Combined Modality Therapy for Primary CNS Lymphoma By Lisa M. DeAngelis, Joachim Yahalom, Howard T. Thaler, and Uma Kher Purpose: Primary CNS lymphoma (PCNSL), formerly rare, is being seen with increased frequency among apparently immunocompetent patients. Conventional treatment has consisted of whole-brain radiotherapy (RT) and corticosteroids, with a median survival of 15 to 18 months and a 3% to 4% 5-year survival. Chemotherapy has been useful in the treatment of recurrent PCNSL. In 1985 we began a treatment protocol using chemotherapy and cranial irradiation for the initial therapy of non-AIDS PCNSL. Patients and Methods: Thirty-one patients (group A) completed the combined modality regimen. All had placement of an Ommaya reservoir and received pre-RT systemic methotrexate, 1 g/m 2 , plus six doses of intraOmmaya methotrexate at 12 mg per dose. A full course of cranial RT (4,000-cGy whole-brain RT plus a 1,440-cGy boost) was followed by two cycles of high-dose cytarabine (ara-C), with each course consisting of two doses of 3 g/m 2 ara-C separated by 24 hours and infused over 3 hours. During this period, 16 additional patients (group

PRIMARY

CNS lymphoma (PCNSL) is an aggressive non-Hodgkin's lymphoma arising within the brain, spinal cord, or leptomeninges. It is often associated with acquired or congenital immunosuppression, but in the past 15 years, its incidence has risen threefold among apparently immunocompetent individuals.' The reason for this increased incidence is unknown. Because the tumor was so rare, large, prospective, therapeutic trials were not possible until recently. Conventional treatment consisted of cranial irradiation and corticosteroids. PCNSL is extremely sensitive to initial treatment in most patients, producing a rapid complete response (CR). Nevertheless, survival is short due to tumor recurrence in greater than 90% of patients within the first year, with a median survival of 12 to 18 months and a 5-year survival of 3% to 4%.2-8 The brain is the primary site of recurrence, but unlike gliomas, relapse in the brain may be at a site distant from the original tumor. Clinical leptomeningeal relapse is common. The leptomeninges are involved pathologically at autopsy in 100% of patients, a result of the periventricular location of most PCNSL lesions.' The eye, an extension of the nervous system, can also be a site of disease either at diagnosis or at relapse. 10 " Systemic lymphoma is apparent in only 7% to 8% of autopsied patients, and the majority of these patients have a single focus of clinically silent disease (eg, cervical lymph node or small pulmonary nodule) that is probably a systemic metastasis from recurrent nervous system tumor.4 ' 6 There is no reported case of occult systemic lymphoma presenting as a CNS mass lesion.

R) were treated with RT alone, either because patients refused chemotherapy or RT was initiated before our consultation; all would have been eligible to participate in the protocol. Follow-up extended through April 1, 1991. Results: Group A had a significantly prolonged time to recurrence (median, 41 months) compared with group R (median, 10 months; P = .003). Although median survival was doubled from 21.7 months for group Rto 42.5 months for group A, this was not statistically significant because of small sample size. More importantly, group R patients received systemic chemotherapy for recurrent PCNSL, which improved survival. Conclusion: The addition of chemotherapy to cranial RT for initial treatment of PCNSL significantly improved disease-free survival and contributed to overall survival; all non-AIDS patients with newly diagnosed PCNSL should be considered for combined modality therapy. J Clin Oncol 10:635-643. © 1992 by American Society of Clinical Oncology.

In 1985, we designed a prospective treatment program for non-AIDS patients with PCNSL using preradiation chemotherapy and cranial radiotherapy (RT). Before then, chemotherapy had been used in isolated patients with recurrent disease, and there was suggestive evidence that a combined modality approach as part of initial therapy prolonged survival.12" 5 We chose drugs that had demonstrated efficacy in PCNSL and that can penetrate the blood-brain barrier, and chose to treat the leptomeninges vigorously in all patients. This comprehensive regimen significantly prolonged disease-free survival. PATIENTS AND METHODS From 1985 through January 1991, all non-AIDS patients with newly diagnosed PCNSL at Memorial Sloan-Kettering Cancer Center (MSKCC) were eligible for treatment with chemotherapy and cranial irradiation. All patients underwent a staging evaluation that included a lumbar puncture, ophthalmologic examination including slit-lamp, abdominal computed tomographic (CT) scan, bone marrow biopsy, chest x-ray, and human immunodeficiency

From the Departmentsof Neurology and Radiation Oncology, and Division of Biostatistics, Memorial Sloan-Kettering Cancer Center, New York; and Department of Neurology and Neuroscience, Cornell University Medical College, New York, NY. SubmittedJune 10, 1991; acceptedNovember 1, 1991. Presented in part at the Annual Meeting of the American Society of ClinicalOncology, Houston, TX, May 1991. Address reprint requests to Lisa M. DeAngelis, MD, Memorial Sloan-KetteringCancerCenter, 1275 York Ave, New York, NY 10021. © 1992 by American Society of Clinical Oncology. 0732-183X/92/1004-0021$3.00/0

Journalof Clinical Oncology, Vol 10, No 4 (April), 1992: pp 635-643

Downloaded from ascopubs.org by EAST CAROLINA UNIVERSITY on August 4, 2019 from 150.216.068.200 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.

635

636

DEANGELIS ET AL

virus-type 1 (HIV-1) serology. All patients had placement of an Ommaya reservoir, regardless of CSF cytologic results. If lymphoma could be established on a frozen section specimen, the reservoir was placed at the time of diagnostic biopsy. 2 Chemotherapy consisted of pre-RT methotrexate 1 g/m , given weekly for two doses along with six doses of intra-Ommaya methotrexate (12 mg per dose) given on a twice-per-week schedule (Fig 1). Leucovorin rescue was given after each dose of systemic (10 mg leucovorin every 6 hours) and intra-Ommaya (10 mg every 12 hours) methotrexate. This was followed by 4,000-cGy wholebrain RT plus a 1,440-cGy boost to the tumor bed. After a 3-week rest, patients received two courses of high-dose cytarabine (ara-C); each course consisted of two doses separated by 24 hours of ara-C 3 g/m 2 infused over 3 hours. Patients received dexamethasone, usually 16 mg/d, during treatment with methotrexate, and were begun on a steroid taper during cranial RT. All patients were off dexamethasone at the start of ara-C. Thirty-one patients were eligible and participated in this combined modality regimen. During this same period, 16 additional patients seen at MSKCC had either refused chemotherapy (n = 1) or had started cranial RT at another institution before our consultation (n = 15). All of these patients had undergone a similar staging evaluation at diagnosis and would have been eligible to receive our protocol. These patients all received at least 4,000-cGy whole-brain RT as the sole initial therapy for PCNSL; no post-RT chemotherapy was administered to any of these patients except at recurrence. All patients with a diagnosis of ocular involvement received 3,000- to 4,000-cGy ocular RT, regardless of treatment group. At recurrence all patients had a repeat cranial CT/magnetic resonance imaging (MRI) scan, Ommaya tap, lumbar puncture, and ophthalmologic examination. At relapse, there was no uniform therapeutic approach; patients were treated with a variety of regimens depending upon the nature and location of their recurrent disease. Follow-up data are complete through April 1, 1991. Response was assessed by follow-up CT/MRI scans according to 16 the criteria reported by Macdonald et al, as well as time to recurrence and survival. A CR was defined as resolution of enhancing tumor, and the patient must be off corticosteroid Diagnosis Dexamethasone 16mg/day Ommaya placement IV MTX (1 gm/M 2 ) Day 1,8 Intra-Ommaya MTX (12 mg/dose) Day 1,4,8,11,15,18 Taper dexamethasone (off by completion of RT)

therapy. A partial response (PR) was defined as at least a 50% reduction in the size of the mass, and the patient must be on a stable or decreasing dose of steroid. A minor response (MR) represented less than a 50% decrease in the size of the tumor with the same steroid restrictions; stable disease (SD) was defined as no objective change in the lesion. Progressive disease (PD) represented an unequivocal increase in tumor size or the appearance of new lesions. Survival curves were drawn using the Kaplan-Meier product-limit method." Survival duration was measured from date of diagnosis to date of death or last follow-up. Potential prognostic factors were evaluated by Cox proportional hazards analysis both for survival and time to recurrence.l" RESULTS Patient Characteristics

Thirty-one patients received the combined modality regimen (group A), and 16 patients received RT alone (group R); all are assessable. There were no differences in the patient characteristics between the two treatment

groups (Table 1). Half of the patients were men and half, women. The median age was 58 years for patients in group A and 63 years in group R. Neither age nor sex had any effect on the time to recurrence or survival. Of

47 patients, 46 had a lumbar puncture at diagnosis; one patient with a large posterior fossa tumor could not safely undergo a spinal tap. Of 46 patients, 17 (37%) had

pathologic evidence of leptomeningeal lymphoma: two had meningeal infiltration demonstrated on biopsy, and 15 had a positive CSF cytologic examination. Five patients had ocular lymphoma; in two, it was clinically silent. No patient had evidence of systemic lymphoma, and all patients were HIV-1 negative. Diagnosisand Pathology

Diagnosis was established on a surgical specimen in 37 of the total 47 patients (Table 2). Of these 37 patients, 20 (54%) had a stereotactic biopsy and 17 (46%), a resection. Three of the 17 resected patients (18%) experienced a severe, permanent postoperative deficit; there were no complications after stereotactic Table 1. PCNSL Patient Characteristics

WBRT 200 cGyx20 (total 4000 cGy) Coned-down 180 cGyx8 (total 1440 cGy) 3 week rest IV cytosine arabinoside (3 gm/M 2 /dose) 1 dose/day for 2 consecutive days 3 week rest 2 IV cytosine arabinoside (3 mg/M /dose) 1 dose/day for 2 consecutive days

Fig 1. Outline of treatment protocol for PCNSL at MSKCC.

Group A n =31)

Group R (n= 16)

16 15

8 8

Total (%)

Sex

Male Female Age (years)

Median Range

58 28-79

63 33-80

31 12 2

16 5 3

Location at diagnosis

Brain Meninges Eyes

Downloaded from ascopubs.org by EAST CAROLINA UNIVERSITY on August 4, 2019 from 150.216.068.200 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.

47(100) 17 (37) 5 (11)

637

CHEMOTHERAPY AND RT FOR PRIMARY CNS LYMPHOMA Table 2. Tissue Diagnosis

Diagnosis Surgery Resection Biopsy CSF Autopsy No tissue Pathology Large cell Immunoblastic Lymphoma NOS Other

Response to Therapy

Group A (n = 31)

Group R (n= 16)

23 8 15 5 3

14 9 5 1 1 -

14 5

8 4

8 1

3 1

Abbreviation: NOS, not otherwise specified.

biopsy. Because CSF cytologic examination showed malignant lymphocytes, biopsy was not performed in six (13%) patients. Four patients did not have tissue confirmation before initiation of treatment. One group R patient without a history of systemic malignancy was presumed to have brain metastases despite MRI scan features suggestive of PCNSL (multiple periventricular lesions that intensely and diffusely enhanced), and was treated with cranial RT. She subsequently developed recurrent cerebral lesions. Further diagnostic and therapeutic options were refused, and PCNSL was confirmed at autopsy. Three patients from group A had homogeneously enhancing mass lesions on CT or MRI scan that disappeared after several days of corticosteroid. One patient had two nondiagnostic biopsies, and the other two did not undergo biopsy because of complete resolution of the masses. These patients were presumed to have PCNSL because of the characteristic radiographic appearance of their lesions and their brisk sensitivity to corticosteroid. The histologic subtypes of lymphoma were comparable in the two groups (Table 2): 45% to 50% of patients had large-cell lymphoma, and 16% to 25% had immunoblastic lymphoma. Immunophenotyping was performed on only eight specimens; seven were B-cell tumors, and one could not be typed. Treatment In group A, all patients completed the full course of systemic and intra-Ommaya methotrexate. Eleven of 31 patients received whole-brain RT alone because multiple lesions precluded the design of an appropriate boost; five of 11 received 5,000-cGy whole-brain RT, and the remaining patients received 4,000 cGy. All patients who completed RT received high-dose ara-C.

After methotrexate and before the start of cranial RT, 22 of 31 patients were assessable for a response to chemotherapy. In nine patients, tumor had been either completely resected (n = 4) or had disappeared after treatment with corticosteroid (n = 1), or films were unavailable for review (n = 4). Of the 22 patients, 14 (64%) had a PR, three an MR (14%), and five had SD (23%). All patients were receiving concurrent corticosteroid during methotrexate, and three of five patients with SD and one of three with an MR to methotrexate had substantial PRs to corticosteroid alone. At the completion of the treatment regimen, 27 of 31 patients from group A had a CR. Two patients had PD; two patients had a substantial PR, but died during RT of intercurrent medical problems due to their neurologic disability (pneumonia) or as a complication of steroid therapy (Pneumocystis carinii pneumonia). Because the majority of the 27 patients had a CR after finishing cranial RT, an independent assessment of the efficacy of high-dose ara-C could not be made. There were six patients with small lesions after RT and before ara-C. All of these patients had had a substantial PR after RT, but small residual enhancing lesions remained. Two developed a CR immediately after completing ara-C; the remaining four patients developed a CR several months after completing ara-C. These patients were neurologically well during this time but had delayed resolution of CT/MRI scan abnormalities, which may represent slow clearing of debris from the CNS. All patients were off corticosteroid at the time of ara-C administration and at the time of CT/MRI assessment of their response. All patients in group R had a CR after finishing RT. Recurrence of PCNSL Two patients in group A and three in group R had PD through treatment and were not evaluated for recurrent disease; all five died within 6 months of diagnosis despite RT and a variety of chemotherapy regimens. Therefore, 29 group A patients and 13 group R patients were assessed for relapse; 10 of 29 (34%) group A and 13 of 13 (100%) group R patients developed recurrent PCNSL. Patients receiving chemotherapy plus RT had a significantly prolonged time to recurrence compared with patients receiving RT alone (P = .003; Fig 2). The median time to recurrence was 41 months for group A and 10 months for group R. Recurrence in the brain was the only site-specific relapse that could be adequately evaluated and compared between treatment groups, and there were significantly more brain relapses in group R

Downloaded from ascopubs.org by EAST CAROLINA UNIVERSITY on August 4, 2019 from 150.216.068.200 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.

638

DEANGELIS ET AL

1.0

I

Fig 2. Time to recurrence for group A patients (0, chemotherapy plus RT; 29 patients, 19 censored) and group R patients (O, RT alone; 13 patients, none censored) patients. Median time to recurrence: group A, 41 months;

e

grup n, Tikmrk Months since diagnosis

(12 of 13) than in group A (nine of 29; P = .0035; Table 3). The median time to recurrence of group A patients was identical when data were analyzed excluding the three patients who lack tissue confirmation of their PCNSL. The location of recurrent disease differed between the two treatment groups. Nine of 10 group A patients developed a parenchymal brain recurrence; four of nine (44%) developed lesions in regions remote from the original site, including one patient who relapsed in both areas. Eight of the nine patients with recurrent brain disease had been treated with the full RT protocol, using 4,000-cGy whole-brain RT plus a 1,440-cGy boost. The one patient with multiple sites of relapse had received only 5,040-cGy whole-brain RT because of multifocal disease at presentation. Therefore, no correlation could be established between the location of recurrent brain disease and the dose of RT delivered as relapses occurred with approximately equal frequency within or outside of the volume of brain receiving the boost. In group R, 12 of 13 had brain relapses, with 92% occurring in new sites. Intramedullary spinal cord lymTable 3. Recurrent PCNSL Group A (n = 29) No.

Median time to recurrence (months) Location Any Brain Meninges Eyes Spinal cord Systemic

%

41 10 9 4 2 0 1

Group R (n = 13) No.

%

10 34 31 14 7 0 3

13 12 6 1 2 0

P

.003 100 92 46 8 15 0

.0035

Iv

monms

idcts

r=.vJ las

ol

up.

phoma developed in two patients from group R, one with a coexistent brain relapse. Leptomeningeal relapse occurred in four (14%) patients from group A, and in six (46%) from group R, although this was not a statistically significant difference. Ocular lymphoma developed in two patients from group A, neither of whom had ocular disease at diagnosis, and in one patient from group R who had a relapse after ocular RT. One group A patient (3%) has recently developed a systemic recurrence with biopsy of a cervical lymph node showing a nonHodgkin's lymphoma, histologically identical to his cerebral lesion. Survival and Treatment of RecurrentDisease The median survival doubled with the addition of chemotherapy from 21.7 months for group R to 42.5 months for group A, but this did not reach statistical significance in part because of the small number of patients in each group (Fig 3). The median survival of group A patients was not affected by the inclusion of the three patients who lacked tissue confirmation at entry onto the study. More importantly, the survival of group R patients was improved by the treatment administered for recurrent disease, and because these patients recurred early in their course, there is longer follow-up for the majority of group R patients treated after recurrence. At the time of relapse, all patients were offered additional treatment, but six patients (three from each group) did not receive further therapy. Four patients refused treatment, one was lost to follow-up, and one patient died suddenly but was presumed to have had recurrent disease based on a history of new neurologic symptoms a few days before death. The average survival for these six patients was 2 months and did not differ

Downloaded from ascopubs.org by EAST CAROLINA UNIVERSITY on August 4, 2019 from 150.216.068.200 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.

639

CHEMOTHERAPY AND RT FOR PRIMARY CNS LYMPHOMA 2.0 0.9 0.6 0.7

0>

0

0.5 0.4

0

Fig 3. Survival curve of group A patients (0; 31 patients, 17 censored) and group R patients (0; 16 patients, seven censored) patients. Median survival: group A, 42.5 months; group R, 21.7 months; P = .228. Tick mark indicates last follow-up.

0.6

o

0.3

o. 0.2 0.1 0.0 0

12

24

48 Months

between the two groups (Table 4). Five patients from group A received subsequent chemotherapy for recurrent brain disease; all received agents that penetrate the blood-brain barrier (BBB; four received high-dose ara-C and one methotrexate, thiotepa, and procarbazine). Three of five (one CR, one PR, and one SD) are still alive with an average follow-up of 10 months. Two group A patients were treated with ocular RT for an ocular recurrence. One patient survived 13.5 months but died of recurrent brain disease that was not treated. The second patient had an ocular and leptomeningeal recurrence; she received intra-Ommaya ara-C in addition to ocular RT and did well for 15 months, at which time she developed recurrent brain and ocular lymphoma that was not treated and died shortly thereafter. The patient with a systemic relapse has been started on systemic chemotherapy using cyclophosphamide, Adriamycin (doxorubicin; Adria Laboratories, Columbus, OH), vincristine, and prednisone (CHOP). Nine patients from group R received additional chemotherapy for recurrent brain disease. A variety of regimens were used and can be divided into two types. Five patients received agents that penetrate the BBB (one received high-dose ara-C; two methotrexate and high-dose ara-C; one methotrexate, thiotepa, and procarbazine; and one modified cyclophosphamide, doxorubicin, etoposide, ara-C, bleomycin, vincristine, and methoTable 4. Treatment of Recurrence

Chemotherapy BBB-permeable CHOP Ocular RT Spinal cord RT None

36

Group A

Group R

(n = 10)

(n = 13)

5 5

9 5 4 -

2 3

1 3

60

72

84

96

108

120

since diagnosis

trexate [ProMACE-Cytabom]), and four received agents the first three of which do not enter a normal BBB (four received CHOP: cyclophosphamide 750 mg/m2, Adriamycin 50 mg/m2, vincristine 1.4 mg/m2, and prednisone equivalent to 60 mg/d). Three of four patients treated with CHOP are dead, and the fourth is lost to follow-up; the average survival was 4 months from the time to recurrence, only 2 months longer than patients who were not treated at all. Of the five patients treated with other agents, four are alive with an average follow-up of 14 months, and one patient is still alive at 28 months. Toxicity Acute toxicity from chemotherapy was mild. There was one episode of methotrexate nephrotoxicity; the patient was supported with high-dose leucovorin and vigorous hydration and made a complete recovery. All patients developed myelosuppression during high-dose ara-C, but it was well-tolerated; two patients required RBC transfusion, and one patient was hospitalized for culture-negative nadir fever and made a complete recovery. No neurologic toxicity was seen with high-dose ara-C. There was no significant acute toxicity from cranial RT. Of note, one patient in each treatment group died of fulminant P carinii pneumonia in the setting of a dexamethasone taper. The patient in group A died at the start of her cranial RT and at autopsy had only a microscopic focus of residual disease; the group R patient died during treatment for recurrent brain disease. Late neurologic toxicity, with dementia and ataxia, was seen in three of 31 (9.7%) group A patients. The incidence of late toxicity is related to long-term survival, and this complication appeared in 11.5% of 1-year survivors (three of 26). One patient died 18 months after diagnosis from progressive dementia; there was no

Downloaded from ascopubs.org by EAST CAROLINA UNIVERSITY on August 4, 2019 from 150.216.068.200 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.

640

DEANGELIS ET AL

clinical evidence of recurrent PCNSL, although autopsy was denied. The other two are still alive at 15+ and 52+ months from diagnosis; neither can ambulate without assistance, and both have deteriorating cognitive function. All three patients were over the age of 60 years at the time of diagnosis and treatment of their PCNSL. No patient from group R developed late toxicity. Prospective neuropsychologic testing was not performed on either group. DISCUSSION

Systemic chemotherapy, including treatment of the leptomeninges, combined with RT for the initial treatment of PCNSL significantly prolongs disease-free survival and contributes to overall survival. Our data demonstrate that PCNSL is a chemosensitive primary brain tumor. The administration of systemic chemotherapy for the treatment of brain tumors has always been hampered, in part, by the limitations of the BBB. Bulky areas of PCNSL have an impaired BBB, which accounts for their intense enhancement seen on CT or MRI scans after the administration of contrast material. This would permit entry of drugs into the region of tumor that does not penetrate a normal BBB. However, we chose agents that not only had activity against non-Hodgkin's lymphoma, but also, in the doses used, could penetrate an intact BBB. This is because PCNSL is disseminated within the neuraxis at diagnosis. Lymphoma can involve the leptomeninges and eye as well as areas of the brain that can be infiltrated by microscopic tumor not evident on cranial scans, including MRI scans; this can extend to other parenchymal regions, such as the spinal cord. The high incidence of relapse in the leptomeninges and brain regions remote from the original tumor site seen in group R patients attests to microscopic disease that is underestimated at diagnosis. Alternative approaches to tumor residing behind an intact BBB include BBB disruption and intraarterial administration of chemotherapy.19 Although this method may be effective, it is cumbersome and may be associated with significant toxicity. More importantly, this method delivers adequate therapy only within the arterial territory undergoing BBB disruption, whereas appropriate systemic chemotherapy will effectively treat the entire CNS parenchyma. The success of our treatment protocol can be attributed not only to the use of high-dose agents that permeate a normal BBB, but also to the vigorous treatment of the leptomeninges using intra-Ommaya methotrexate. Whereas systemic chemotherapy was de-

signed to treat parenchymal brain disease, intraOmmaya chemotherapy was administered to insure adequate therapy to the CSF. Systemic high-dose methotrexate (2 3.0 g/m 2) can produce therapeutic levels within the CSF; however, these levels are not sustained for more than 24 hours. 20 High-dose intravenous ara-C can also yield significant drug concentrations in the CSF; however, CSF levels are unpredictable and are of short duration. 21,22Intra-Ommaya methotrexate gives therapeutic levels in the CSF that persist for 48 hours, and two doses may be administered per week.23'24 Use of an Ommaya reservoir as opposed to repeated lumbar puncture assures drug delivery into the CSF and provides better drug distribution within the entire subarachnoid space. 24 All of our patients received intra-Ommaya methotrexate whether or not we could demonstrate the meningeal lymphoma we knew to be present. In fact, not only did group A patients have fewer brain and no spinal cord recurrences, there were also fewer meningeal relapses, although statistical significance was not reached because of the small number of patients. Therefore, adequate therapy directed to the CSF is an important component of treatment for PCNSL; it is not known whether or not high-dose systemic chemotherapy with methotrexate or ara-C can accomplish the same goal. Although this was not a randomized prospective study, we believe a comparison between our two treatment groups is valid for several reasons. Although patients in group R did not receive a uniform dose and schedule of RT, they all received an adequate dose using modern radiotherapy techniques. The minor variations in total dose administered to this group have not been shown to affect time to recurrence or survival. Most importantly, the time to recurrence of group R patients is comparable to all reported series of patients treated with RT alone. There were no differences in clinical features between the two groups, and group R patients would have been offered the chemotherapy protocol had they not begun RT early. Median survival was not statistically different between the treatment groups; however, the addition of chemotherapy to cranial irradiation has a biologically significant impact on survival. The median survival of group A patients is substantially different from every reported series in the literature using RT alone and is the longest reported median survival for any therapeutic regimen.2 -9 Survival of group R patients was prolonged by the addition of chemotherapy at the time of recurrence. If further therapy had not been instituted, group R would have followed the well-established rapid demise of

Downloaded from ascopubs.org by EAST CAROLINA UNIVERSITY on August 4, 2019 from 150.216.068.200 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.

641

CHEMOTHERAPY AND RT FOR PRIMARY CNS LYMPHOMA

untreated patients at relapse, giving a median survival only a few months different from the median time to recurrence (10 months). However, the prolonged survival of group A patients led to the appearance of late toxicity not previously seen in PCNSL patients. Delayed neurologic sequelae of treatment usually occur more than 1year after completion of therapy, and the diagnosis can only be established in the absence of tumorY Therefore, the incidence of late toxicity is directly proportional to the percentage of patients with long-term disease-free survival. Age is known to be an important predisposing factor to the late consequences of cranial irradiation. The very young (age < 3 years) are particularly susceptible, but it also appears from our data that the elderly may be more vulnerable. Neither the specific injurious components of a therapeutic regimen nor the individual cellular mechanisms of injury have been identified. Cranial RT for a variety of cerebral neoplasms is known to produce significant toxicity in long-term survivors. 25-27 The only previous report of

delayed toxicity after treatment for PCNSL occurred in a patient treated with whole-brain RT alone28; however, permanent sequelae of RT alone were not seen in group R patients because all developed recurrent tumor. Chemotherapy can potentiate the toxicity produced by RT alone, particularly with methotrexate, but we do not know whether or not chemotherapy significantly contributed to the development of delayed toxicity in group A patients independent of enhanced survival.2 Because of our concern regarding the potential for late toxicity when combining methotrexate and cranial RT, we chose to administer the methotrexate before cranial RT; this sequence of administration reduces the risk of leukoencephalopathy and, at least experimentally, may be protective against late toxicity.29,30 In addition, preradiation chemotherapy permits an assessment of drug efficacy as most patients have complete resolution of their mass lesions after RT. We also used a relatively decreased dose of systemic methotrexate in an effort to reduce the cumulative amount of drug, which is a known risk factor for leukoencephalopathy. Although prospective studies have not been performed, retrospective reports suggest a dose response of PCNSL to cranial RT with improved responses at doses > 4,000 cGy.2 ,7Retrospective analyses for patients with systemic diffuse large-cell lymphoma treated with RT suggest a plateau for local control after 4,500 to 5,000 cGy; however, local recurrence was still observed in 20% to 38% of patients. 133 We administered 4,000 cGy to the whole brain and added 1,440 cGy to the region of

greatest involvement for a total of 5,440 cGy to the area of bulk disease. Because parenchymal brain recurrences in group A patients occurred with equal frequency within and out of the boosted volume, we cannot be certain that the boost enhanced our local control rate. Therefore, the optimal dose of cranial RT for PCNSL remains uncertain. Given our current information, we recommend that all patients with PCNSL be treated with 4,000- to 5,000-cGy whole-brain RT in 200- to 180-cGy daily fractions. The benefit of a boost remains unclear. The diagnosis of intracranial mass lesions is greatly simplified with the availability of CT and MRI scans. Although the precise nature of lesions cannot be ascertained from the radiographic appearance alone, there are specific characteristics of PCNSL that strongly suggest that diagnosis. Multiple, periventricular masses that densely and diffusely enhance after contrast administration are features of PCNSL seen in the majority of patients that distinguish it from either glioma or metastatic brain tumors.34' 35 When these radiographic features are observed, PCNSL is the leading diagnosis, and corticosteroids should be withheld until tissue confirmation has been obtained, unless the patient is herniating.3 6 As seen in three of our patients, corticosteroids are cytolytic in PCNSL and can cause lesions to shrink or disappear, thus eliminating the opportunity to establish the diagnosis with certainty."3738 This sensitivity to cortico-

steroid is unique to PCNSL, and is not seen with any other cerebral malignancy; nevertheless, tissue should be obtained for pathologic confirmation and steroid responsiveness should be used diagnostically only if the lesion has resolved completely and tissue is unobtainable. The diagnostic surgical approach differs from malignant gliomas where complete resection has a clear 39 survival advantage over biopsy alone." Stereotactic biopsy is the method of choice to establish the diagnosis of PCNSL; complete resection carries a high risk of permanent postoperative neurologic deficit and does not add to survival. Although PCNSL is histologically identical to aggressive non-Hodgkin's lymphomas that occur in the periphery, it remains confined to the nervous system, even at recurrence. At the time of diagnosis, none of the 47 patients had evidence of systemic lymphoma. Systemic evaluation with bone marrow and body CT scans is not necessary in the staging of a patient with newly diagnosed PCNSL, but extensive staging of the nervous system is essential, as many sites of disease, particularly the eyes and leptomeninges, are clinically silent. Even at

Downloaded from ascopubs.org by EAST CAROLINA UNIVERSITY on August 4, 2019 from 150.216.068.200 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.

642

DEANGELIS ET AL

recurrence, systemic involvement is rare, and usually clinically unimportant. The occasional patient with a systemic recurrence will require therapy, and conventional regimens are useful, as in our one group A patient. Our regimen significantly improved disease-free survival; however, patients did recur, and some patients did not respond to any initial therapy. All curative chemotherapeutic regimens for the treatment of comparable systemic non-Hodgkin's lymphomas use combination chemotherapy, relying heavily on alkylating agents. Further advances in the treatment of PCNSL are likely to include the use of multiagent regimens with BBB pene-

tration, and we are currently investigating the potential benefit of adding thiotepa and procarbazine to this regimen. If the efficacy of chemotherapeutic regimens can be substantially enhanced, PCNSL may become the first primary brain tumor that can be treated with chemotherapy alone. At present, any non-AIDS patient with PCNSL, including elderly patients, should be considered for preradiation chemotherapy, which was associated with a minimum of toxicity. ACKNOWLEDGMENT We gratefully acknowledge the support and helpful criticism of Dr Jerome B. Posner.

REFERENCES 1. Eby NL, Grufferman S, Flannelly CM, et al: Increasing incidence of primary brain lymphoma in the US. Cancer 62:24612465, 1988 2. Berry MP, Simpson WJ: Radiation therapy in the management of primary malignant lymphoma of the brain. J Rad Oncol Biol Phys 7:55-59, 1981 3. Freeman CR, Shustik C, Brisson ML, et al: Primary malignant lymphoma of the central nervous system. Cancer 58:11061111, 1986 4. Henry JM, Heffner RR, Dillard SH, et al: Primary malignant lymphomas of the central nervous system. Cancer 34:1293-1302, 1974 5. Littman P, Wang CC: Reticulum cell sarcoma of the brain. Cancer 35:1412-1420, 1975 6. Murray K, Kun L, Cox J: Primary malignant lymphoma of the central nervous system. J Neurosurg 65:600-607, 1986 7. Sagerman RH, Collier CH, King GA: Radiation therapy of microgliomas. Radiology 149:567-570, 1983 8. Woodman R, Shin K, Pineo G: Primary non-Hodgkin's lymphoma of the brain. A review. Medicine 64:425-430, 1985 9. Schaumburg HH, Plank CR, Adams RD: The reticulum cell sarcoma-microglioma group of brain tumours. A consideration of their clinical features and therapy. Brain 95:199-212, 1972 10. Appen RE: Posterior uveitis and primary cerebral reticulum cell sarcoma. Arch Ophthalmol 93:123-124, 1975 11. Rockwood EJ, Zakov ZN, Bay JW: Combined malignant lymphoma of the eye and CNS (reticulum-cell sarcoma). J Neurosurg 61:369-374, 1984 12. Abelson HT, Kufe DW, Skarin AT, et al: Treatment of central nervous system tumors with methotrexate. Cancer Treat Rep 65:137-140, 1981 (suppl 1) 13. Ervin T, Canellos GP: Successful treatment of recurrent primary central nervous system lymphoma with high-dose methotrexate. Cancer 45:1556-1557, 1980 14. Loeffler JS, Ervin TJ, Mauch P, et al: Primary lymphomas of the central nervous system: Patterns of failure and factors that influence survival. J Clin Oncol 3:490-494, 1985 15. Kawakami Y, Tabuchi K, Ohnishi R, et al: Primary central nervous system lymphoma. J Neurosurg 62:522-527, 1985 16. Macdonald DR, Cascino TL, Schold SC Jr, et al: Response criteria for phase II studies of supratentorial malignant glioma. J Clin Oncol 8:1277-1280, 1990

17. Kaplan EL, Meier P: Non-parametric estimations from incomplete observations. J Am Stat Assoc 53:457-481, 1958 18. Cox DR: Regression models and life-tables. J R Stat Soc 34:187-220, 1972 19. Neuwelt EA, Frenkel EP, Gumerlock MK, et al: Developments in the diagnosis and treatment of primary CNS lymphoma. Cancer 58:1609-1620, 1986 20. Borsi JD, Moe PJ: A comparative study on the pharmacokinetics of methotrexate in a dose range of 0.5 g to 33.6 g/m 2 in children with acute lymphoblastic leukemia. Cancer 60:5-13, 1987 21. Slevin ML, Piall EM, Aherne GW, et al: Effect of dose and schedule on pharmacokinetics of high-dose cytosine arabinoside in plasma and cerebrospinal fluid. J Clin Oncol 1:546-551, 1983 22. Capizzi RL, Yang J-L, Cheng E, et al: Alteration of the pharmacokinetics of high-dose Ara-C by its metabolite, high Ara-U in patients with acute leukemia. J Clin Oncol 1:763-771, 1983 23. Ettinger LJ, Chervinsky DS, Freeman AI, et al: Pharmacokinetics of methotrexate following intravenous and intraventricular administration in acute lymphocytic leukemia and non-Hodgkin's lymphoma. Cancer 50:1676-1682, 1982 24. Shapiro WR, Young DF, Mehta BM: Methotrexate: Distribution in cerebrospinal fluid after intravenous, ventricular and lumbar injections. N Engl J Med 293:161-166, 1975 25. DeAngelis LM, Delattre JY, Posner JB: Radiation-induced dementia in patients cured of brain metastases. Neurology 39:789796, 1989 26. Hochberg FH, Slotnick B: Neuropsychologic impairment in astrocytoma survivors. Neurology 30:172-177, 1980 27. Lieberman AN, Foo SH, Ransohoff J, et al: Long term survival among patients with malignant brain tumors. Neurosurgery 10:450-453, 1982 28. Merchut MP, Haberland C, Naheedy MH, et al: Long survival of primary cerebral lymphoma with progressive radiation necrosis. Neurology 35:552-556, 1985 29. Bleyer WA: Neurologic sequelae of methotrexate and ionizing radiation: A new classification. Cancer Treat Rep 65:89-98, 1981 (suppl 1) 30. Geyer JR, Taylor EM, Milstein JM: Radiation, methotrexate, and white matter necrosis: Laboratory evidence for neural radioprotection with pre-irradiation methotrexate. Int J Rad Oncol Biol Phys 15:373-375, 1988 31. Bush RS, Gospodarowicz M, Sturgeon J, et al: Radiation

Downloaded from ascopubs.org by EAST CAROLINA UNIVERSITY on August 4, 2019 from 150.216.068.200 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.

643

CHEMOTHERAPY AND RT FOR PRIMARY CNS LYMPHOMA therapy of localized non-Hodgkin's lymphoma. Cancer Treat Rep 61:1129-1136, 1977 32. Fuks Z, Kaplan HS: Recurrence rates following radiation therapy of nodular and diffuse malignant lymphomas. Radiology 108:675-684, 1973 33. Bush RS, Gospodarowicz M: The place of radiation therapy in the management of patients with localized non-Hodgkin's lymphoma, in Rosenberg SA, Kaplan HS (eds): Malignant Lymphomas: Etiology, Immunology, Pathology, Treatment. New York, NY, Academic, 1982, pp 485-502 34. Thomas M, Macpherson P: Computed tomography of intracranial lymphoma. Clin Radiol 33:331-336, 1982 35. Schwaighofer BW, Hesselink JR, Press GA, et al: Primary

intracranial CNS lymphoma: MR manifestations. Am J Neuroradiol 10:725-729, 1989 36. DeAngelis LM: Primary central nervous system lymphoma: A new clinical challenge. Neurology 41:619-621, 1991 37. Singh A, Strobos RJ, Singh BM, et al: Steroid-induced remissions in CNS lymphoma. Neurology 32:1267-1271, 1982 38. Vaquero J, Martinez R, Rossi E, et al: Primary cerebral lymphoma: The "ghost tumor." Case report. J Neurosurg 60:174176, 1984 39. Wood JR, Green SB, Shapiro WR: The prognostic importance of tumor size in malignant gliomas: A computed tomographic scan study by the Brain Tumor Cooperative Group. J Clin Oncol 6:338-343, 1988

Downloaded from ascopubs.org by EAST CAROLINA UNIVERSITY on August 4, 2019 from 150.216.068.200 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.