Journal of Neuro-Oncology 14: 263--270,1992. © 1992KluwerAcademic Publishers. Printedin the Netherlands. Clinical Study
The prognostic significance of postoperative residual contrast enhancement on CT scan in pediatric patients with medulloblastoma
Jeffrey P. Bourne, 1 Russell Geyer, 1 Mitchel Berger, 2 Brian Griffin 3 and Jerrold Milstein 4 Pediatrics 1, Neurological Surgery e, Radiation Oncology 3, Neurology 4, Children's Hospital & Med Center, University of Washington, Seattle, Washington, USA
Key words: medulloblastoma, pediatric brain tumor, computerized tomography Abstract
The clinical and therapeutic features of 20 patients with medulloblastoma treated at Childrens' Hospital and Medical Center, Seattle, from 1980 to 1987, were retrospectively analyzed with regard to prognosis. The overall actuarial 5-year survival rate was 63 %, with 57% of patients free from recurrence at 5 years. Residual contrast enhancement on CT scans taken immediately postoperatively was associated with a significantly decreased 5-year recurrence-free survival rate; the 5-year recurrence-free survival rate was 100% for those patients without post-operative residual enhancement compared to 41% for those patients with residual enhancement. A high risk group of patients with residual contrast enhancement persisting one year following diagnosis was identified. No patient in this group survived without disease progression. Other factors, including sex, age at diagnosis, evidence of tumor dissemination, or extent of surgical resection as reported by the neurosurgeon, did not significantly influence prognosis.
Introduction
During the last two decades the combination of surgical resection, craniospinal axis irradiation and perhaps adjuvant chemotherapy has significantly increased the survival rate of children with medulloblastoma, with recent studies indicating 5-year survival rates of approximately 60% [1-9]. Nonetheless, a proportion of patients continues to have tumor recurrence. Identification of those patients at higher risk of recurrence may allow early intensification of therapy, and hopefully, improved outcome. Several prognostic factors have been described for children with medulloblastoma, including sex of patient, age at diagnosis, evidence of tumor dissemination, degree of surgical resection, and the presence of contrast enhancing lesions on CT scans taken post-operatively [3~5, 8, 10-18]. This report considers the influence of these prog-
nostic factors on 5-year recurrence-flee survival in a group of pediatric patients with medulloblastoma, placing special emphasis on the presence or absence of contrast-enhancing lesions identifiable on sequential post-operative CT scans.
Materials and methods
From 1980 through 1987, twenty-seven consecutive children (less than 16 years old) with newly diagnosed medulloblastoma were treated at Children's Hospital and Medical Center (CHMC) in Seattle, WA. Of these, twenty-one patients had serial postoperative contrast-enhanced CT scans available for review. One patient expired three months after tumor excision as a result of respiratory failure not related to medulloblastoma and was excluded from the present study.
264 Of the twenty patients in the study group, five had gross total or partial resection of their primary tumor at another institution, while fifteen were operated upon in our institution.
Age and sex The sex distribution was 11 males and 9 females (male/female ratio of 1.2 : 1).
Postoperative myelogram records were available for all patients, which allowed accurate Chang Mstage assessment. Three patients were described intraoperatively as having gross nodular seeding in the cerebellar subarachnoid space (M2) and this was also identified on CT scan. Two additional patients were noted to have lumbar drop metastases on postoperative myelogram (M3). Thus, five patients were regarded as having evidence of disseminated disease (5/20, 25%) at the time of diagnosis.
Radiation treatment Surgical treatment Two patients were not irradiated immediately postoperatively because of their age (both were one year old). One of these patients subsequently received craniospinal irradiation, while the other patient received only adjuvant chemotherapy. Thus, 19 patients (90%) received postoperative radiation treatment. Treatment was standardized with regard to dose and volume irradiated. The majority of irradiated patients received 5400 cGy over six weeks to the primary site of tumor (range 50005500 cGy), while the neuroaxis doses ranged from 2400 cGy to 3600 cGy. Those patients with metastatic disease outside the posterior fossa received boost doses to those areas for a total dose of 5000 to 5400 cGy.
The primary tumor resection was total in fourteen patients (70%) and subtotal (50-99% resection) in six patients (30%). These estimates represented the opinion of the surgeon at the completion of the operative procedure rather than the post-operative CT findings. Eighteen out of 20 patients (90%) had a shunting procedure undertaken perioperatively. There was no surgical mortality in this series, nor were there any episodes of bacterial meningitis or postoperative complications other than V-P shunt obstructions in four of twenty patients.
CT scan evaluation Adjuvant chemotherapy Adjuvant chemotherapy was initially given to 19/20 (95%) patients. Fourteen patients received '8 Drugs in One Day' chemotherapy [19] and five patients received CCNU, Vincristine and prednisone.
Reports dicated by the consulting radiologist at the time of the contrast-enhanced CT scans were reviewed for description of abnormal enhancement. In addition, all CT scans were reevaluated by the authors. Observation of any abnormal residual enhancement, regardless of size, was noted. In all cases, this was at least adjacent to the primary site; in all patients with subarachnoid disease, subarachnoid enhancement was also noted.
Staging Staging of medulloblastoma has been based upon the Chang operative staging system [13]. Chang T-stage assessment was not possible in the present study due to incomplete or imprecise records of surgical assessment of tumor volume and location.
Statistical considerations Survival rates (S) and recurrence-free survival rates (RFS) were computed using the life table method [20]. The significance of variation (p value)
265 was determined using the Mantel-Haenszel procedure for statistical testing of the difference between survival curves [21] (values of p < 0.05 were considered statistically significant).
Results
Surgical treatment
Patients who underwent total resection as reported by the operating surgeon (N = 14) tended to relapse less frequently than those who underwent a lesser surgical procedure (N = 6), although this difference was not statistically significant: 5 year RFS of 71% versus 33% (p = 0.187).
Survival
Eleven patients are surviving with no evidence of recurrent disease 42 to 92 months from diagnosis (median 63 months), while nine patients have had progressive disease six to 92 months from diagnosis (median 25 months). Two and 5-year actuarial survival (S) rates for all patients (N = 20) are 95% and 63 %, respectively. Two and 5-year recurrence-free (RFS) survival rates for all patients (N = 20) are 90% and 57%, respectively.
Age and sex
Age at diagnosis did not significantly predict for outcome although there was a tendency for those diagnosed at 2 years old or less to relapse more often. Five patients were 2 years old or less at diagnosis and 15 patients were older than two years of age; five year RFS was 40% and 62%, respectively (p = 0.14). Girls (N = 9) tended to relapse more often than boys (N = 11) but this also was not significant: 5 year RFS 44% versus 66% (p = 0.13).
Dissemination
Patients who had evidence of tumor dissemination at the time of diagnosis (N = 5) tended to relapse more frequently than those had no evidence of tumor dissemination (N = 15), but this was also not a significant difference: 5 year RFS of 40% versus 62% ( p = 0.17).
Post resection C T First postoperative C T scan. All twenty patients had an evaluable contrast-enhanced CT scan taken in the initial postoperative period. This scan occurred during the interval between resection and the start of radiation treatment for all patients receiving radiation treatment. The median postoperative interval to this scan was 3 days (mean 4 days, range 1-10 days). Patients with a residual enhancing lesion on the first postoperative CT scan were significantly more likely to relapse than those without residual enhancement. A residual enhancing lesion was present in 14/20 (70%). Five year RFS was 41% (N = 14) for these patients compared with 100% (N = 6) when no residual was present (p = 0.035) (Fig. 1). Subsequent C T scans. Subsequent contrast-enhanced CT scans were evaluated at approximately 3-month intervals for all patients. In the subset of patients without a residual enhancing lesion on the first postoperative CT scan (N = 6), none were noted to develop enhancing lesions on subsequent CT scans taken during the entire follow-up period. In the subset of patients with a residual enhancing lesion on the first postoperative CT scan (N = 14), eight patients (8/14, 57%) were noted to have resolution of their enhancing lesions on CT scans taken during their first postoperative year. In this group, the median duration of enhancement (time until first CT scan without a residual enhancing lesion) was 197 days (mean 211 days, range 108-369 days). Of those patients with a residual enhancing lesion on the first postoperative CT scan which did not resolve (6/14, 43%), two patients had clinical
266
1
I
0
0
1
- Residual E.
~
50]
t.,
ol,
. 0
.
I00
................. I..,I.-.I.-..-I.-Ii
, 20
.
.
40
. p.=i;,.035.
- Residual ,..,...,...,\
L,,=ll
60
l
40
E. at 1 yr.
~-"-I-IIIII
80
: ~i.+ Re,?idual t.
,
............................
,
p = 0.0014
60 20
MONTHS
Fig. 1. Recurrence-free curve comparing patients with residual enhancement on first postoperative CT scan versus those with no residual enhancement on the first CT scan.
and radiographic tumor recurrence within the first year (175 and 256 days postoperatively). The remaining four patients had persistent residual enhancing lesions on all CT scans taken during the first year postoperatively and thereafter for the duration of their follow-up periods (Fig. 4, 5). All four of these patients eventually relapsed (16, 23, 28 and 37 months from diagnosis). There was a significant difference noted in 5-year RFS when comparing all patients surviving recurrence-free at the end of the first year with no residual enhancement on CT scan (14/18) versus those surviving recurrence-free at the end of the first year with persistent residual enhancement on CT scan at one year (4/18): Projected 5-year RFS in those patients without residual enhancement at one year was 82% versus 0% for those with sustained residual enhancement at one year (p = 0.00014) (Fig. 2).
Considering only those 12 patients with residual enhancement on the first postoperative CT scan who survived at least one year, those patients in whom residual enhancement resolved within one year (8/12) had a 72% 5-year RFS compared with a 0% 5-year survival among those patients (4/12) in whom residual enhancement persisted at one year (p = 0.000632) (Fig. 3).
Discussion
The overall survival rate of our patients, 63% at 5 years, is similar to other recent reports [1-8, 22]. In
~+ 0
I
0
•
•
I
20
'
•
Residual E. at I yr. I
40 MONTHS
'
"
I
60
'
8'O
Fig. 2. Recurrence-free survival curve comparing all patients without residual enhancement on CT at one year postoperatively ( . . . ) versus recurrence-free patients with residual enhancement on CT at one year postoperatively ( ).
these studies, the five year survival rates ranged from 53% [22] to 71% [5] (median 63%). A number of prognostic factors in medulloblastoma have been described. Regarding gender, several previous studies have identified females as having a more favorable prognosis than males [5, 12, 14, 15, 18, 22, 23] while many other studies have found no significant difference in recurrence-free survival on the basis of sex [3, 6, 11, 16, 17, 24, 25]. We found a tendency for females to relapse more often, but this was not statistically significant. In our experience, children aged 2 years or less at the time of diagnosis appeared to have a smaller chance of recurrence-free survival than did older patients, but this was not a statistically significant observation. This is in agreement with other studies which have also noted a tendency for younger children to relapse more frequently [2, 11, 16, 22]. Other investigators, however, have found no difference in recurrence-free survival on the basis of age at diagnosis [3, 5, 6]. It has been suggested that the higher rate of recurrent disease among younger children may be partially explained by a higher incidence of tumor dissemination at diagnosis in younger patients [10, 15, 16]. In our present study, three of the five patients 2 years old or younger had evidence of tumor dissemination at the time of resection (60%). In contrast, only two of the fifteen patients three years or older at diagnosis had evidence of tumor dissemination (13%).
267 100
Residue1 E, et 1 9r.
80 t~
+ t 1 \ ........Ill
&O
la.
tl
40
p-
20 0
\ |
0
•
•
I
20
•
r
000632
+Residual [ at I gr,
k T
40
•
•
T
60
•
•
I
80
HONTHS
Fig. 3. Recurrence-free survival curve comparing patients in whom residual enhancement resolved within one year ( . . . ) versus those with persistent enhancement greater than one year
(
).
The incidence of dissemination at diagnosis for all patients in the present series was 25% (5/20). This compares with a 16% to 46% incidence reported by others [10, 11, 15]. In our present study, patients with evidence of dissemination tended to relapse more frequently than those that had no evidence for dissemination, but this observation did not reach statistical significance. Most other studies have corroborated the poor prognosis associated with disseminated disease at diagnoses [3, 5, 6, S, 16]. Our data suggests a trend toward a more favorable outcome associated with complete tumor resection as observed by the operating neurosurgeon. Others have also reported a relationship between extent of surgical resection and outcome [11, 16, 22]. A recent study found complete resection to be the most significant favorable prognostic factor [5]. On the other hand, one study found no difference in outcome between those with complete resection or with partial resection [3]. In our study, while gross total resection was reported by the operating neurosurgeon in 14 cases, residual postoperative enhancement on postoperative CT scan was present in eight of these cases. Conversely, in all instances in which the postoperative CT scan demonstrated no residual enhancement, the surgical impression was that of gross total resection. Most studies reported to date have included many patients diagnosed prior to the CT era, in which the neurosurgical impression of ex-
Fig. 4. Postoperative residual enhancement.
tent of resection could not be confirmed. This may in part explain the discrepancy between studies in regard to the prognostic significance of extent of surgical resection. The only prognostic factor examined by our study which reached statistical significance in predicting 5-year RFS was the presence or absence of residual contrast enhancement on CT scans taken postoperatively. We examined this prognostic factor in a number of ways. First, we noted that the absence of residual contrast enhancement on the first postoperative CT scan significantly predicted for favorable outcome in this group of patients as compared with those patients with residual contrast enhancement on the initial postoperative scan. Residual enhancing lesions were present in 14/20 (70%). Five-year RFS was 41% (N = 14) for these patients compared with 100% (N = 6) when no residual enhancement was present (p = 0.035). Similar findings were recently reported by Jenkin [5] who found that of 28 patients who had an evaluable CT scan during the interval between resection and the start of radiation treatment (0-22 days), fifteen had a residual tumor mass (54%). Five-year RFS was 42% for these patients compared with 82% in those patients with no residual enhancing tumor present (p = 0.02) [5]. We found the presence of persistent residual
268
Fig. 5. Persistent enhancement 1 year following completion of radiotherapy. This patient went on to have locally progressive tumor.
contrast enhancement on CT scans taken one year or more postoperatively to be highly predictive of poor outcome. A persistent residual enhancing lesion was present on CT scans taken one year or greater in 4/18 (22%) progression-free patients. Five year RFS in those patients with sustained residual enhancement at one year was 0%, versus 82% in those without residual enhancement at one year (p = 0.00014). These findings were, once again, similar to those observed by Jenkins [5]. In that study, 9/47 (19%) had a measurable tumor mass on CT scans taken after completion of radiation treatment. The median interval from the end of radiation to the first post-treatment scan was 240 days (mean 385 days, minimum 5, maximum 2696 days) [5]. The 5-year RFS for the patients with a residual mass was 44% compared with 79% for patients with no residual disease (p = 0.004). In both our study and the Jenkins study, it was observed that many of the patients with evidence of residual disease on the first postoperative CT scan have resolution of this residual disease sometime after radiation therapy during the first year postoperatively. We determined that of those patients with evidence of residual enhancing lesions on the first postoperative CT scan, there was a significant-
ly worse recurrence-free survival in those patients who continued to have sustained residual enhancement on CT scans taken one year or more postoperatively. All four patients with sustained residual enhancement on CT scans taken one year or more postoperatively eventually relapsed, while only two of the eight patients who initial postoperative residual enhancement resolved within the first year had progression of disease. Controversy exists regarding the optimal time postoperatively to obtain a contrast enhanced scan to distinguish residual tumor from artifactual enhancement secondary to surgical trauma. In the dog brain, contrast enhancement first appears at day seven following resection [26]. However in a recent study evaluating contrast enhancement at the resection margin following lobectomy for epilepsy not associated with a tumor, postoperative CT enhancement was documented as early as the third day following surgery [27]. Cairncross and Wakai [28, 29] have demonstrated postoperative contrast enhancement within three to five days following tumor resection emphasizing the possibility of spurious enhancement at the resection cavity margin depending upon when the postoperative scan is taken. Usually, the artifactual enhancement from surgical trauma appears as a thin rim of enhancement as opposed to nodular enhancement. While the typical artifactual 'rim enhancement' was not observed in any of our patients, we cannot completely exclude the possibility that resolving postoperative enhancement represented surgical artifact in some cases. In those patients with postoperative enhancement which subsequently resolved, the prognosis appears to be relatively favorable, whether this reflects resolution of artifactual enhancement or complete tumor response to therapy. On the other hand, those patients who did not have resolution of their enhancing mass by one year following diagnosis all had tumor progression at the site of enhancement, arguing strongly that persistent residual enhancement represented tumor. Of those patients with persistent enhancement one year following diagnosis, one patient, who had been treated only with chemotherapy because of young age at diagnosis, underwent a second sur-
269 gery at the time of disease progression. Histopathologic examination of resected tissue was consistent with ganglioneuroblastoma; this patient has received no further therapy and remains free of tumor two years following her second surgical procedure. Whether the remaining three patients would have benefited from early second-look surgery prior to disease progression and perhaps intensification of adjuvant therapy remains an unanswered but critical question. This series of patients with medulloblastoma suggests that persistence of postoperative enhancement on CT scan is associated with a high probability of tumor progression. The significance of this finding, as well as its therapeutic implications, should be evaluated in a large multi-institutional trial.
6.
7.
8.
9.
10.
11.
Acknowledgements 12.
This study was supported in part by NIH K08 Clinical Investigator Award No. CA01451-02 under J. Russell Geyer, M.D. and NIH K08 Clinical Investigator Award No. NS01253-03 under Mitchel Berger, M.D.
References
13.
14.
15. 16.
1. Bellani FF, Gasparini M, Lombardi F, Zucali R, Lucarelli G, Migliavacca F, Moise S, Nicola G: Medulloblastoma. Results of a sequential combined treatment. Cancer 54: 1956-61, 1984 2. Brand WN, Schneider PA, Tokars RP: Long term results for a pilot study of low dose cranial-spinal irradiation for cerebellar medulloblastoma. Int J Radiat Oncol Biol Phys 13: 1641-1645, 1987 3. Caputy AJ, McCullough CC, Manz HJ, Patterson K, Hammock MK: A review of the factors influencing the prognosis of medulloblastoma. J Neurosurg 66: 88--87, 1987 4. Hughes EN, Shillito J, Sallan SE, Loeffler JS, Cassady JR, Tarbell N J: Medulloblastoma at the Joint Center for Radiation Therapy between 1968 and 1984. The influence of radiation dose on the patterns of failure and survival. Cancer 61: 1992-1998, 1988 5. Jenkin MB, Goddard K, Armstrong D, Becker L, Berry M, Chan H, Doherty M, Greenberg M, Hendrick B, Hoffman H, Humphreys R, Sonley M, Weitzman S, Zipursky A: Posterior fossa medulloblastoma in childhood: Treatment
17.
18.
19.
20.
Results and a proposal for a new staging system. Int J Rad Oncology Bio! Phys 19: 265-274, 1990 Kopelson G, Linggood RM, Keinman GM: Medulloblastoma. The identification of prognostic subgroups and implications for multimodality therapy. Cancer 51: 312-319, 1983 Levin VA, Rodriguez LA, Edwards MSBWW, Liu HC, Fulton D, Davis RL, Wilson CB, Silver P: Treatment of medulloblastoma with procarbazine, hydroxyurea, and reduced radiation doses to whole brain and spine. J Neurosurg 68: 383--387, 1988 Packer RJ, Sutton LN, Rorke LB, Littman PA, Sposto R, Rosenstock JG, Bruce DA, Shut L: Prognostic importance of cellular differentiation in medulloblastoma of childhood. J Neurosurg 61: 296-301, 1984 Stiller CA, Lennox EL: Childhood medulloblastoma in Britain 1971-77: an analysis of treatment and survival. Br J Cancer 48: 835-841, 1983 Allen JC, Epstein F: Medulloblastoma and other primary malignant neuroectodermal tumors of the CNS - the effect of patients' age and extent of disease on prognosis. J Neurosurg 57: 446-451, 1982 Berry MP, Jenkin RDT, Keen CW, Nair BD, Simpson WJ: Radiation treatment of medulloblastoma. A 21 year review. J Neurosurg 55: 43-51, 1981 Bloom HJG: Medulloblastoma: prognosis and prospects. Int J Radiat Oncol Biol Phys 2: 1031-1033, 1977 Chang CH, Housepian EM, Herbert C: An operative staging system and megavoltage radiotherapeutic technic for cerebellar medulloblastomas. Radiology 107: 359-362, 1973 Chatty EM, Earle KM: Medulloblastoma: A report of 201 cases with emphasis on the relationships of histologic variants to survival. Cancer 28: 977-983, 1971 Deutsch M: Medulloblastoma-staging and treatment outcome. Int J Radiat Oncol Biol Phys 14: 1103-1107, 1988 Evans AE, Jenkin RDT, Sposto R, Ortega JA, Wilson CB, Wara W, Ertel IJ, Kramer S, Chang CH, Leiken SL, Hammond GD: The treatment of medulloblastoma. The results of a prospective randomized trial of radiation therapy with and without chloroethyl-cyclohexyl nitrosurea, vincristine and prednisone. J Neurosurg 72:572-582,-1990 Harisiadis L, Chang CH: Medulloblastoma in children: A correlation between staging and results of treatment. Int J Radiat Oncol Biol Phys 2: 833-841, 1977 MacFarland DR, Horwitz H, Saenger EL, Bahr GK: Medulloblastoma: A review of prognosis and survival. Br J Radiol 42: 198--214, 1969 Pendergrass TW, Milstein JM, Geyer JR, Mulne AF, Kosnik EJ, Morris JD, Heideman RL, Ruyrnann FB, Stuntz JT, Bleyer WA: Eight drugs in one day chemotherapy for brain tumors: Experience in 107 children and rationale for preradiation chemotherapy. J Clin Oncol 5(8): 1221-1281, 1987 Cutler SJ, Ederer F: Maximum utilization of the life table method in analyzing survival. J Chronic Dis 8: 699-712, 1958
270 21. MantelN, HaenszelW: Statistical aspects of the analysis of data from retrospective studies of disease. JNCI 22: 719748, 1959 22. Tait DM, Thornton-Jones H, Bloom HJG, Lemerle J, Morris-Jones P: Adjuvant chemotherapy for medulloblastoma: the first multi-centre control trial of the international society of Paediatric Oncology (SIOPI) Eur J Cancer 26: 464469, 1990 23. Hope-StoneH: Results of treatment of medulloblastoma. J Neurosurg 32: 83-88, 1970 24. Landberg TG, Lindgren ML, Cavallin-Stahl EK: Improvements in the radiotherapy of medulloblastoma 1946-1975. Cancer 45: 670-678, 1980 25. Quest DO, Brisman R, Antunes JL, Housepian EM: Period of risk for recurrence in medulloblastoma. J Neurosurg 48: 159-163, 1978 26. Jeffries BF;Kishore PRS, Singh KS, Ghatak NR, Krempa
J: Contrast enhancement in the postoperative brain. Radiology 139: 409413, 1981 27. Laohaprasit V, Silbergeld DL, Ojemann GA, Eskridge JM, Winn HR: Postoperative CT contrast enhancement following lobectomy for epilepsy. J Neurosurg 73: 392-395, 1990 28. Cairncross JG, Pexman JW, Rathbone MP, DelMaestro RF: Postoperative contrast enhancement in patients with brain tumor. Ann Neurol 17: 570-572, 1985 29. Waikai S, Andoh Y, Ochiai C, Inoh S, Nagai M: Postoperative contrast enhancement in brain tumors and intracerebral hematomas: CT study. J Comput Assist Tomograph 14: 267-271, 1990
Address for offprints: R. Geyer, Children's Hospital & Med Center, Hematology Dept. 4800 Sand Point Way NE, Seattle, WA 98105, USA