Late Recurrence of Primitive Neuroectodermal Tumor/Medulloblastoma

ILENE 6. LEFKOWITZ, MD,” ROGER J. PACKER, MD,*vts* STEPHEN G. RYAN, MD,$ NARlN SHAH, MD,§ JANE ALAVI, MD,\( LUCY 8. RORKE, MD,**T LESLIE N. SUTTON, ME),”*# AND LUIS SCHUT, MD’*#

The period of risk for recurrence of primitive neuroectodermal tumor/medulloblastoma (PNET/MB) is not clearly defined. With current treatment since more than 50% of children with PNET/MB can be expected to survive for at least 5 years after diagnosis, determining the evidence of “late” recurrence is of increasing concern. Collins has stated that patients with embryonal tumors who survive, disease free, for a period of time equal to the age at diagnosis plus 9 months can be declared cured. This, so-called Collins’ law has been applied to patients with PNET/MB. To determine the incidence of “late” recur- rence, factors which impact on recurrence and applicability of Collins’ law, the authors studied all patients diagnosed with PNET/MB at the Children’s Hospital of Philadelphia, Hospital of the Uni- versity of Pennsylvania, Philadelphia, and Geisinger Medical Center, Danville, Pennsylvania, between 1970 and 1984. For the 44 patients in this study, the disease-free survival at 5, 10, and 12 years was 54%,41% and 30%, respectively. For children surviving 5 years, the actuarial survival at 10 years was 75% and at 12 years, 51%. Age, sex, dose of radiotherapy, chemotherapy, or extent of surgery were not predictive of late relapse. Recurrence in three of seven patients (43%) occurred outside the ‘‘period of risk” as predicted by Collins. It appears that the “period of risk” for recurrent central nervous system tumors after PNET/MB is as yet undefined and probably indefinite.

Cancer 62826430,1988.

HE PRIMITIVE neuroectodermal tumor/medullo- T blastoma (PNET/MB) is an embryonal neoplasm which accounts for approximately 20% of pediatric brain tumors.’ It is most commonly found in the cere- bellar vermis, where it is also known as the “medullo- blastoma.”

Controversy concerning the cell of origin and there- fore the nomenclature, of similarly appearing tumors in other central nervous system locations exists. In a recent review, it was suggested that until the cell of origin of these tumors can be better established and specific markers can be identified, that all such tumors, indepen- dent of their location whether the neuro-axis, be classi- fied as “primitive” neuroectodermal tumors (PNET/

Presented in part at the Child Neurology Society, Boston, October 1986.

From the ‘Neuro-Oncology Program, Children’s Hospital of Phila- delphia; Departments of ?Pediatrics, $Neurology, lINeuro-pathology, #Neurosurgery, and I( Medicine, University of Pennsylvania, Philadel- phia; and Weisinger Medical Center, Danville, Pennsylvania.

Supported in part by the Forerder Fund for Excellence, Neuro-On-

The authors thank Linda Cella for secretarial assistance. Address for reprints: Roger J. Packer, MD, Division of Neurology,

Children’s Hospital of Philadelphia, 34th & Civic Center Boulevard, Philadelphia, PA 19 104.

cology .

Accepted for publication January 22, 1988.

MB).’ The dismal prognosis formerly associated with PNET/MB has improved considerably with improve- ments in surgery and postoperative care and the intro- duction of local and craniospinal radiation the rap^.^,^ More recently, the use of adjuvant chemotherapy in certain high-risk patients appears to have further en- hanced ~urvival.~ With current therapy, as high as 50% to 70% of patients with PNET/MB can be expected to survive, disease free, for 5 years after De- spite these encouraging trends, sporadic late recurrences of PNET/MB have been well documented both in indi- vidual case reports and in several longitudinal stud- ies?,6,’o*’2,16 Some authors have contended that patients who survive beyond the period of time of predicted re- lapse for embryonic tumors described by Collins (Col- lins’ law) have a high likelihood of c ~ r e . ’ ~ , ~ ’ , ’ ~

We examined the records of patients at our institu- tions who survived at least 5 years after the diagnosis of PNET/MB in order to determine more precisely the incidence of late recurrence and to identify any clinical and histologic features which might be associated with late relapse.

Materials and Methods The medical records of all patients at The

Children’s Hospital of Philadelphia, Geisinger Medical

826

No. 4 RECURRENT PNET Lefkowitz et a[. 827

TABLE 1. Clinical-Histoloeic Features of Patients With Late Relaase

Initial treatment Relapse

Age at Initial Rad, local Rad, CS Chemo- Time after Histologic diagnosis Sex Tumor site histologic type Surgery (rad) (rad) therapy diagnosis Site type

5.5 yr M PF

12yr, 1 1 mo F PF

5 Yr F PF

8yr ,7mo M Thalbrain- stem

10yr,2mo F PF

29 yr M PF

14 yr M PF

PNET, with Total

PNET,with Total

PNET, Undiff Subtotal

PNET, with Subtotal

glial Diff resection

glial Diff resection

resection

neuronal resection Diff

PNET, Undiff Subtotal resection

PNET Subtotal

PNET Subtotal resection

resection

5140

5000

5000

5200

5500

5250

5400

4140

4000

3600 (to C,)

4000

4000

4000

3200

Yes

Yes

No

Yes

No

Yes

No

8yr ,5mo PF Anaplastic

6 yr, 4 mo PF and PNET, Undiff

8.5 yr PF and PNET, Undiff

8 yr, 5 mo Thalamus; PNET,

glioma

LMS

LMS

LMS neuronal and glial

12yr,5 mo PF PNET with anaplastic glioma

1 1 yr PF PNET

5.5 yr PFand PNET LMS

PF: posterior fossa; LMS leptomeningeal spread; Thal: thalamus; Undiff undifferentiated; CS: craniospinal, PNET: primitive neuroec-

Center, Danville, Pennsylvania, and the Hospital of the University of Pennsylvania, Philadelphia, who were treated for PNET/MB between January 1970 and Au- gust 198 1 were examined. Over the time period, 44 pa- tients with PNET/MB were diagnosed and 24 were alive and in continuous remission for 5 years since initial therapy.

The 24 patients who survived 5 years without evi- dence of disease were a median of 9 years at the time of diagnosis (range, 5-29 years). They have been followed for a median of 8 years, 1 1 months (range, 5-1 5 years). All patients were seen yearly for a neurologic examina- tion and contrasted computerized tomography (CT) scan. However, magnetic resonance imaging (MRI) was not routinely performed in these patients.

Patients were treated in a standardized fashion. Gross total surgical resection was attempted in all patients. Routine preoperative ventriculoperitoneal shunting was not performed; permanent cerebrospinal fluid drainage was utilized postoperatively only if there was persistent hydrocephalus. All patients were treated with cranio- spinal radiotherapy supplemented with local radiother- apy to the primary site. Patients received a median dose of 3600 rad to the craniospinal axis and a median dose of 5200 rad to the tumor. Eight patients also received adjuvant chemotherapy (CCNU and vincristine).

Where possible, histologic specimens from the origi- nal, and, if applicable, recurrent tumors were examined by one of us (L.B.R.). Clinical and pathologic data then were analyzed to identify any factors which might be predictive of late recurrence. Parameters which were

todermal tumor; Diff differentiated; Rad radiation.

analyzed included sex, age at diagnosis, extent of resec- tion, extent of disease at diagnosis (T and M stage as suggested by Chang et aZ.19), tumor histologic type, the dose of cranial and local radiotherapy, and the use of chemotherapy.

Results Of the 24 long-term survivors, nine had recurrent

central nervous system (CNS) mass lesions at the site of their original tumor. One of these patients was found to have an asymptomatic posterior fossa mass lesion at the primary site on routine follow-up computed tomogra- phy (CT) 5.5 years after his original tumor was diag- nosed; this patient has refused further treatment. A sec- ond patient, who was asymptomatic, had a mass lesion seen on CT, performed as a surveillance study 10 years after initial diagnosis. Radionecrosis was found at reop- eration.

The remaining seven patients with recurrent mass le- sions all had histologically confirmed tumor recurrence, are described in Table 1 . All had symptoms of recurrent posterior fossa mass at the time of detection of their recurrent lesions. The patients with recurrent tumors were a median 8 years 7 months of age at the time of original diagnosis. Four were male and three female.

Recurrence occurred at a median of 8 years 5 months from diagnosis. The actuarial disease-free survival rate for our entire population 5 years after diagnosis was 54%, but fell to 41% at 10 years and 30% at 12 years (Fig. 1). For patients surviving 5 years, the predicted rate of disease-free survival at 10 years was 75% and at 12 years

828 CANCER August 15 1988 Vol. 62

0

-l

C

0 40 80 120 160 TIME (months)

FIG. 1 . Actuarial survival from diagnosis for 44 patients with PNETIMB.

was 5 1 %. Three of seven patients developed recurrence at a period of time of greater than their original age at diagnosis plus 9 months. This occurred 1.3 years, 2.3 years and 2.2 years after this period of time, respectively.

There were no significant differences between those 5-year survivors who relapsed and those who did not with respect to age at diagnosis, sex, dose of radiation, extent of surgery, treatment with chemotherapy, or pres- ence of differentiation on original tumor histologic study (Table 2).

At the time of recurrence, one patient had an ana- plastic glioma and two others had tumors with mixed undifferentiated and glial elements. The original tumor

TABLE 2. 5-Year Disease-Free Survivors: Clinical Comparison of Late Relapsers Versus Continuous Responders

Median age at diagnosis

Sex (male/female) Surgical resection

Radiation

(range)

(total/subtotal)

Local (median) Craniospinal

(median) Chemotherapy (yes/no)

Late relapsers (N = 7)

9.5 yr (5-29 yr) 413 215

5200 rad 4000 rad

413

Continuous responders (N = 20)

8 yr, 10 mo (5-14 Yr) loll0 7/13

5 140 rad 3600 rad

911 1

TABLE 3. Reported 5-Year and 10-Year Survival Rates for Patients With Primitive Neuroectodermal Tumor/Medulloblastoma

No. of % 5-yr % IO-yr Reference Yr vatients survival survival

Hershatter ef al.” Farwell et al. ’’ Park ef al. ‘ ’ Berry et aL6 Chin and

Maruyamag Harisiadis and

ChangI4 Mealy and Hall4 Broadbent et ~ 1 . ’ ~ King and

Sagerman” Bloom ef aL3 Aroni2 Lefkowitz et al.

(current series)

1940- 1983 1935-1979 1950- 1980 1958- 1978

1964-198 1

1963-1975 1953-1973 1940-1976

1960- I972 1950-1964 1946-1968

1970- 1983

127 33% 21% 143 22% 13% 144 47% (k4.8) 42% (f5.1) 122 56% 43%

23 35% 35%

58 40% 31% 45 41% 22% 31 29% 29%

24 38% 0% 82 32% 26% 14 46% 35%

44 54% 41%

in two of these patients had dial differentiation. One other patient, whose original tumor was differentiated, had an undifferentiated tumor at the time of relapse. The remaining three patients had undifferentiated tumors at diagnosis and relapse (Table 1).

Six of the seven patients with recurrent tumor have died; their median postrecurrence survival was 7 months. All tumors recurred at the primary site; in four patients this was associated with concurrent leptomen- ingeal dissemination.

Discussion

There are a large number of series reporting long-term survivors of PNET/MB, and they have shown a variable decline in survival due to recurrent tumor between 5 and 10 years (Table 3). Although it was recognized as early as 1953 that craniospinal radiation had a major role in treating PNET/MB, standardization of doses and methods of delivery, as well as the use of megavoltage rather than orthovoltage irradiation, is relatively new. This creates difficulty in comparing different series and drawing conclusions from any one series.

King and Sagerman,” in a series published in the American Journal of Roentgenology in 1975, reported 24 patients treated with radiation alone during the years 1960 to 1972. The 38% 5-year survival in this study is comparable to other series from this time, but the 10- year survival approaches 0%. However, only three chil- dren in this series were alive 6 years after diagnosis, so that this marked decline in survival takes on less signifi- cance. Other series report less dramatic but definite in- creasing mortality after 5 years. The largest reported series, 144 patients seen at the Hospital for Sick Chil- dren in Toronto, over a 30-year period, had a 5-year

No. 4 RECURRENT PNET

survival of 47% declining to 42% by 10 years.” Hershat- ter et al., in a report of 127 patients treated with varying amounts and volumes of radiotherapy between 1940 and 1983 found an actuarial survival at 5 years of 33%, which fell to 21% by 10 years.” Mealy and Hall4 re- ported “late” relapse, 3 years after treatment, in seven of 40 patients with PNET/MB. However, those patients who died greater than 5 years after diagnosis had re- ceived reduced dosages of radiotherapy and some had relapsed years before their death.

In contrast to these series reporting some decline in survival after 5 years, Broadbent et al.” and Chin and Maruyama’ found no patients who had their initial re- lapse 5 years after diagnosis. So, in conclusion, the inci- dence of relapse greater than 5 years after diagnosis in the series reported to date is difficult to determine. It is unclear whether late relapse is due to what would now be considered less than optimal therapy. It is also un- clear whether many of the “late” deaths were in children who had initially had disease recurrence within 5 years of diagnosis, but died greater than 5 years after diag- nosis, since distinction between survival and disease-free survival was not made in these series. In our series, using in all patients what is presently accepted as necessary doses of radiotherapy, we found a decline in survival between 5 and 10 years after treatment and no plateau to the survival curve (Fig. 1).

Collins’ law also has been used to assess the probabil- ity of “cure” for patients with PNET/MB. Collins’ law states that disease-free survival of greater than age at diagnosis plus 9 months is equivalent to cure.2’s22 Al- though this law was first applied to Wilm’s tumor, an embryonal kidney neoplasm, its proponents have stated it is appropriate to look at all embryonal tumors in this fashion.23 This implies that all primitive tumors have an intrinsic, unalterable growth rate. After “definitive” treatment, if any malignant cells remain, it will take the same amount of time to become clinically apparent as it did originally; assuming that the primitive neoplastic (or preneoplastic) cells were present from embryogenesis. The PNET/MB is the most common embryonal tumor of the central nervous system.

It is clear, in our series, that recurrent PNET/MB did not obey Collins’ law in three patients. In two patients relapse occurred 2 years outside the period of risk. Mealy and Hall suggested up to 10% of PNET/MB sur- vivors will develop recurrent PNET/MB or other brain neoplasms after the period of risk.4 Park et al. from Toronto Children’s Hospital have reported two relapses of 28 patients (7.1%) past the period of risk; one 9 months (1.4 X risk) and the other 2 years and 5 months (1.5 X risk) later.” A review of the literature reveals 15 additional cases of recurrent brain tumor in patients past the period of predicted r i ~ k . ’ ~ , ~ ~ Multiples of the risk

. Lefkowitz et al. 829

period range from 1.1 to 6.0. There is a tendency in these reported cases for recurrent PNET/MB to occur within 2 years of the risk period whereas other tumor types appear later.

Recurrence of PNET/MB within a short time (less than 2 years) outside the period of risk has been ex- plained by some authors as reflecting altered cell ki- netics, presumably due to previous radiation therapy.I7 It can also be speculated that the first tumor was com- pletely erradicated but that other cells in the area, with other growth kinetics predestined to become PNETI MB, began proliferating and produced a second PNET/ MB. This theory can be possibly evoked in patients with a complete resection and adequate craniospinal irradia- tion.

Brain tumors which occur many years past the period of risk are often of a somewhat different histology.24 The PNET/MB has been demonstrated to undergo spon- gioblastic dedifferentiation both in tissue culture and clinically in the direction of a s t r o c y t ~ m a . ~ ~ ? ~ ~ Alterna- tively, there is evidence that radiation to the central ner- vous system can produce tumors. Low-dose radiation (less than 1000 rad) used for the treatment of tinea ca- pitis has been associated with meningiomas and sar- c o m a ~ . ~ ~ Several researchers have reported radiation in- duced astrocytoma in nonhuman primate^^'-^' as well as humans after high-dose radiation for meningioma, ~raniopharyngiorna’~,~~,~~ and malignancies of the head and neck.33

It is interesting that two of the three tumors in our series which recurred outside the “period of risk” as predicted by Collins’ law, had foci of glioma within them. In the tumor which relapsed the longest (2.3 years after the period of risk) no areas of PNET could be identified; the tumor was composed of anaplastic glioma. Since, in these two cases the original tumor had areas thought to represent glioma, these lesions cannot be truly considered second tumors. Making distinctions between recurrent and secondary tumors even more dif- ficult is determining exactly what these foci of glioma- tous appearing tissue on light microscopic examination of the original specimens represent. At the time of origi- nal diagnosis, glial specific markers (such as glial fibril- lary acidic protein [GFAP]) were not available and even if glial-specific staining were not positive, that would not definitively prove that areas of anaplastic glioma were not present.34 Our current experience, as well as a review of the literature, leads us to favor the hypothesis of radia- tion induction of these recurrent tumors in some cases, especially when the second tumor is primarily a glioma.

Thus, in our experience, the period of risk for recur- rence of the primary tumor or possibly of development of secondary tumor in patients with PNET/MB appears to be as yet undefined and probably indefinite. This has

8 30 CANCER August 15 1988 Vol. 62

obvious connotations as regards counseling patients and their families, and determining how long surveillance studies need to be done for detection of tumor redevel- opment at the primary site. Currently, we are following all long-term survivors with yearly MRI scans; although all our patients in this series with recurrent tumors be- came symptomatic before detection by neuroimaging. Since other lesions, such as delayed radionecrosis, may mimic recurrence, it seems that surgical re-exploration is indicated in those patients who develop new posterior fossa masses. However, these recommendations are tempered by the fact that presently available treatment for these “recurrent” neoplasms usually is ineffective, with most patients dying of disease within 1 year of diagnosis.

REFERENCES

1. Young JL, Miller W. Incidence of malignant tumors in U. S. children. J Pediatr 1975; 86:254-258.

2. Rorke LB, Gilles FH, Davis RL, Becker LE. Revision of the World Health Organization classification of brain tumors for child- hood brain tumors. Cancer 1985; 56:1869-1886.

3. Bloom HJG, Walker EWK, Henk JM. The treatment and prog- nosis of medulloblastoma in children: A study of 82 verified cases. Am J Roentgenol Radium Ther Nucl Med 1969; 105:43-62.

4. Mealy J, Hall PV. Medulloblastoma in children. J Neurosurg

5. Allen JC. Childhood brain tumors: Current status of clinical trials in newly diagnosed and recurrent disease. Pediatr Clin North Am 1985;

6 . Berry MP, Jenkin RDT, Keen CW et al. Radiation treatment for medulloblastoma: A 21 year review. J Neurosurg 1981; 55:43-5 1 .

7. Bloom HJG. Intracranial tumors response and resistance to ther- apeutic endeavors, 1970-1980. Int J Radiat Oncol Biol Phys 1982;

8 . Caputy AJ, McCullough DC, Manz HJ el al. A review of the factors influencing prognosis of medulloblastoma. J Neurosurg 1987;

9. Chin HW, Maruyama Y. Early response and long-term results in the radiotherapy of childhood medulloblastoma. J Neuro-Oncoll983; 153-59.

10. Hershatter BW, Halperin EC, Cox EB. Medulloblastoma: The Duke Medical Center experience. Int J Radiat Oncol Biol Phys 1986;

1 1 . Park TS, Hoffman HJ, Hendrick EB, et al. Medulloblastoma: Clinical presentation and management. J Neurosurg 1983; 58543- 552.

1977; 4656-64.

32:633-65 1 .

8~1083-1113.

66~80-87.

12: 177 1-1777.

12. Aron BS. Medulloblastoma in children: Twenty-two years expe- rience with radiation therapy. Am JDis Child 1971; 121:314-317.

13. Farwell JR, Dohromann GJ, Flannery JT. Medulloblastoma in childhood An epidemologic study. J Neurosurg 1984; 61:657-664.

14. Harisiadis L, Chang CH. Medulloblastoma in children: A cor- relation between staging and results of treatment. Int J Radiat Oncol Biol Phys 1977; 2:833-841.

15. King GA, Sagerman RH. Late recurrence of medulloblastoma. Am J Roentgenol 1975; 123:7-12.

16. Kosmaki S, Kosmaki R, Choi H. Radiation and drug-induced intracranial neoplasms with angiographic demonstration. Neurol Me- dico-Cherurg 1977; 1755-62.

17. Latchaw JP, Hahn JF, Moylan DJ et al. Medulloblastoma: Pe- riod ofrisk reviewed. Cancer 1985; 55:186-189.

18. Quest DO, Besman R, Antunes JL, Housepian EM. Period of risk for recurrence in medulloblastoma. J Neurosurg 1978; 48: 159- 163.

19. Chang CH, Houseplan EM, Herbert C. An operative staging system and a megavoltage radiotherapeutic technique for cerebellar medulloblastoma. Radiology 1969; 93: 1351-1359.

20. Broadbent VA, Barnes ND, Wheeler TK. Medulloblastoma in childhood Long-term results of treatment. Cancer 198 1; 48:26-30.

21. Collins VP. Wilm’s tumor: Its behavior and prognosis. J La State Med Soc 1955; 107:474-480.

22. Collins VP. The treatment of Wilm’s tumor. Cancer 1958; 11:89-94.

23. Shuman RM, Alvord EC, Leech RW. The biology of childhood ependymomas. Arch Neuroll975; 32:731-739.

24. Dell S, Cohen M, Kushner M. Period of risk hypothesis after prophylactic radiotherapy. Childs Brain 198 1; 8:372-38 1 .

25. Pearl GS, Mira SS, Miles ML. Glioblastoma multiform occur- ring 13 years after treatment of medulloblastoma. Neurosurgery 1980;

26. Raimondi AJ, Tomita T. Medulloblastoma in childhood. Acfa Neurochir 1979; 50:127-138.

27. Modan B, Baidatz D, Mart H et al. Radiation-induced head and neck tumors. Lancet 1974; 1:277-279.

28. Haymaker W, Rubinstein LJ, Miquel J. Brain tumors in irra- diated monkeys. Acta Neuropathol (Berl) 1972; 20267-277.

29. Kent SP, Pickering JE. Neoplasms in monkeys (macaca mu- latta): Spontaneous and irradiation induced. Cancer 1958; 1 1: 138- 147.

30. Krupp JH. Nine-year mortality experience in proton exposed macaca mulatta. Radiat Res 1976; 67:244-25 1 .

3 1. Jones A. Supervoltage x-ray therapy of intracranial tumors. Ann R CollSurg Engl 1960; 27:310-354.

32. SOB RL, Donaldson SS, York CH. Malignant astrocytoma fol- lowing radiation therapy of a craniopharyngioma. J Neurosurg 1978;

33. Marus GM, Levin CV, Rutherford GS. Malignant glioma fol- lowing radiotherapy for unrelated primary tumors. Cancer 1986;

34. Bonnin JM, Rubinstein LJ. Immunohistochemistry of central nervous system tumors: Its contribution to neurosurgical diagnosis. J Neurosurg 1984; 60:1121-1133.

6:546-55 1 .

48:622-627.

58:886-894.