- Email: [email protected]
Extended anterior craniofacial resection for intracranial extension of malignant tumors
Extended Anterior Craniofacial Resection for Intracranial Extension of Malignant Tumors Mark H. Bilsky,
MD, Dennis H. Kraus, MD, Elliot W. Strong, MD, Louis B. Harrison, Philip H. Gutin, MD, Jatin P. Shah, MD, New York, New York
OBJECTIVE: To review our experience with anterior craniofacial resection for malignant neoplasms with intracranial extension. Survival was analyzed in terms of presence of intracranial extension, extent of intradural disease, tumor histology, and histological status of margins. PATIENTS: In a retrospective review made at a tertiary cancer facility, 26 of the 115 consecutive patients undergoing craniofacial resection for malignant lesions of the anterior skull base had intracranial extension, defined as dural and/or brain extension. Survival was evaluated with the Kaplan-Meier product limit method, and comparisons between individual subgroups were performed using the log-rank test. RESULTS: Patients with intradural extension have a statistically worse disease-specific survival than patients without intracranial extension (P = 0.05). Surgical margins and tumor histology impact on survival. The incidence of local complications was 42% and of systemic complications, 6%. CONCLUSION: Anterior craniofacial resection is indicated for patients with resectable disease. The complication rate is comparable with that of patients without intracranial extension. Gross total resection with histologically negative margins portends a better prognosis. Esthesioneuroblastoma has a better prognosis than other tumor types. Am J Surg. 1997;174:565-566.0 1997 by Excerpta Medica, Inc.
raniofacial surgery of tumors involving the anterior skull base has been reported from a number of centers.lm9 The initial application of this technique for the resection of orbital tumors was reported by Dandy’ in 1941. Ketcham et al* expanded the role of craniofacial resection for tumors of the sinonasal region, which involved the anterior skull base. The early application of this technique was restricted to tumors of the paranasal sinus approaching the skull base, but limited to the sinuses with or
C
From the Neurosurgery (MHB, PHG) and Head and Neck (DHK, EWS, JPS) Services, Department of Surgery, and the Brachytherapy Service (LBH), Department of Radiation Oncology, Memorial Sloan-Kettering Cancer Center, New York, New York. Requests for reprints should be addressed to Dennis H. Kraus, MD, Box 285, 1275 York Avenue, Memorial Sloan-Kettering Cancer Center, New York, New York 10021. Presented at the 43rd Annual Meeting of the Society of Head and Neck Surgeons, Cancun, Mexico, April lo-12,1997.
0 1997 by Excerpta All rights reserved.
Medica,
Inc.
MD,
without unilateral orbital involvement. Intracranial extension had been considered a contraindiciation to surgery. Recent studies, however, have shown the percentage of patients undergoing anterior craniofacial resection (ACFR) for tumors with intracranial extension ranges from 27% to 57%.4-9 Long-term palliation and the potential for cure are thought to justify this aggressive approach to these advanced malignant tumors of the skull base. Intracranial involvement is widely recognized as a negative prognostic factor for long-term outcome. Other variables that impact on outcome include histology, extent of resection, and surgical margins. This paper assesses the role of craniofacial resection in patients with malignant tumors involving the skull base that extend to the dura and/or brain.
PATIENTS
AND
METHODS
Of the 115 patients who underwent ACFR at Memorial Sloan-Kettering Cancer Center for malignant tumors involving the anterior skull base between 1973 and 1995, 26 patients (23%) had intracranial involvement. Ages ranged from 11 to 70 years with a mean of 46 years. Seventeen patients were male and 9 were female. Histologic diagnoses varied widely and included the following: sarcoma (6), squamous cell carcinoma (5), adenoid cystic carcinoma (4), esthesioneuroblastoma (4), adenocarcinoma (3), anaplastic carcinoma (3), and mucoepidermoid carcinoma (1). The most common sites of origin as judged by the epicenter of the tumor on radiographic evaluation and intraoperatively were the nasal cavity and ethmoid sinus (Table I). Preoperative radiographic assessment included computed tomography (CT) scan and magnetic resonance imaging (MRI) scan to determine the extent of the lesion. Additionally, patients underwent an extent-of-disease work-up that included chest radiography, routine blood chemistries, and liver function tests. Seventeen patients had dural invasion only and 9 patients had both dural and brain involvement. Two patients demonstrated tumor extension adjacent to but not involving the cavernous sinus. Resection of the anterior skull base as well as dural resection was required in all patients. Twenty patients required orbital exenteration. Twenty patients had galeal-pericranial flap reconstruction, and 6 had pericranial flap reconstruction only. Seven patients additionally had free flap reconstruction, most commonly employing the rectus abdominis muscle. Technical details of ACFR have been published elsewhere, but will be briefly described here. A subarachnoid drain is placed preoperatively at the L4-5 interspace to facilitate cerebrospinal fluid drainage during the procedure. Approximately 200 cc of spinal fluid are drained in small aliquots throughout the procedure in order to provide brain 0002-961 O/97/$1 7.00 PII SOOO2-961 0(97)00172-4
565
TABLE
I Tumor
Dfsease-Specihc
Epicenter Number
Nasal Ethmoid Lacrimal Frontal Orbital Total
Survival
of Patients 9 8 4 4 1 26
Dual
Involvement
0.2 oi5
0
relaxation. With adequate cerebrospinal fluid drainage, only minimal brain retraction is intermittently required throughout the procedure. Patients receive preoperative steroids and phenytoin, which are rapidly tapered in the postoperative period. Vancomycin hydrochloride, ceftazadime sodium, and metronidazole are started preoperatively and continued until the drains are removed. The intracranial portion of the procedure is generally performed first. A bicoronal skin incision is fashioned. The posterior aspect of the scalp flap is elevated, and the pericranial graft is initially dissected. The pericranial graft dissection is then carried approximately 2 cm.anterior to the bicoronal skin incision, at which point the galea is dissected with the pericranium in order to create the flap for reconstruction of the skull base. Leaving a cuff of galea on both the anterior and posterior aspects of the scalp incision facilitates closure of the bicoronal incision at the end of the case. A single burr hole is placed over the sagittal sinus and the craniotomy bone flap is created with the lateral extent at the midpupillary line. A portion of the nasion attached to the craniotomy bone flap is incised using small osteotomes in order to facilitate access to the anterior central skull base. Meticulous dissection of the frontal sinus mucosa and resection of the posterior wall of the sinus are then carried out. The dura is dissected off the anterior skull base. The olfactory nerves are sacrificed bilaterally. The dura is resected according to the extent of invasion. Pericranium is generally employed to close the ensuing dural defect in a water-tight fashion. Appropriate cuts are then made in the cribriform plate, fovea ethmoidalis, planum sphenoidale, and oribtal roof depending on the extent of the tumor. The transfacial incision is performed through a lateral rhinotomy or Weber-Ferguson incision depending on the extent of resection. The resection incorporates the lateral nasal wall, septum, and ethmoid contents in continuity with the bony anterior skull base. Larger resections include the maxilla, orbit, infratemporal fossa, and other involved structures. Every attempt is made to perform an en bloc resection even with the presence of intracranial involvement. Following tumor resection, four small holes are drilled in the planum sphenoidale and superolateral fovea ethmoidalis in order to secure the galeal-pericranial graft for the skull base reconstruction. In the instance of extended maxillectomy and/or orbital exenteration, a rectus abdominis free flap is used for reconstruction of the cranial base and facial defect. A drain is placed in the epidural space. The craniotomy bone flap with the attached upper half of the nasal bones is replaced and secured with miniplates for fa566
THE AMERICAN
JOURNAL
OF SURGERY~
VOLUME
174
10 Years
15
20
Figure 1. Kaplan-Meier curves comparing disease-specific survival in patients with intracranial extension and patients without intracranial extension. The difference is statistically significant (P = 0.004).
rmease-specmc survival
1 0.8
h
‘X
I 6
Negative
0.4
0.2 Positive A--
0
5
0
10
1-A 15
20 J
Figure vival in positive cant (P
2. Kaplan-Meier curves comparing disease-specific surpatients with negative histologic margins and patients with histologic margins. The difference is statistically signifi= 0.05).
cial and calavarial reconstruction. All soft tissue incisions are then closed. In this series, gross total resection was achieved in all patients. Fourteen patients had negative microscopic margins and 12 patients had positive margins, as judged by permanent section histopathologic review. Of the patients with positive margins, 7 of 17 (41%) had dural invasion only and 5 of 9 (55%) had brain invasion. Nine patients had preoperative radiation therapy and presented for resection of persistent or recurrent disease. Six of these patients also received various regimens of preoperative chemotherapy. Thirteen patients received postoperative radiation therapy. Radiation doses ranged from 5,500 to 7,100 cGy. One additional patient received radioactive iodine seeds adjacent to the cavernous sinus and an additional boost of 2,520 cGy external beam radiation therapy. Four patients received no adjuvant therapy. One patient died of sepsis in the immediate postoperative period. The second patient had a gross total resection of an adenocarcinoma with negative microscopic margins. One additional Datient had radiation theraw for retinoblastoma prior to developing a postradiation sarcoma of the ethmoid and was not a candidate for further radiation therapy. The final patient had an anaplastic carcinoma that recurred prior to planned postoperative radiation therapy. Routine follow-up and radiographic imaging were performed on all patients at regular intervals. The sites of re.
NOVEMBER
I
1997
1CBANIOFACIAL
TABLE
of Cancer
Complications Number
Cerebrospinal fluid fistula Wound/bone infection Delay cognitive function Hematoma Tension pneumocephalus Cavernous carotid injury Meningitis Death
of Patients 3 3 2 2 2 1 1 1
currence were documented. Survival analysis was performed using the Kaplan-Meier product limit method,” and comparison between individual subgroups was performed using the log-rank test. ‘r Dichotomous variables were compared with the chi-square test.t2
RESULTS The disease-specific survival for patients undergoing ACFR with intracranial extension in this series was 40% at 2 years and 33% at 5 years. This compares with a diseasespecific survival of 74% at 2 years and 64% at 5 years for patients undergoing ACFR without intracranial extension at our institution (Figure 1). Comparison between these two groups is statistically significant (P = 0.004). There was a trend toward worsened survival in patients who showed brain invasion versus dural invasion only, but this difference was not statistically significant (P = 0.48). Patients with dural invasion only had both a 2- and 5-year survival of 44% and those with brain invasion had a 2- and 5-year survival of 33% and 17%, respectively. Seven patients remain alive without evidence of disease at a mean of 5.6 years. Four patients are alive with evidence of recurrent disease. Fifteen patients have died of disease or other causes. One patient died in the immediate postoperative period secondary to sepsis. Seventeen patients demonstrated recurrent disease. The patterns of failure include the following: local (5), distant (4), local/distant (3), regional/distant (3), and local/regional/distant (2). The disease-specific survival in patients with negative histologic margins was 54% at both 2- and 5-year follow-up compared with patients with positive histologic margins who had a 2- and 5-year survival of 15% and O%, respectively (Figure 2). This is a statistically significant difference (P = 0.05). However, local control was achieved in 8 patients (58%) with negative margins and 6 patients with THE
of
Alive with Disease
0 1 1 3 1 1 0 7
Ill
Complication
INTRACRANIAL
EXTENSION/BILSKY
ET AL
by Histology
No Evidence Disease
Sarcoma Squamous cell carcinoma Adenoid cystic carcinoma Esthesioneuroblastoma Adenocarcinoma Anaplastic carcinoma Mucoepidermoid carcinoma Total
TABLE
FOR
II Outcome
Type
RESECTION
AMERICAN
Dead of Disease
1 2 0 1 0 0 0 4
5 2 3 0 2 2 1 15
positive margins (55%). The difference in survival is attributed to distant metastatic disease in patients with positive margins. Prior treatment did not appear to impact on outcome. Of the 9 patients who were operated on for persistent or recurrent disease, 3 remain without evidence of recurrence and 6 are dead of disease. The 17 patients undergoing surgery as primary treatment include 4 who are without evidence of disease, 4 who are alive with disease, and 9 who are dead of disease. The small number of patients in each histologic subgroup precludes statistical analysis; however, esthesioneuroblastoma portends the best prognosis. Three patients are alive without evidence of recurrent disease and 1 patient is alive with disease. All of the 6 patients with sarcoma had highgrade lesions. Of these patients, 4 are dead of disease, 1 is alive with recurrent disease, and 1 died in the immediate postoperative period. Patients with the remaining histologies have at least 1 patient each with no evidence of disease, except mucoepidermoid carcinoma in which the single pa+ tient with this entity is dead of disease (Table II). Major, local complications occurred in 11 patients (42%), and 2 patients (8%) had systemic complications. Eight of 17 patients (47%) with dural invasion only and 3 patients (33%) with brain invasion had complications. On the other hand, patients undergoing ACFR without intracranial extension had a 30% complication rate. This difference was not statistically different (P = 0.36). The most common complications were CSF fistula (3) and bone plate infection (3) (Table III).
COMMENTS The widespread application of ACFR for tumors extending through the anterior skull base to involve the dura and brain requires close examination. In this series, patients undergoing ACFR for tumors showing intracranial extension have a worse disease-specific survival than patients without intracranial extension. Further, there is a trend toward worsened survival with brain invasion when compared with dural invasion only, although this difference was not statistically significant. This difference confirms the conclusion of most other series in which intracranial extension is thought to be a neg ative prognostic variable. Van Tuyl and Gussack reported a series of 21 patients with a mean disease-survival of 9.5 months. Patients who did not require dural resection had a 83% rate of local control versus 22% in patients requiring dural resection. The majority of patients with local recurJOURNAL
OF SURGERY”
VOLUME
174
NOVEMBER
1997
567
CRANIOFACIAL
RESECTION
FOR
INTRACRANIAL
EXTENSION/BILSKY
rences were found to have direct brain extension or spinal cord seeding of the tumor through the cerebrospinal fluid that was thought to be secondary to involvement of the dura. Kraus4 also found dural extension to be a negative prognostic factor with 86% of patients undergoing ACFR involving the intracranial cavity died of disease. Catalan0 et al8 also reported intracranial involvement as a major prognostic variable. Among the 73 patients undergoing ACFR, 25 patients (34%) demonstrated dural and/or brain involvement. Of the 13 patients demonstrating recurrent disease, 9 patients (63%) had intracranial involvement. Four patients (33%) with dural involvement only and 5 (38%) with brain involvement developed recurrent disease. Two studies did not show intracranial involvement to impact on survival. Bohrer et al5 reported a series of 53 patients with 30 patients (57%) showing dural involvement and 8 (15%) showing brain invasion. No significant difference was seen in patients with or without dural involvement at 1 and 3 years, and no difference was seen at 1 year with cerebral involvement. There were no 3-year survival data for patients with brain invasion. Additionally, Richtsmeier et al6 reviewed 32 patients undergoing ACFR, of whom 26 had malignant disease and 11 had intracranial extension. Five (45%) of these patients died of disease. Despite negative gross intraoperative tumor margins, 12 patients were found to have positive microscopic margins on final histologic review. The disease-specific survival in the group with positive margins was worse than for patients in whom negative margins were achieved. The rate of local recurrence was essentially the same in both groups. Patients with positive margins generally died of distant metastases. This may reflect a more advanced, aggressive tumor at the time of resection. No statistical conclusions could be drawn from evaluation of the histologic tumor types owing to the small numbers in each group. Esthesioneuroblastoma patients appeared to have the best prognosis. Among the 4 patients with this disease, 3 are alive without recurrent disease and 1 is alive with local recurrence. The diagnosis of high-grade sarcoma portends a poor prognosis. The local complication rate in this series was 42% and the systemic complication rate was 8%. This rate is consistent with the relatively high major complication rate found in other series that ranges from 28% to 50%.4+9 A single postoperative death occurred, for a mortality rate of 4%. In this series, the complication rate was not significantly different with or without the presence of intracranial extension. This experience confirms our current approach to malignant tumors involving the anterior skull base with intracranial extension. Patients with isolated dura and/or limited brain invasion are candidates for potentially curative resec-
568
THE AMERICAN
JOURNAL
OF
SURGERY@
VOLUME
174
ET AL 1
tion, regardless of previous treatment. Contraindications to surgery include extension of tumor into the cavernous sinus, optic chiasm invasion, involvement of an only seeing eye, and metastatic disease. Massive frontal lobe involvement is a relative contraindication with the decision partly dependent on the histologic grade of the tumor. The decision to operate reflects the belief in an ability to obtain a complete resection with negative histologic margins.
CONCLUSION Anterior craniofacial resection for malignant tumors with intracranial extension is indicated for selected patients with resectable disease. The survival rate is lower for patients with dural and/or brain involvement than for patients without intracranial extension. However, the complication rates are similar between these two groups. A trend was observed for improved survival for patients with dural involvement alone in comparison with patients with brain involvement. In this series, poor prognostic indicators include positive pathologic margins and histological tumor type. Patients with esthesioneuroblastoma have a better prognosis than other tumor types.
REFERENCES 1. Dandy WE. Orbital Tumors: Results Following the Transcranial Obendue A&k. New York: Oskar Priest; 1941:1154. 2. Ketcham AS, Wilkins RH, Van Buren JM, Smith RR. A combined intracranial facial approach to the paranasal sinuses. Am J Surg. 1963;106:698-703. 3. Shah JP, Sundarasan N, Galicich JH, Strong EW. Craniofacial resection for tumors involving the base of skull. 1988;154:352. 4. Kraus DH, Sterman BM, Levine HL, et al. Factors influencing survival in ethmoid sinus cancer. Arch Otolaryngol Head Neck Surg. 1992;118:367-372. 5. Bohrer PS, Donald PJ, Wengen DF, et al. The significance of dural and cerebral invasion by skull base malignancies. Skull Base Surgery. Update I. Amsterdam: Kugler Publications; 1993:133-138. 6. Richtsmeier WJ, Briggs RJ, Koch WM, et al. Complications and early outcome of anterior craniofacial resection. Arch Otoloaryngol Head Neck Surg. 1992;118:913-917. 7. Van Tuyl R, Gussack GS. Prognostic factors in craniofacial surgery. Laryngoscope. 1991;101:240-244. 8. Catalan0 PJ, Hecht CS, Biller HF, et al. Craniofacial resection. an analysis of 73 cases. Arch Otofuryngol Head Neck Surg. 1994;120:1203-1208. 9. Shah JP, Kraus DH, Arbit E, et al. Craniofacial resection for tumors involving the anterior skull base. Otoluryngol Head Neck Surg. 1992;4:387. 10. Kaplan E, Meier I’. Non-parametric estimation for incomplete observation. J Am Statistical Assoc. 1958;53:457-481. 11. Peto R, Pike MC, Armitage P, et al. Design and analysis of randomized clinical trials requiring prolonged observation of each patient. Part 2. Analysis and examples. BrJ Cancer. 1977;35:1-39. 12. Fleiss JL. Statistical methods for rates and proportions. New York: Wiley Csr Sons; 1973:24-26.
NOVEMBER
1997
MD, Dennis H. Kraus, MD, Elliot W. Strong, MD, Louis B. Harrison, Philip H. Gutin, MD, Jatin P. Shah, MD, New York, New York
OBJECTIVE: To review our experience with anterior craniofacial resection for malignant neoplasms with intracranial extension. Survival was analyzed in terms of presence of intracranial extension, extent of intradural disease, tumor histology, and histological status of margins. PATIENTS: In a retrospective review made at a tertiary cancer facility, 26 of the 115 consecutive patients undergoing craniofacial resection for malignant lesions of the anterior skull base had intracranial extension, defined as dural and/or brain extension. Survival was evaluated with the Kaplan-Meier product limit method, and comparisons between individual subgroups were performed using the log-rank test. RESULTS: Patients with intradural extension have a statistically worse disease-specific survival than patients without intracranial extension (P = 0.05). Surgical margins and tumor histology impact on survival. The incidence of local complications was 42% and of systemic complications, 6%. CONCLUSION: Anterior craniofacial resection is indicated for patients with resectable disease. The complication rate is comparable with that of patients without intracranial extension. Gross total resection with histologically negative margins portends a better prognosis. Esthesioneuroblastoma has a better prognosis than other tumor types. Am J Surg. 1997;174:565-566.0 1997 by Excerpta Medica, Inc.
raniofacial surgery of tumors involving the anterior skull base has been reported from a number of centers.lm9 The initial application of this technique for the resection of orbital tumors was reported by Dandy’ in 1941. Ketcham et al* expanded the role of craniofacial resection for tumors of the sinonasal region, which involved the anterior skull base. The early application of this technique was restricted to tumors of the paranasal sinus approaching the skull base, but limited to the sinuses with or
C
From the Neurosurgery (MHB, PHG) and Head and Neck (DHK, EWS, JPS) Services, Department of Surgery, and the Brachytherapy Service (LBH), Department of Radiation Oncology, Memorial Sloan-Kettering Cancer Center, New York, New York. Requests for reprints should be addressed to Dennis H. Kraus, MD, Box 285, 1275 York Avenue, Memorial Sloan-Kettering Cancer Center, New York, New York 10021. Presented at the 43rd Annual Meeting of the Society of Head and Neck Surgeons, Cancun, Mexico, April lo-12,1997.
0 1997 by Excerpta All rights reserved.
Medica,
Inc.
MD,
without unilateral orbital involvement. Intracranial extension had been considered a contraindiciation to surgery. Recent studies, however, have shown the percentage of patients undergoing anterior craniofacial resection (ACFR) for tumors with intracranial extension ranges from 27% to 57%.4-9 Long-term palliation and the potential for cure are thought to justify this aggressive approach to these advanced malignant tumors of the skull base. Intracranial involvement is widely recognized as a negative prognostic factor for long-term outcome. Other variables that impact on outcome include histology, extent of resection, and surgical margins. This paper assesses the role of craniofacial resection in patients with malignant tumors involving the skull base that extend to the dura and/or brain.
PATIENTS
AND
METHODS
Of the 115 patients who underwent ACFR at Memorial Sloan-Kettering Cancer Center for malignant tumors involving the anterior skull base between 1973 and 1995, 26 patients (23%) had intracranial involvement. Ages ranged from 11 to 70 years with a mean of 46 years. Seventeen patients were male and 9 were female. Histologic diagnoses varied widely and included the following: sarcoma (6), squamous cell carcinoma (5), adenoid cystic carcinoma (4), esthesioneuroblastoma (4), adenocarcinoma (3), anaplastic carcinoma (3), and mucoepidermoid carcinoma (1). The most common sites of origin as judged by the epicenter of the tumor on radiographic evaluation and intraoperatively were the nasal cavity and ethmoid sinus (Table I). Preoperative radiographic assessment included computed tomography (CT) scan and magnetic resonance imaging (MRI) scan to determine the extent of the lesion. Additionally, patients underwent an extent-of-disease work-up that included chest radiography, routine blood chemistries, and liver function tests. Seventeen patients had dural invasion only and 9 patients had both dural and brain involvement. Two patients demonstrated tumor extension adjacent to but not involving the cavernous sinus. Resection of the anterior skull base as well as dural resection was required in all patients. Twenty patients required orbital exenteration. Twenty patients had galeal-pericranial flap reconstruction, and 6 had pericranial flap reconstruction only. Seven patients additionally had free flap reconstruction, most commonly employing the rectus abdominis muscle. Technical details of ACFR have been published elsewhere, but will be briefly described here. A subarachnoid drain is placed preoperatively at the L4-5 interspace to facilitate cerebrospinal fluid drainage during the procedure. Approximately 200 cc of spinal fluid are drained in small aliquots throughout the procedure in order to provide brain 0002-961 O/97/$1 7.00 PII SOOO2-961 0(97)00172-4
565
TABLE
I Tumor
Dfsease-Specihc
Epicenter Number
Nasal Ethmoid Lacrimal Frontal Orbital Total
Survival
of Patients 9 8 4 4 1 26
Dual
Involvement
0.2 oi5
0
relaxation. With adequate cerebrospinal fluid drainage, only minimal brain retraction is intermittently required throughout the procedure. Patients receive preoperative steroids and phenytoin, which are rapidly tapered in the postoperative period. Vancomycin hydrochloride, ceftazadime sodium, and metronidazole are started preoperatively and continued until the drains are removed. The intracranial portion of the procedure is generally performed first. A bicoronal skin incision is fashioned. The posterior aspect of the scalp flap is elevated, and the pericranial graft is initially dissected. The pericranial graft dissection is then carried approximately 2 cm.anterior to the bicoronal skin incision, at which point the galea is dissected with the pericranium in order to create the flap for reconstruction of the skull base. Leaving a cuff of galea on both the anterior and posterior aspects of the scalp incision facilitates closure of the bicoronal incision at the end of the case. A single burr hole is placed over the sagittal sinus and the craniotomy bone flap is created with the lateral extent at the midpupillary line. A portion of the nasion attached to the craniotomy bone flap is incised using small osteotomes in order to facilitate access to the anterior central skull base. Meticulous dissection of the frontal sinus mucosa and resection of the posterior wall of the sinus are then carried out. The dura is dissected off the anterior skull base. The olfactory nerves are sacrificed bilaterally. The dura is resected according to the extent of invasion. Pericranium is generally employed to close the ensuing dural defect in a water-tight fashion. Appropriate cuts are then made in the cribriform plate, fovea ethmoidalis, planum sphenoidale, and oribtal roof depending on the extent of the tumor. The transfacial incision is performed through a lateral rhinotomy or Weber-Ferguson incision depending on the extent of resection. The resection incorporates the lateral nasal wall, septum, and ethmoid contents in continuity with the bony anterior skull base. Larger resections include the maxilla, orbit, infratemporal fossa, and other involved structures. Every attempt is made to perform an en bloc resection even with the presence of intracranial involvement. Following tumor resection, four small holes are drilled in the planum sphenoidale and superolateral fovea ethmoidalis in order to secure the galeal-pericranial graft for the skull base reconstruction. In the instance of extended maxillectomy and/or orbital exenteration, a rectus abdominis free flap is used for reconstruction of the cranial base and facial defect. A drain is placed in the epidural space. The craniotomy bone flap with the attached upper half of the nasal bones is replaced and secured with miniplates for fa566
THE AMERICAN
JOURNAL
OF SURGERY~
VOLUME
174
10 Years
15
20
Figure 1. Kaplan-Meier curves comparing disease-specific survival in patients with intracranial extension and patients without intracranial extension. The difference is statistically significant (P = 0.004).
rmease-specmc survival
1 0.8
h
‘X
I 6
Negative
0.4
0.2 Positive A--
0
5
0
10
1-A 15
20 J
Figure vival in positive cant (P
2. Kaplan-Meier curves comparing disease-specific surpatients with negative histologic margins and patients with histologic margins. The difference is statistically signifi= 0.05).
cial and calavarial reconstruction. All soft tissue incisions are then closed. In this series, gross total resection was achieved in all patients. Fourteen patients had negative microscopic margins and 12 patients had positive margins, as judged by permanent section histopathologic review. Of the patients with positive margins, 7 of 17 (41%) had dural invasion only and 5 of 9 (55%) had brain invasion. Nine patients had preoperative radiation therapy and presented for resection of persistent or recurrent disease. Six of these patients also received various regimens of preoperative chemotherapy. Thirteen patients received postoperative radiation therapy. Radiation doses ranged from 5,500 to 7,100 cGy. One additional patient received radioactive iodine seeds adjacent to the cavernous sinus and an additional boost of 2,520 cGy external beam radiation therapy. Four patients received no adjuvant therapy. One patient died of sepsis in the immediate postoperative period. The second patient had a gross total resection of an adenocarcinoma with negative microscopic margins. One additional Datient had radiation theraw for retinoblastoma prior to developing a postradiation sarcoma of the ethmoid and was not a candidate for further radiation therapy. The final patient had an anaplastic carcinoma that recurred prior to planned postoperative radiation therapy. Routine follow-up and radiographic imaging were performed on all patients at regular intervals. The sites of re.
NOVEMBER
I
1997
1CBANIOFACIAL
TABLE
of Cancer
Complications Number
Cerebrospinal fluid fistula Wound/bone infection Delay cognitive function Hematoma Tension pneumocephalus Cavernous carotid injury Meningitis Death
of Patients 3 3 2 2 2 1 1 1
currence were documented. Survival analysis was performed using the Kaplan-Meier product limit method,” and comparison between individual subgroups was performed using the log-rank test. ‘r Dichotomous variables were compared with the chi-square test.t2
RESULTS The disease-specific survival for patients undergoing ACFR with intracranial extension in this series was 40% at 2 years and 33% at 5 years. This compares with a diseasespecific survival of 74% at 2 years and 64% at 5 years for patients undergoing ACFR without intracranial extension at our institution (Figure 1). Comparison between these two groups is statistically significant (P = 0.004). There was a trend toward worsened survival in patients who showed brain invasion versus dural invasion only, but this difference was not statistically significant (P = 0.48). Patients with dural invasion only had both a 2- and 5-year survival of 44% and those with brain invasion had a 2- and 5-year survival of 33% and 17%, respectively. Seven patients remain alive without evidence of disease at a mean of 5.6 years. Four patients are alive with evidence of recurrent disease. Fifteen patients have died of disease or other causes. One patient died in the immediate postoperative period secondary to sepsis. Seventeen patients demonstrated recurrent disease. The patterns of failure include the following: local (5), distant (4), local/distant (3), regional/distant (3), and local/regional/distant (2). The disease-specific survival in patients with negative histologic margins was 54% at both 2- and 5-year follow-up compared with patients with positive histologic margins who had a 2- and 5-year survival of 15% and O%, respectively (Figure 2). This is a statistically significant difference (P = 0.05). However, local control was achieved in 8 patients (58%) with negative margins and 6 patients with THE
of
Alive with Disease
0 1 1 3 1 1 0 7
Ill
Complication
INTRACRANIAL
EXTENSION/BILSKY
ET AL
by Histology
No Evidence Disease
Sarcoma Squamous cell carcinoma Adenoid cystic carcinoma Esthesioneuroblastoma Adenocarcinoma Anaplastic carcinoma Mucoepidermoid carcinoma Total
TABLE
FOR
II Outcome
Type
RESECTION
AMERICAN
Dead of Disease
1 2 0 1 0 0 0 4
5 2 3 0 2 2 1 15
positive margins (55%). The difference in survival is attributed to distant metastatic disease in patients with positive margins. Prior treatment did not appear to impact on outcome. Of the 9 patients who were operated on for persistent or recurrent disease, 3 remain without evidence of recurrence and 6 are dead of disease. The 17 patients undergoing surgery as primary treatment include 4 who are without evidence of disease, 4 who are alive with disease, and 9 who are dead of disease. The small number of patients in each histologic subgroup precludes statistical analysis; however, esthesioneuroblastoma portends the best prognosis. Three patients are alive without evidence of recurrent disease and 1 patient is alive with disease. All of the 6 patients with sarcoma had highgrade lesions. Of these patients, 4 are dead of disease, 1 is alive with recurrent disease, and 1 died in the immediate postoperative period. Patients with the remaining histologies have at least 1 patient each with no evidence of disease, except mucoepidermoid carcinoma in which the single pa+ tient with this entity is dead of disease (Table II). Major, local complications occurred in 11 patients (42%), and 2 patients (8%) had systemic complications. Eight of 17 patients (47%) with dural invasion only and 3 patients (33%) with brain invasion had complications. On the other hand, patients undergoing ACFR without intracranial extension had a 30% complication rate. This difference was not statistically different (P = 0.36). The most common complications were CSF fistula (3) and bone plate infection (3) (Table III).
COMMENTS The widespread application of ACFR for tumors extending through the anterior skull base to involve the dura and brain requires close examination. In this series, patients undergoing ACFR for tumors showing intracranial extension have a worse disease-specific survival than patients without intracranial extension. Further, there is a trend toward worsened survival with brain invasion when compared with dural invasion only, although this difference was not statistically significant. This difference confirms the conclusion of most other series in which intracranial extension is thought to be a neg ative prognostic variable. Van Tuyl and Gussack reported a series of 21 patients with a mean disease-survival of 9.5 months. Patients who did not require dural resection had a 83% rate of local control versus 22% in patients requiring dural resection. The majority of patients with local recurJOURNAL
OF SURGERY”
VOLUME
174
NOVEMBER
1997
567
CRANIOFACIAL
RESECTION
FOR
INTRACRANIAL
EXTENSION/BILSKY
rences were found to have direct brain extension or spinal cord seeding of the tumor through the cerebrospinal fluid that was thought to be secondary to involvement of the dura. Kraus4 also found dural extension to be a negative prognostic factor with 86% of patients undergoing ACFR involving the intracranial cavity died of disease. Catalan0 et al8 also reported intracranial involvement as a major prognostic variable. Among the 73 patients undergoing ACFR, 25 patients (34%) demonstrated dural and/or brain involvement. Of the 13 patients demonstrating recurrent disease, 9 patients (63%) had intracranial involvement. Four patients (33%) with dural involvement only and 5 (38%) with brain involvement developed recurrent disease. Two studies did not show intracranial involvement to impact on survival. Bohrer et al5 reported a series of 53 patients with 30 patients (57%) showing dural involvement and 8 (15%) showing brain invasion. No significant difference was seen in patients with or without dural involvement at 1 and 3 years, and no difference was seen at 1 year with cerebral involvement. There were no 3-year survival data for patients with brain invasion. Additionally, Richtsmeier et al6 reviewed 32 patients undergoing ACFR, of whom 26 had malignant disease and 11 had intracranial extension. Five (45%) of these patients died of disease. Despite negative gross intraoperative tumor margins, 12 patients were found to have positive microscopic margins on final histologic review. The disease-specific survival in the group with positive margins was worse than for patients in whom negative margins were achieved. The rate of local recurrence was essentially the same in both groups. Patients with positive margins generally died of distant metastases. This may reflect a more advanced, aggressive tumor at the time of resection. No statistical conclusions could be drawn from evaluation of the histologic tumor types owing to the small numbers in each group. Esthesioneuroblastoma patients appeared to have the best prognosis. Among the 4 patients with this disease, 3 are alive without recurrent disease and 1 is alive with local recurrence. The diagnosis of high-grade sarcoma portends a poor prognosis. The local complication rate in this series was 42% and the systemic complication rate was 8%. This rate is consistent with the relatively high major complication rate found in other series that ranges from 28% to 50%.4+9 A single postoperative death occurred, for a mortality rate of 4%. In this series, the complication rate was not significantly different with or without the presence of intracranial extension. This experience confirms our current approach to malignant tumors involving the anterior skull base with intracranial extension. Patients with isolated dura and/or limited brain invasion are candidates for potentially curative resec-
568
THE AMERICAN
JOURNAL
OF
SURGERY@
VOLUME
174
ET AL 1
tion, regardless of previous treatment. Contraindications to surgery include extension of tumor into the cavernous sinus, optic chiasm invasion, involvement of an only seeing eye, and metastatic disease. Massive frontal lobe involvement is a relative contraindication with the decision partly dependent on the histologic grade of the tumor. The decision to operate reflects the belief in an ability to obtain a complete resection with negative histologic margins.
CONCLUSION Anterior craniofacial resection for malignant tumors with intracranial extension is indicated for selected patients with resectable disease. The survival rate is lower for patients with dural and/or brain involvement than for patients without intracranial extension. However, the complication rates are similar between these two groups. A trend was observed for improved survival for patients with dural involvement alone in comparison with patients with brain involvement. In this series, poor prognostic indicators include positive pathologic margins and histological tumor type. Patients with esthesioneuroblastoma have a better prognosis than other tumor types.
REFERENCES 1. Dandy WE. Orbital Tumors: Results Following the Transcranial Obendue A&k. New York: Oskar Priest; 1941:1154. 2. Ketcham AS, Wilkins RH, Van Buren JM, Smith RR. A combined intracranial facial approach to the paranasal sinuses. Am J Surg. 1963;106:698-703. 3. Shah JP, Sundarasan N, Galicich JH, Strong EW. Craniofacial resection for tumors involving the base of skull. 1988;154:352. 4. Kraus DH, Sterman BM, Levine HL, et al. Factors influencing survival in ethmoid sinus cancer. Arch Otolaryngol Head Neck Surg. 1992;118:367-372. 5. Bohrer PS, Donald PJ, Wengen DF, et al. The significance of dural and cerebral invasion by skull base malignancies. Skull Base Surgery. Update I. Amsterdam: Kugler Publications; 1993:133-138. 6. Richtsmeier WJ, Briggs RJ, Koch WM, et al. Complications and early outcome of anterior craniofacial resection. Arch Otoloaryngol Head Neck Surg. 1992;118:913-917. 7. Van Tuyl R, Gussack GS. Prognostic factors in craniofacial surgery. Laryngoscope. 1991;101:240-244. 8. Catalan0 PJ, Hecht CS, Biller HF, et al. Craniofacial resection. an analysis of 73 cases. Arch Otofuryngol Head Neck Surg. 1994;120:1203-1208. 9. Shah JP, Kraus DH, Arbit E, et al. Craniofacial resection for tumors involving the anterior skull base. Otoluryngol Head Neck Surg. 1992;4:387. 10. Kaplan E, Meier I’. Non-parametric estimation for incomplete observation. J Am Statistical Assoc. 1958;53:457-481. 11. Peto R, Pike MC, Armitage P, et al. Design and analysis of randomized clinical trials requiring prolonged observation of each patient. Part 2. Analysis and examples. BrJ Cancer. 1977;35:1-39. 12. Fleiss JL. Statistical methods for rates and proportions. New York: Wiley Csr Sons; 1973:24-26.
NOVEMBER
1997