation for epilepsy, when the patients presented with intractable seizures based on cerebral hemiatrophy and infantile hemiplegia. Unlike the poor postoperative cognition of the tumor patients having undergone an anatomic hemispherectomy, those with seizures and infantile hemiplegia were the reverse in that there was nearly always significant cognitive improvement (McFie 1961; Smith and Sugar 1975). This was the consensus by neuropsychologists who commented upon it in the literature.

Most disappointingly, the foregoing excitement about the beneficial outcomes of the operation did not last long, for it was about the same time when there began to appear in the literature the odd article describing peculiar complications in the longer postoperative courses of some patients. The original article by Noetzel in 1940 was followed by other articles, which ushered in the evidence of delayed complications that led to the demise of the operation of anatomical HSPY (McKissock 1953; Cabieses et al. 1957; Laine et al. 1964; Oppenheimer and Griffith 1966; Griffith 1967; Falconer and Wilson 1969; Wilson 1970a, b).

The initially described complications were not uniform, and there was no standard acceptance as to why the complications occurred. However, gradually it became evident that the common complications included subdural appearance of a membrane or more than one membrane, fluid in the resective cavity that was xanthochromic or bloody, granular ependymitis, and often hydrocephalus (Noetzel 1940; Adams 1983). The latter could be managed, but the former failed any reasonable management. It soon became a recognizable hemorrhagic syndrome that eventually was simply designated a “delayed” intracranial hemorrhage. Its actual location was not easily identified nor absolute. It was labeled by some as “subdural” (Oppenheimer and Griffith 1966; Falconer and Wilson 1969; Wilson 1970a, b), and there was nearly always associated evidence of hemosiderin in the brain and spinal cord (Falconer and Wilson 1969; Wilson 1970b). This hemorrhagic complication became associated with a number of names with various designations, but eventually the term “superficial cerebral hemosiderosis” (SCH) probably became the one used most commonly (Rasmussen 1973, 1975, 1983; Rasmussen and Andermann 1983).

No source of bleeding was found to which one might attach a specific physiological etiology nor was there any recognition suggesting a potential prophylaxis or certain satisfactory treatment of the complication. The notion arose that it was the large area left by the hemispherectomy that somehow gave rise to the bleeding, which led to various suggestions as to how the cavity might be reduced in volume. Oppenheimer and Griffith were the first to suggest managements of the complications; these included the disconnection of a part of the hemisphere, but leaving it in place, thus reducing the cavity size (1966), which was echoed by Wilson (1970b). Another suggestion was to simply use a corpus callosotomy as a replacement for hemispherectomy (Wilson et al. 1975). Rasmussen reviewed the MNI results of both aHSPYs as well as multi-lobar resections for the management of intractable epilepsy. Approximately a third of the former developed the SCH syndrome, but none of the latter developed the hemorrhagic complication (1973, 1983). He and his colleagues considered that “this syndrome results after anatomically complete hemispherectomy from lack of adequate support of the remaining hemisphere within the skull and its consequent vulnerability to minimal jolts to the head or brief physiological increases in intracranial pressure (e.g., coughing, sneezing) leading to repeated leakage of red blood cells into the intracranial cavity” (Tinuper et al. 1988).

In order to reduce or abolish the SCH syndrome, Rasmussen, in 1968, began carrying out operations associated with leaving in place (and connected) at the least epileptogenic 25–33 % of the hemisphere in patients with infantile hemiplegia, an operation that he referred to as a “subtotal hemispherectomy.” However, the seizure outcomes in these patients were less efficacious (45 % vs 59 %), but as expected, the SCH syndrome was not observed (Rasmussen 1983). This “cost of a significant lessening of the effectiveness of the cortical excision” in the subtotal hemispherectomies led Rasmussen to further revise the operation by “disconnecting the remaining segment of the bad hemisphere from the rest of the brain by sectioning the white matter down to the medial pial surface,” which involved “the remaining frontal or occipital pole ….. separated from the corpus callosum and the upper brain stem, resulting in a functional complete but anatomical subtotal hemispherectomy” (Rasmussen 1983; Tuniper et al. 1988; Villemure and Rasmussen 1990). The latter, or a modification of it, is what eventually developed and came to be called a “functional hemispherectomy” (fHSPY). This fHSPY was joined by a more recent, less invasive, technique, which has been labeled a “hemispherotomy,” in which nearly the whole of the disconnected hemisphere is left intact (Delalande et al. 1992, 2001). This will be discussed briefly in Sect. 12.3.

Thus, in summary, as clinical experience accumulated, it became evident that this late hemorrhagic complication occurred many years (a minimum of 3–4 years and up to 20+ years) postoperatively after an aHSPY, that it involved approximately 25–33 % of the aHSPYs carried out, and that 50 % of those who developed the complication died. Thus, in the final analysis, one out of every seven or eight cases undergoing an aHSPY died some years later by complications attributable to the operative procedure, which had been conducted many years earlier! As a result, shortly after the middle of the twentieth century, the use of the aHSPY as an operative procedure for the surgical management of epilepsy gradually came to an end.

There have been many examples of fHSPYs with slightly different alterations in methodology. Independently, I have used the same technical fHSPY, as described in this section, since the 1970s (see Girvin and Baessa 2006). It is not dissimilar to the final revision of what was used at the Montreal Neurological Institute by Villemure and Rasmussen and their colleagues (1993). It has been a very satisfactory procedure in my hands, and, as a result, I have thus seen no reason to change it, on the basis of what I interpret to be minor suggestions of alterations in the literature. The reader may be interested in reviewing many of these modifications, which have appeared in the literature since the turn of the century (Schramm 2002; Binder and Schramm 2006; Almeida et al. 2006). I will not comment on these fHSPYs, as I truly believe that it is what a surgeon develops on his/her own, or learned during residency, that best determines the best procedure for him/her. The article of Binder and Schramm outlines the advantages that they feel exist in their “transsylvian functional hemispherectomy” as compared with the Rasmussen version. I could quite honestly make a similar comparison of the advantages of the one I will describe in the following narrative, compared with Binder’s and Schramm’s method, but I do not feel that it is terribly useful. In considering the many different techniques that may emerge from a given surgical operation, one should be aware of Cushing’s very relevant statement in Keen’s Surgery: Its Principles and Practice, “Many of the views to be expressed are purely personal ones and are given with the full knowledge that the instruments and operative details found satisfactory to one individual may be entirely unsuited to the operative requirements of others” (1910, p.261). For those in academic institutions with neurosurgery residency programs, these operations provide significantly rewarding learning experiences for residents!

My fHSPY consists of the removal of the temporal lobe and the lower part of the Rolandic cortex, but leaving the remaining disconnected hemisphere (frontal and posterior parietal and occipital lobes and the superior Rolandic cortex) in place, with no mortality, a paucity of morbidity, and nearly always with the abolition of seizures or greater than a 90 % reduction in seizures. A significant portion of the patients is off anticonvulsant medication and free of seizures. It is a safe procedure and easily performed. I have arbitrarily divided the technical procedure into a continuum of five relatively individually isolated surgical steps, which I have labeled: (1) radical temporal lobectomy (TLY), (2) central (Rolandic) corticectomy, (3) corpus callosotomy, (4) posterior medial and orbitofrontal incision, and (5) antero-posteromedial parietal incision.

The indications for an fHSPY are varied. That for which it was originally indicated in the management of intractable epilepsy related particularly to infantile hemiplegia and it remains the most appropriate procedure for treatment of this condition and one of the most efficacious surgical managements of intractable epilepsy. However, it also has become a very satisfactory operation for certain patients with Rasmussen’s encephalitis, hemi-megalencephaly, and a variety of other less common diseases, e.g., some cases of Sturge-Weber disease, porencephalic cysts, etc. King et al. were the first to consider the use of hemispherectomy in the surgical management of hemimegalencephaly (1985). The reader who is interested in the complexities of hemimegalencephaly is referred to more comprehensive reviews in the literature (Dekaban and Sakuragawa 1977; DeMyer 1999; Di Rocco and Iannelli 2000).

One feature of the surgical conduct of either type of hemispherectomy is the difference between the cases with large ventricles as compared to those in which ventriculomegaly is not present. This difference is not minor by any means. The surgery of the latter takes much longer. In those with large ventricles (infantile hemiplegia with typical hemispheric atrophy, some porencephalic cases, and occasional Rasmussen’s encephalitis), the whole operation can be easily carried out within 2–3 h. However, those with small ventricles, or anatomically distorted small ventricles, which may especially be seen in hemimegalencephaly with significantly distorted anatomy, can take up to 5–6 h or even more infrequently. In these latter cases it takes more time to incise the much thicker mantles of parenchyma, more time to sort out distorted anatomy, more time to easily find and follow the small ventricles, more time to achieve satisfactory hemostasis, and more time to retract parenchyma. Any one of these potential eventualities can head the list of added labor-intensive features in the conduct of fHSPYs in which there are small and distorted ventricles and/or postoperative complications, especially in infants below the age of 9 months (see DiRocco and Iannelli 2000).

Another aspect of the fHSPY, which is different from general neurosurgery, is that the vascular supply to the disconnected parts of the hemisphere that are being left in place, i.e., the posterior parietal, occipital, and frontal lobes, should be preserved. This is another aspect of the fHSPY, which is more tedious in the group of cases with small ventricles and larger volumes of parenchyma. Adams indicated the importance of good hemostasis in this operative procedure, noting that it should be “immaculate” (1983)!

The preferred scalp incision is not dissimilar to that used for the aTLY, as noted and illustrated in Chap. 6 (Sect. 6.​2.​2). It does not need to be any larger than the interrupted line in Fig. 12.1a that is the same as that for an aTLY demonstrated in Fig. 6.​1, if the operation is for disorders in which there is significant hydrocephalus, e.g., infantile hemiplegia. However, if it is for one of the conditions with small ventricles and large amounts of parenchyma, then it needs to be somewhat more posterior and superior, reaching approximately the temporo-occipital junction in the former and within 2–3 cm of the midline in the latter, as illustrated by the solid line in Fig. 12.1a. This is more easily understood by referring to Figs. 12.1b, c. B illustrates the case of the very large ventricles, which characterize the cases of infantile hemiplegia. The lateral ventricle is not only extremely enlarged, but the anterior extent of the large trigone is more anteriorly situated because of the atrophy of the parenchyma of the temporal lobe, the insula, and the forebrain nuclei, i.e., thalamus and hypothalamus. C illustrates the usual (normal) position of the trigone as a result of the normal or above normal amount of parenchyma, both cortical and in the gray matter nuclei of the forebrain. Thus, the scalp incision must be carried more posteriorly and medially in order to easily achieve the second (see Sect. 12.2.5) and third (see Sect. 12.2.6) steps, respectively, of the fHSPY described in the following sections.

This is a radical extension of the aTLY in that it is not “anterior” to any extent. It is a true radical complete temporal lobectomy (TLY), extending back to the trigone of the lateral ventricle, e.g., near the junction of the temporal and occipital lobes. Its surgical technique is the same as that outlined earlier for the aTLY (Sect. 6.​2). The major difference is its extent, with the posterior resection (PRL) line being at the coronal cross section of the posterior trigone of the lateral ventricle; thus, its posterosuperior extent will be very close to, at, or even behind the posterior extent of the Sylvian fissure. This is illustrated in Fig. 12.2a. This is the initial stage of the fHSPY, and the very posterior removal greatly facilitates the other two steps that isolate the posterior part of the hemisphere, which will end up being disconnected from any remaining functioning cortex. The posterior cerebral artery (PCA) must be preserved, as it will be irrigating the cerebral parenchyma that will be disconnected, but otherwise left intact. This also applies to any significantly sized middle cerebral arterial (MCA) branches that cross the posterior resection line, which take the label of en passage arteries. Other than these points, this step is simply a more radical removal of the temporal lobe cortex than the aTLY, but its surgery is roughly the same. In those with marked atrophy, e.g., infantile hemiplegia, the TLY is much easier and less time consuming than the small ventricle/large amount of parenchyma disorders, e.g., especially megalencephaly. Figure 12.2b illustrates the surgical bed of the temporal lobectomy.

This “central corticectomy” or “Rolandic window” is the second stage of the fHSPY; it provides full access to the lateral ventricle, which is required for step three, which consists of the corpus callosotomy. Figure 12.3a outline its boundaries: (1) inferior, Sylvian fissure; (2) posterior, the extension superiorly of the posterior resection line (PRL) of the TLY; (3) anterior, mid-frontal, dorsolateral incision at roughly a coronal level of the sphenoid wing; and (4) superior, sufficient horizontal cortical and subcortical incision to achieve access to the roof of the body of the lateral ventricle, through which the corpus callosotomy will be conducted.