[1] 7w Factors of morbidity in hemispherectomies: Surgical technique!pathology

[1] 249w Objective: The objective of this paper is to evaluate factors of surgical morbidity from different techniques of hemispherectomy with emphasis on causative pathology. Patients and methods: Thirty patients underwent hemispherectomy in our institution from 1987 to 2003, two presented with Sturge-Weber Syndrome (SWS), sixteen with Rasmussen's Syndrome (RS), eight with established hemispheric lesions (EHL), and four with cortical development malformations (CDM). Six surgeons operated on three patients using anatomical hemispherectomies (AH), 11 patients using functional hemispherectomy (FH), and 16 patients employing hemispherotomy (HT). Surgical technique and causative pathology were studied independently as factors of morbidity in hemispherectomy. Results: Overall mean surgical time was 11:50G3:20 h and increased proportionately in pathologies with larger hemispheres. Blood transfusion was particularly influenced by the approach adopted by our team of anesthesiologists, independently of technique or pathology. Pathology was the most important factor related to hydrocephalus as two out of four patients with CDM needed ventriculoperitoneal shunt whilst none with EHL or SWS. Four patients undergoing HT and one FH presented residual bridges connecting the hemispheres, three were reoperated and are seizure free. Two patients with CDM did not improve their seizures worthwhile with surgery and other two (one with RS and other with CDM) were waiting a second procedure due to incomplete inter-hemispheric disconnection. Five patients presented infection and one died after developing meningoencephalitis. Conclusion: Hemispherectomies are procedures where pathology and surgical technique interact narrowly. Therefore, in order to study surgical morbidity or outcome, both pathology and technique have to be analyzed independently.

[1] 73w Anatomical hemispherectomy (AH) has been used in treating seizure since the 1930s. However, it was almost abandoned in the 1960s after reports of postoperative fatalities caused by hydrocephalus, hemosiderosis, and trivial head traumas [1]. Despite serious complications, patients' remarkable improvement encouraged surgeons to modify the AH in order to lessen its morbidity whilst preserving efficacy. Efforts to improve the technique gave rise to several unique procedures, which remain in use to this day.

[2] 133w Given that all approaches are able to achieve complete isolation of the damaged hemisphere, sufficient to render the patient free of seizures [2], the rationale behind choosing a particular technique should be based on its related morbidity. Unfortunately, comparing the morbidity of different techniques drawing from the literature is not straightforward. Some authors describe blood loss, operative time, and risk of developing either hydrocephalus or hemosiderosis, as drawback parameters related specifically to the technique [3,4]. However, it would be reasonable to assume that other factors, such as the causative pathology, also bias these parameters. Furthermore, most papers give an account of a series of cases operated on by the same individual and as a consequence the experience and ability of the surgeon in question are bound to have a considerable influence on results.

[3] 55w In the literature, pathology is already considered the main determinant for seizure control, where its importance for intra and postoperative course has been cited but as yet not fully addressed. The objective of this paper is to evaluate the surgical morbidity of different techniques of hemispherectomy with emphasis on pathology as a distinct morbidity factor.

[1] 69w Overall mean surgical time was 11:50G3:20 h, varying from 6:30 to 19:00 h. Comparing different pathologies, surgical time increased proportionately with larger hemispheres though figures were not statistically significant. Hemispherectomies on patients with CDM took on average 13:30 h, while on patients with RS or SWS the time stood at 12:00 h on average, whilst patients with EHL took 11:30 h. The technique employed did not affect procedure length.

[2] 113w All but one patient in this series received a blood transfusion during surgery. On average, each patient received 3.3G1.4 units of packed red blood cells. Despite transfusion, hemoglobin levels fell 1.0G1.4 g/dl, comparing pre and immediate postoperative values. Hemoglobin values on immediate postoperative hemogram averaged 12.5G1.5 g/dl. Over the first days after the procedure, levels of hemoglobin continued to drop, by an average of 2.6G2.0 g/dl, and six patients had a second transfusion, usually when their hemoglobin levels fell below 10 g/dl. Hemoglobin reached its lowest values between the second and the fifth postoperative day. There were no statistical differences among groups of pathologies, or surgical techniques in relation to blood transfusion requirement.

[3] 130w Ten patients developed leukocyte counts of over 20,000 cells per mm 3 on the blood sample collected within 12 h after surgery. This group evolved with an average decline in hemoglobin of 3.8G1.7 g/dl over the proceeding day. In contrast, patients with leukocyte counts below 20,000 cells per mm 3 presented an average decline of only 2.0G1.7 g/dl Fig. 1. Schematic drawing of groups of hemispherectomy techniques: Group 1 includes techniques based on complete cortical removal and is divided in two subgroups: 1A (intra ventricular approach) and 1B (extra ventricular approach). Group 2 assembles disconnective techniques and is divided in three subgroups: 2A (functional hemispherectomy), 2B (vertical approach), and 2C (lateral approach). The drawings were based on previously reported techniques that are described in details in the original papers [3,9,25,27,29].

[4] 22w in hemoglobin (P!0.01). Variations in early postoperative leukocyte counts were not linked to causative pathology, technique, intra-operative blood transfusion, or surgical length.

[5] 30w Data from each patient regarding surgical time, age, hemoglobin variation, and the first postoperative white blood count, separated by pathology and surgical technique are shown on Tables 1 and 2.

[6] 36w Three patients developed hydrocephalus three, four, and six months respectively, after the hemispherectomy, and underwent ventriculoperitoneal shunts, one with RS and two with CDM. Two had undergone procedures from Group 2C and one from Group 1A.

[7] 110w After surgery, MRI detected a residual bridge in the corpus callosum connecting the hemispheres in five patients (four with RS and one with CDM, four had undergone FH, Group 2A, and one HT, Group 2C). All these patients developed postoperative seizures and, on closer investigation, scalp EEG disclosed epileptic activity spreading between the hemispheres. Three patients with RS underwent a second operation to complete disconnection of the corpus callosum and became seizure free thereafter. Intense electrical activity was revealed by electrocorticography over the insula in one of these patients where the area was then removed during the operation. The other two patients with incomplete disconnection are awaiting a second procedure.

[8] 92w Twenty-one patients received follow-up over one year, three over less than a year, whilst one deceased, and five were lost during follow-up. A total of 17, out of 21, were seizure free or had a 90%, or greater, improvement in their seizures. Four patients remained with seizures, which recurred few days after the surgery. Two of these were awaiting a second procedure, made necessary due to incomplete disconnection of the corpus callosum (one with RS and other with CDM) and the remaining two did not improve significantly with surgery (both had CDM).

[9] 381w Table 1 Patient's data separated by causative pathology ID Path. Tech. Age Surg. time HB1 HB2 HB3 HB1-HB2 HB2-HB3 BT WBC 1 CDM 2C 2.2 17:00 15.8 12.1 9.7 3.7 2.4 2 15600 2 CDM 2A 13 12:30 15.2 13.6 9.6 1.6 4 3 21600 3 CDM 2C 1.7 18:00 14.9 14.5 8.6* 0.4 5.9 3.2 30000 4 CDM 1A 0.6 6:30 13.4 14.1 10.1 K0.7 4 3.5 14200 5 EHL 2C 16 9:00 14.7 12.1 11.7 2.6 0.4 3 12620 6 EHL 2C 19 13:00 14.2 12.8 9.6 1.4 3.2 2 37700 7 EHL 2C 7 12:00 11.3 13.3 8* K2 5.3 2 26300 8 EHL 2C 33 10:30 16.6 11.9 10.2 4.7 1.7 0 14370 9 EHL 2C 16 12:30 11.6 10.4 10.3 1.2 0.1 5 9800 10 EHL 2C 8 13:30 13.3 10.8 10.2 2.5 0.6 4 17830 11 EHL 2C 18 11:30 14.2 12.5 12 1.7 0.5 4 11760 12 EHL 2C 38 9:30 15.4 14 10.8 1.4 3.2 1 22780 13 RS 2A 15 11:00 13.9 14.7 9.7 K0.8 5 4 12100 14 RS 2C 11 14:00 12.4 11.8 11.5 0.6 0.3 4 15200 15 RS 2C 3 19:30 13.6 12.7 10.1 0.9 2.6 3 12130 16 RS 2C 7 15:30 12.2 12.5 10 K0.3 2.5 4 13200 17 RS 2A 8 8:00 12.9 11.9 10 1 1.9 3 10300 18 RS 2A 14 8:50 15 13 9.7 2 3.3 5 13300 19 RS 1A 21 9:00 15.5 14 8.1* 1.5 5.9 5 23900 20 RS 2A 4.5 13:00 12.9 12.5 10.5 0.4 2 3 22100 21 RS 2A 13.5 14:00 14 13.5 10.5 0.5 3 4 15800 22 RS 2C 16 15:30 13.5 13.6 8.8* K0.1 4.8 5 22400 23 RS 2C 7 15:00 12.4 13.2 9.9 K0.8 3.3 4 23000 24 RS 2A 6.5 13:00 13.5 12.3 11.8 1.2 0.5 3 27600 25 RS 2A 5.2 8:30 12.8 13.9 8.8* K1.1 5.1 2 11600 26 RS 2A 5.5 8:15 12.5 11.9 11.7 0.6 0.2 3,3 16600 27 RS 2A 8 10:30 11.5 10.5 10.2 1 0.2 4 12000 28 RS 2A 21 8:00 14.5 12.5 7.8* 2 4.7 5 13400 29 SWS 2C 4.5 8:00 13.1 11.7 11.5 1.4 0.2 1 13590 30 SWS 1A 0.7 8:00 10 7.5** 12.1 2.5 0 2,9 14100

[10] 99w Legends: ID (patient's identification), Pathol. (causative pathology), Tech. (used surgical technique, specified by groups reported on Fig. 1), Age (age at surgery in years), Surg. Time (surgical time from skin to skin), HB1 (preoperative hemoglobin), HB2 (first postoperative hemoglobin), HB3 (hemoglobin lowest postoperative value), HB1-HB2 (difference between preoperative and immediate postoperative hemoglobin), HB2-HB3 (difference between immediate and the lowest postoperative hemoglobin), BT (blood transfusion in units of packed red blood cells), and WBC (first postoperative white blood cells count). * Patients that receive a second blood transfusion. ** The patient received blood transfusion right after his arrival at ICU.

[11] 91w Seventeen patients presented at least one peak of axillary temperature S38.5 8C. Ten patients underwent at least one lumbar puncture to rule out CSF infection. There was no pathogen growth on CSF but three patients presented leukocyte counts on CSF of over 1000/mm 3 and were treated for meningitis with broad-spectrum antibiotics. One of them deceased on the 27 postoperative day. He developed an incisional fistula and the necropsy revealed meningoencephalitis. Patients without bacterial meningitis averaged 109G136 leukocytes/mm 3 , 106G69.4 mg/dl of protein, and 50G13.9 mg/dl of glucose in CSF.

[12] 61w Three other patients developed CSF fistula, two being incisional and one nasal (two had RS and one EHL). Two resolved with continuous lumbar drainage and antibiotics. One developed osteomyelitis requiring bone flap removal, where dura mater was patched during the procedure. Two other patients developed skin infection, but were treated with antibiotics only. Patients' postoperative complications are listed on Table 3.

[13] 53w There was no occurrence of unexpected neurological deficit aggravation. Two patients with RS and one with SWS had residual fine motricity in the affected hand before surgery, which was lost after the procedure. No patient in this series presented hemosiderosis or late-onset hydrocephalus. As mentioned before, one patient deceased after developing postoperative menigoencephalitis.

[14] 462w Table 2 Patient's data separated by surgical technique ID Tech. Path. Age Surg. time HB1 HB2 HB3 HB1-HB2 HB2-HB3 BT WBC 7 2C EHL 7 12:00 11.3 13.3 8* K2 5.3 2 26300 22 2C RS 16 15:30 13.5 13.6 8.8* K0.1 4.8 5 22400 5 2C EHL 16 9:00 14.7 12.1 11.7 2.6 0.4 3 12620 11 2C EHL 18 11:30 14.2 12.5 12 1.7 0.5 4 11760 14 2C RS 11 14:00 12.4 11.8 11.5 0.6 0.3 4 15200 8 2C EHL 33 10:30 16.6 11.9 10.2 4.7 1.7 0 14370 16 2C RS 7 15:30 12.2 12.5 10 K0.3 2.5 4 13200 23 2C RS 7 15:00 12.4 13.2 9.9 K0.8 3.3 4 23000 29 2C SWS 4.5 8:00 13.1 11.7 11.5 1.4 0.2 1 13590 15 2C RS 3 19:30 13.6 12.7 10.1 0.9 2.6 3 12130 12 2C EHL 38 9:30 15.4 14 10.8 1.4 3.2 1 22780 6 2C EHL 19 13:00 14.2 12.8 9.6 1.4 3.2 2 37700 3 2C CDM 1.7 18:00 14.9 14.5 8.6* 0.4 5.9 3,2 30000 1 2C CDM 2.2 17:00 15.8 12.1 9.7 3.7 2.4 2 15600 10 2C EHL 8 13:30 13.3 10.8 10.2 2.5 0.6 4 17830 9 2C EHL 16 12:30 11.6 10.4 10.3 1.2 0.1 5 9800 17 2A RS 8 8:00 12.9 11.9 10 1 1.9 3 10300 2 2A CDM 13 12:30 15.2 13.6 9.6 1.6 4 3 21600 20 2A RS 4.5 13:00 12.9 12.5 10.5 0.4 2 3 22100 24 2A RS 6.5 13:00 13.5 12.3 11.8 1.2 0.5 3 27600 26 2A RS 5.5 8:15 12.5 11.9 11.7 0.6 0.2 3,3 16600 13 2A RS 15 11:00 13.9 14.7 9.7 K0.8 5 4 12100 21 2A RS 13.5 14:00 14 13.5 10.5 0.5 3 4 15800 25 2A RS 5.2 8:30 12.8 13.9 8.8* K1.1 5.1 2 11600 18 2A RS 14 8:50 15 13 9.7 2 3.3 5 13300 27 2A RS 8 10:30 11.5 10.5 10.2 1 0.2 4 12000 28 2A RS 21 8:00 14.5 12.5 7.8* 2 4.7 5 13400 30 1A SWS 0.7 8:00 10 7.5** 12.1 2.5 0 2,9 14100 4 1A CDM 0.6 6:30 13.4 14.1 10.1 K0.7 4 3,5 14200 19 1A RS 21 9:00 15.5 14 8.1* 1.5 5.9 5 23900 See legend at Table 1. Table 3 Surgical complications ID Inc. Disc. VPS CSF fist. Inf. 1 C 3 C C 5 C Osteom. 12 C Mening.* 14 Mening. 15 C Mening. 16 C 17 C 18 C 19 C 20 C 26 Skin Inf. 27 Skin Inf. 28 C ID (patients' identification as listed on table 1), Inc. Disc. (incomplete disconnection), VPS (ventriculoperitoneal shunt), CSF fist. (CSF fistulae), Inf (Infection), Skin inf (skin infection), Mening. (meninigitis), Osteom. (osteomyelitis). *Patient deceased.

[1] 92w This series represents a unique combination of several surgeons performing different hemispherectomy techniques within the same institution. In such a context, differences among techniques cannot be attributed to externally acquired experience as all surgeons were at the same point in their learning curves. Our results better reflect the reality in groups having experience in epilepsy surgery but who are not yet entirely familiar with hemisperectomy. The parameters of surgical time, blood transfusion, hydrocephalus, postoperative seizures, infection, and mortality/unexpected-neurological-deficits were chosen because they are frequently used to justify development of new surgical approaches.

[2] 56w Reduction in surgical time has been cited as an advantage of the techniques from group 2. In our series the surgical time did not differ appreciably amongst the different techniques. Nevertheless, surgical procedures on patients with CDM were approximately 20% longer than those on patients with EHL, although these differences did not translate to noticeable morbidities.

[3] 291w Bleeding is still a concern during hemispherectomies, however, comparing data from different series is not feasible as authors use different parameters to measure blood loss. Kestle et al. [5] and Schramm et al. [6] reported progressive reduction in blood loss or transfusion necessity when comparing techniques from Group 1B, Group 2A and Group 2C. Jonas et al. [7], on the other hand, reported differences in blood loss among pathologies, being higher in patients with hemimegalencephaly. In our series, 29 patients received packed red cells intra-operatively. On average, the amount transfused was higher than in other series but was not linked to pathology or technique [6,8]. Our increased blood volume may reflect higher intra-operative bleeding or, lower tolerance threshold for postoperative anemia adopted by our team of anesthesiologists. Most of our patients left the operating theater with hemoglobin levels above 12 g/dl and none presented hemodynamic instability during the procedure. In the ICU, hemoglobin continued to drop for a few days and six patients needed a second blood transfusion when their hemoglobin levels fell below 10 mg/dl. Notably, postoperative progressive anemia was not associated with intra-operative blood loss, blood transfusion, technique, pathology, or patient age. Initially, it was thought to be the result of bleeding from the subgaleal drain. However, there was a strong association between hemoglobin reduction and leukocyte counts of over 20,000/mm 3 in the earliest postoperative hemogram. As fever and hemodynamic abnormalities are frequent in the postoperative period, it is reasonable to suppose that this fall in hemoglobin may be related to some sort of systemic inflammatory response, however, further studies are needed to clarify this. Thus, the amount of blood loss during and after the surgery is attributable to a combination of surgical technique, pathology and inflammatory factors.

[4] 145w Hydrocephalus is a frequent morbidity following hemispherectomies and may develop under two different circumstances: either within months of the procedure, or years later. Rasmussen believed subacute hydrocephalus was a defect of CSF absorption secondary to the huge removal of the subarachnoid space [9]. Late onset hydrocephalus, for its part, has been attributed to either superficial cerebral hemosiderosis or low-pressure hydrocephalus [10]. In our series however, the most important factor associated to subacute hydrocephalus was pathology, whereby 50% of our patients with CDM needed ventriculoperitoneal shunt. Compiling series from the literature whose shunts were reported by surgical technique and pathology, there is a clear preponderance of subacute hydrocephalus in patients with CDM (Table 4). Since patients with CDM have larger hemispheres, the higher The first figure refers to total of patients reported by the cited series and the second number to total of ventriculoperitoneal shunts inserted.

[5] 391w incidence of hydrocephalus may be a consequence of debris in the surgical cavity, which are more abundant after operating on a thicker parenchyma. The series of Jonas et al. [7] reported higher incidence of shunting in all groups of pathologies although was not included in the table since a proportion of his patients underwent ventriculoperitoneal shunt routinely after hemispherectomy [11]. Therefore, it is not feasible to ascertain the actual rate of hydrocephalus in their series. Di Rocco and Iannelli [12] found correlation between subacute hydrocephalus and patient age, yet not technique. Our results may corroborate these findings to some extent, as patients with CDM are usually considered surgical candidates earlier than in other pathologies. Papers published in the sixties considered superficial cerebral hemosiderosis (SCH) as an inevitable and fatal complication of hemispherectomy [13]. Despite the small number of our patients who underwent anatomical hemispherectomy (Group 1A), there were no instances of late onset hydrocephalus or SCH in our series. In 1973, Rasmussen [14] reported the MNI experiment with follow up over thirty years, of patients with anatomical and subtotal hemispherectomy. The author concluded that a small portion of remnant brain would be sufficient to protect against SCH. This may account to some extent for the lack of such problem in the current literature, as more recent series have used techniques that leave a portion of the brain in the cavity. Although hydrocephalus and SCH may manifest several decades after the procedure, it is the exception, not the rule [15]. The average interval between surgery and SCH has been described as 8 years [9]. As Rasmussen's technique was developed in the seventies, one could expect several cases of SCH by now if the technique were not protective against such a condition. On the other hand, in patients who have undergone AH today (Group 1A), widespread use of CT scan may allow early diagnosis and treatment for bleeding in the surgical cavity shown to be effective in preventing the development of SCH [16]. Finally, Cook et al. [17] presented 34 patients who underwent AH, in which ventriculoperitoneal shunt was used almost routinely, with at least 14 years of follow up and none presented SCH. Therefore, although SCH is still related to AH (group 1A), it may not be considered a common cause of morbidity today so long as patients receive proper follow up.

[6] 188w Currently, causative pathology, not surgical technique, is considered the main factor for postoperative seizure control [18,19]. However, procedures other than from group 1A may leave bridges between the hemispheres or epileptogenic areas over either the insula or the basal frontal lobe that perpetuate seizures spreading [20,21]. Incomplete disconnection occurred in 18.5% of our cases, the same incidence as reported by Peacock et al., [11] in patients operated on with functional hemispherectomy (Group 2A). Shimizu and Maehara [8] reported this in 9% of his patients using modified peirinsular hemispherotomy, while Schramm et al., [6] reported none in their series (both author employed techniques from Group 2C). Finally, Kossoff et al. [18] reported five patients out of 111 reoperated upon after initial hemidecortication (group 1B) due to persistent seizures and residual tissue identified on MRI. Incomplete disconnection may happen using almost all techniques, although it may be reduced according to surgeons' experience. Pathologies with thicker cortical mantles, such as hemimegalencefalia, may increase the chances of its occurrence. Therefore, incomplete disconnection should not be considered a complication but an expected morbidity, at least, at the beginning of the surgeon's learning curve.

[7] 149w Considering operative mortality and major morbidity together, hemispherectomy is still a risky procedure, regardless of technique or pathology. Our series presented a rate of postoperative fever similar to what is reported in the literature. Temperature elevation after hemispherectomy is a common phenomenon and has been attributed aseptic meningitis. On the other hand, we had a high rate of complications like infection and CSF fistulae. We credited such drawback to the teaching characteristic of our institution, where several in-training medical teams contributed to surgeries. One patient deceased in our series due to meningoencephalitis. Other authors have reported brain stem lesion, excessive blood loss, brain swelling, generalized hypoxia, infection, locked-in syndrome, and shunt failure as causes of death or permanent sequelae [6][7][8][21][22][23][24][25]. The death rate reached up to 5.7% in recent series and, even in a series without death or permanent sequelae, life-threatening operative complications occurred in 10% of cases [26].

[1] 54w Hemispherectomies are procedures where pathology and surgical technique interact narrowly. In order to define causes of morbidity both factors have to be analyzed independently. From this standpoint, it is not possible to infer that a specific technique of hemispherectomy has less morbidity or better outcome if results are not adjusted for different causative pathologies.

[1] 77w Thirty patients underwent hemispherectomy in our institution from 1987 to 2003, these being fourteen females and sixteen males with an average age of 11.5G8.9 years, ranging from 7 months to 38 years. Pre-operative work up included a thorough history of the disease, video-EEG with seizure recordings, neuropsychological evaluation, CT scan, and MRI. The Wada test was used only when language lateralization was uncertain. A multidisciplinary group analyzed the data before a final agreement on surgery was reached.

[2] 49w In this series, two patients presented with Sturge-Weber Syndrome (SWS), sixteen with Rasmussen's Syndrome (RS), eight with established hemispheric lesions, sequelae of previous vascular or traumatic event, with or without porencephalic cyst (EHL), and four with cortical development malformations (CDM) (three with hemimegalencephaly and one with hemispheric cortical dysplasia).

[3] 153w In order to analyze our data and compare our results to the literature, hemispherectomy techniques currently in use were allocated to one of two main groups. Group 1 included those techniques that completely remove the cortex from the hemisphere, whilst Group 2 comprised those associated to partial cortical removal and fiber disconnection. Group 1 presented two sub-divisions based on ventricular integrity where Group 1A included techniques that involve opening the lateral ventricles (called anatomical or classical hemispherectomy) and Group 1B those that kept them closed (called hemidecortication or hemicorticectomy). Group 2 was sub-divided into three: functional hemispherectomy (FH), as described by Rasmussen (Group 2A), techniques that disconnect the hemisphere from a vertical approach (Group 2B), and techniques disconnecting the hemisphere from a lateral approach (Group 2C). In the literature techniques included in the groups 2B and 2C are frequently known as hemispherotomy. Fig. 1 shows schematic drawings of techniques included within each group.

[4] 37w Six surgeons operated on thirty patients. Three patients underwent anatomical hemispherectomies (Group 1A), 11 patients underwent the functional hemispherectomy (Group 2A) and 16 were operated on employing three different techniques from the lateral approach group (Group 2C).

[5] 49w Surgical time (from skin incision to skin closure), blood transfusion, hydrocephalus, postoperative seizures, infection, and mortality/unexpected-neurological-deficits were reviewed as parameters of morbidity. At least three samples of peripheral blood were examined for each patient, one preoperatively, one within 12 h following surgery, and another on the third postoperative day.

[6] 51w Mean values and standard deviation were obtained from all analyzed data. The two groups with different leukocyte counts were compared using the Mann-Whitney test, whereas the c 2 test was employed to compare groups with diverse surgical times and diverse volumes of blood transfusion. Statistical significance was considered present for P!0.05.