Central Nervous System Relapse in Malignant Lymphomas: Risk Factors and Implications for Prophylaxis MATERIALS AND METHODS

Transcription

1 Central Nervous System Relapse in Malignant Lymphomas: Risk Factors and Implications for Prophylaxis By Joseph P. Litam, Fernando Cabanillas, Terry L. Smith, Gerald P. Bodey, and Emil J. Freireich The records of 292 patients with malignant lymphoma other than Hodgkin s disease, registered in our protocols from 1967 to 1977, were reviewed to identify those with central nervous system (CNS) involvement. Thirty-one patients were encountered with this complication, an incidence of 11 %. Patients with a diffuse histology had a higher frequency of CNS recurrences (27/174 = 16%) in contrast to only 4/118 (3%) for those with nodular types. However, if only patients with diffuse histology in CR are considered, the frequency of CNS relapse is 13.5% (13/98). The risk factors that predict for the development of this complication were studied using multivariate analysis. Diffuse poorly differentiated lymphocytic and diffuse undifferentiated lymphomas were found to be associated with a high risk of CNS relapse. Prior chemotherapy. bone marrow involvement, age <35, and extranodal disease were also identified as high-risk factors. Using the information generated by a logistic regression model, patients with malignant lymphoma of diffuse type can be classified into three categories when first seen: low-risk group, intermediate, and high-risk group. CNS prophylaxis is recommended for the intermediate and high-risk group, while only close follow-up is advised for the lowrisk group patients who have one adverse characteristic. T HE FREQUENCY ofcentral nervous system (CNS) relapses in patients with malignant lymphoma, other than Hodgkin s disease, has been previously reported as ranging from 5% to 29%.i.5 The CNS relapse rate in malignant lymphomas appears to have increased since the development of effective combination chemotherapy for these disorders. This complication has been uniformly associated with a poor prognosis. Although prophylactic CNS treatment has been recommended for patients with malignant lymphoma, it is not clear which patients should be considered as candidates for this approach. Furthermore, it is not clear which therapeutic modalities should be utilized for prophylaxis. Ihis study attempts to identify those pretreatment variables that could predict for a high risk of CNS relapse in patients with malignant lymphoma. Also, by studying the clinical behavior of this disorder, we attempt to determine which approach to prevention would be most desirable. MATERIALS AND METHODS General Considerations The study group consisted of 292 consecutive patients who were treated in the Department of Developmental Therapeutics, M. D. Anderson Hospital and Tumor Institute, from 1967 to 1977 with From the Departments of Developmental Therapeutics and Biomathematics, The University of Texas System Cancer Center, M. D. Anderson Hospital and Tumor Institute, Houston, Texas. Supported in part by Grants CA and from the National Cancer Institute, National Institutes of Health, USPHS. Bethesda, Md. Submitted March ; accepted August 1. /979. Address reprint requests to Fernando Cabanillas, M.D.. M. D. Anderson Hospital, 6723 Bertner, Houston, Texas /979 by Grune & Stratton, Inc /79/ $0l.00/0 Blood, Vol. 54, No. 6 (December),

2 1250 LITAM ET AL. four different protocols: COP regimens,6 7 CHOP-HOP,8 CHOP-BIeo,9 and CHOP-Bleo-Levamisole. #{176} The tumors were classified according to Rappaport s criteria after review by one of the members of the staff of our pathology department. The clinical staging was performed according to recommendations of the Ann Arbor conference. 2 Patients with Burkitt s lymphoma and immunoblastic Iymphadenopathy were also included in this study. Definitions Central nervous system (CNS) disease was established by any one of the following: ( 1 ) presence of lymphoma cells in the cerebrospinal fluid, (2) postmortem histologic diagnosis of CNS lymphoma, (3) histologic proof of CNS disease by biopsy or surgery, (4) use of myelogram, brain scan, and/or CT scan. Two patients who did not meet the above criteria were included because they developed facial nerve palsy associated with elevation of cerebrospinal fluid protein (158 and 92 mg/too ml) without Iymphoma cells in the CSF cytologic examination. Both showed temporary improvement after treatment with radiotherapy. In addition, both patients developed progressive CNS symptoms and signs suggestive of relapse of their disease. Bone marrow involvement was considered to be present whenever lymphoma cells were identified by bone marrow aspiration and/or biopsy. A bone marrow biopsy was considered mandatory to adequately evaluate this possibility. Patients who only had bone marrow aspirations were excluded from the analysis of this variable, although they were included in the analysis of the other variables. A bulky tumor was defined as any palpable abdominal mass exclusive of the liver or spleen or any mass visible in flat plate of the abdomen or shown to produce displacement of the abdominal viscera (not including ureters) as shown by studies using contrast media. Mediastinal masses were also considered as bulky if there was more than merely prominence of the hilum. If a tumor that might qualify as bulky was resected prior to chemotherapy, it was not classified as bulky. Complete remission (CR) was defined as a disappearance of all signs and symptoms of disease. Partial remission (PR) was defined as greater than 50% but less than 100% reduction in the sum of maximum diameters of all measurable lesions. All other patterns of response, including mixed response, stable disease, and progressive disease, were classified as failures. All patients were evaluated for response after every 3-4 cycles of treatment. All studies initially interpreted as demonstrating presence of tumor, i.e., lymphangiogram, gallium scan, bone marrow aspirations and biopsy, chest x-ray, etc., were repeated in order to assess response status. The treatment for CNS relapse usually consisted of intrathecal therapy (cytosine arabinoside, 100 mg, or methotrexate, 20 mg) given twice a week until the CSF was cleared of lymphoma cells. Maintenance was then given once a week or every other week for 3 mo. Radiotherapy (ranging from 2500 to 3000 rad) to the brain was also used. In the case of cord compression, radiotherapy to the cord and/or lamintectomy was used. Statistical Considerations Survival curves for patients with and without CNS relapse were calculated by the method of Kaplan and Meier, 3 and the difference between the curves was tested for statistical significance using a generalized Wilcoxon test. 4 A chi-square test was used to test differences in CNS relapse rates by pretreatment variables considered individually. In addition to determining which individual pretreatment factors were related to development of CNS relapse, further analyses were carried out to determine the combination of patient characteristics most predictive of CNS relapse. A logistic regression model 5 was fitted for patients with diffuse histologies and resulted in an equation that relates probability of CNS relapse to a linear function of patient characteristics. The model has the following form: In a0 + a, (x, -,) a (Xk - l -pj where p represents the probability of CNS relapse, the a(s) are regression coefficients, the x(s) represent patient characteristics, and the are the means of the characteristics. The magnitude of the regression coefficients determines the contribution of each characteristic to the prediction of CNS relapse. Characteristics entered the model stepwise, and only those whose addition resulted in a significant reduction in maximized log-likelihood were included. Characteristics considered for inclusion in the model were: histology (I = DPDL or DUL, 2 - other), age (I - <35, 2 = 35), prior

3 RISK FACTORS IN CNS LYMPHOMAS 1251 chemotherapy (0 = no, I = yes), stage (3 = I, II, Ill, 4 = IV), bulkiness (0 = no, I = yes), marrow involvement (0 = no, I = yes), extranodal disease other than marrow (0 = no, I = yes), and chemotherapy (1 = COP, 2 = CHOP). Application of the resulting model to data for a given patient allows the estimation of probability of CNS relapse for that patient; thus, some relative determination of risk may be made for patients with a combination of favorable and unfavorable characteristics. RESULTS General Considerations There were 31 patients with CNS lymphoma among the 292 patients, a frequency of I 1%. More than one-half of these patients (20) had meningeal involvement alone. The others were cord compression (4), parenchymal (4), meningeal and cord compression (2), and meningeal and parenchymal (I). Of the 292 patients, 185 achieved a complete remission (CR) after systemic chemotherapy. The frequency of CNS lymphoma as thefirst site ofrelapse among the 185 patients in CR was 6.5% (12/185). Ihe median time from onset of CR to CNS relapse was 7 mo (range, mo). There were 2 additional patients in CR who experienced bone or bone marrow recurrences simultaneously with the onset of CNS disease. Eight of the I 2 patients whose first site of recurrent disease was the CNS subsequently developed lymphomatous involvement of other organs. The median interval from CNS recurrence to systemic recurrence in these 8 patients was 3 mo (range, mo). Ihree patients died with CNS lymphoma and never developed evidence of recurrent systemic disease. The survival of these 3 patients, as measured from CNS relapse to death, was 3 days, 3 mo, and 4 mo. Iwo patients remained alive at 12+ mo and 30+ mo since the onset of therapy for their CNS lymphoma. Both patients with prolonged CNS disease remissions have documented CNS involvement. Both patients had a positive CSF cytology. In addition, one of them had a positive CI scan of the brain. In both patients, all the abnormal findings found initially reverted to normal after treatment. One of these patients remains in CR and the other patient has recently relapsed systemically, although his CNS disease is still in remission. The latter patient is included in the group of 8 who developed systemic relapse after CNS recurrence. One of these patients received chemotherapy through an Ommaya shunt, inserted early during his CNS relapse. They also received whole brain radiotherapy. Among the remaining I 7 patients with CNS lymphoma, one was found to have CNS disease at the time of initial presentation. Eight patients had achieved a partial remission of their systemic disease, and the first manifestation of progressive disease was CNS lymphoma. The remaining 8 patients had developed systemic recurrences following initial response to therapy, and subsequent to the systemic relapse, developed CNS lymphoma. Considering all 31 patients who developed CNS lymphoma, the median duration from onset of systemic chemotherapy to the time of CNS relapse was 7 mo. In 80% of these patients, CNS lymphoma was first diagnosed 5 mo or more after the onset of systemic chemotherapy. The median interval from the initial diagnosis of lymphoma to CNS relapse was 13 mo. The survival from start of systemic treatment for patients with and without CNS disease is compared in Fig. I. The median survival for patients who developed CNS relapse was 10 mo, and the median survival for patients without CNS relapse was

4 1252 LITAM ET AL o Total Foil No CNS Relapse CNS Relapses Alive 36 mos. P #{176}.40.2O 0 48 Months Fig. 1. Survival patients with and without CNS lymphoma. 36 mo; approximately half the patients in the latter group remain alive. The difference in the 2 survival curves was highly statistically significant (p < 0.01). There were 15 patients who responded to CNS therapy as measured by the improvement or disappearance of CNS symptoms, notably nerve palsy and the clearing of malignant cells in the CSF as well as normalization of previously elevated CSF protein and WBC count. Twelve patients did not respond to treatment and 4 patients received no treatment because they were diagnosed at the time of postmortem. The median survival (measured from CNS relapse to death) of responders was 7.5 mo compared to only 1 mo for nonresponders. Influence of Histology Patients with diffuse lymphoma had a much higher frequency of CNS recurrence than patients with nodular lymphoma. Only 3% (4/118) of the patients with nodular lymphoma developed CNS relapse compared to 16% (27/174) of all patients with diffuse lymphoma (p < 0.01) (Table 1). However, if only patients with diffuse histology in CR are considered, the frequency of CNS lymphoma as Table 1. Incidence of CNS Disease by Nodularity and Histologic Type Number of Patients Number of With CNS Disease Characteristics Patients (%) All patients (1 1 %) Nodular histologies (3%) NPDL 85 3 (4%) NHL 19 1 (5%) N. mixed 14 0 (0%) Diffuse histologies (16%) DHL (12%) DPDL 11 4(37%) D. mixed 4 1 (25%) DUL 7 3 (43%) Others 15 2(13%) lncludes patients with DWDLL. immunoblastic lymphadenopathy, unclassified lymphomas.

5 RISK FACTORS IN CNS LYMPHOMAS 1253 the first site of relapse was lower: 11.2% (11/98). CNS lymphoma occurred most frequently in patients with diffuse poorly differentiated lymphocytic lymphoma (DPDL, 37%) and in patients with diffuse undifferentiated lymphoma (DUL, 43%) (Table I). Of the 11 patients with DPDL, 5 were considered to have lymphoblastic lymphoma. The group with DUL included 2 patients with Burkitt s lymphoma, both of whom developed CNS disease. Ihere were 4 patients with DWDLL, who were included under other diffuse histologies in Tables I and 2. None of these 4 patients developed CNS recurrences. Table 2. Incidence of CNS Disease by Patient Characteristics, Type of Chemotherapy and Response-Diffuse Histology Only Characteristics No. Patients No. With CNS Disease (%) p Value All diffuse (16%) Prior therapy None (17%) XRT 47 2 (4%) <.01 Chemo ± XRT 10 5 (50%) Histology DHL, DMx, other diffuse (13%).01 DPDL. DUL 18 7 (39%) Age < (32%) (13%) (10%) > (12%) Marrow involvement No (15%).05 Yes 19 7 (37%) Extranodal disease other than marrow No 63 5 (8%).08 Yes (20%) Stage 13 2 (15%) III 34 1 (3%).07 IV (19%) Chemotherapy COP 53 5 (9%) CHOP-HOP 21 1 (5%).13 CHOP-Bleo (19%) CHOP-Bleo-Lev 37 9 (24%) Tumor bulkiness Nonbulky (20%).19 Bulky 82 9 (11%) Response CR 98 13(13.5%) PR (20%) Fail 13 1 (8%).36 Early death 10 0 (0%) Inevaluable 2t Sex Male (15%) 76 Female (17%) Three more patients achieved CR but developed systemic relapse before CNS relapse. tboth patients had no evidence of disease at start of treatment.

6 1254 LITAM ET AL. Relation of Pretreatment Variables to Risk of CNS Relapse Sixty percent of patients in this study had diffuse lymphoma, and because of the higher incidence of CNS relapse associated with this histologic type, further study of pretreatment variables related to CNS relapse was confined to this subgroup. In Table 2, the CNS relapse rate is given for several pretreatment patient characteristics, type of chemotherapy, and response to chemotherapy. CNS relapse rates of greater than 30% were noted for patients who had received prior chemotherapy, those under the age of 35, those with DPDL or DUL histologies, and those with bone marrow involvement. Prior chemotherapy is felt to be an important prognostic factor only because it reflects the mean duration of survival prior to completion. Therefore, patients surviving long enough to get additional therapy will be at higher risk of developing CNS relapse. A somewhat higher frequency of CNS relapse was also associated with stage IV disease and with involvement at extranodal sites other than bone marrow. Using the information on patient characteristics listed above, a logistic regression model was fitted and resulted in the following equation: 1 n ( = - I (histology - I.88) + I.939 (prior chemo -.06) l -p,i + I.80 1 (marrow ) - 1. I 5 1 (age - I.78) (other extranodal -.67) According to this model, histology (DPDL or DUL), prior chemotherapy, marrow involvement, age under 35, and other extranodal involvement are associated with an increased probability of CNS relapse. Variables are listed in the order in which they entered the model; thus, histology was the single factor most closely associated with CNS relapse, prior chemotherapy was the variable most important in combination with histology, and so on. Stage of disease was not included in the model because it did not contribute significantly to the fit of the model after other variables were included in the equation. Even though there was a correlation between DPDL and DUL histologies and age under 35, both of these variables were included, since both contributed independently to the model s capability to predict CNS relapse. Using information on each patient s pretreatment characteristics, the regression equation was solved to obtain a probability of CNS relapse for each patient in the study. Patients were then grouped according to these predictions, and observed CNS relapse rates were compared (Table 3). Within every category, the observed rates of CNS relapse were within the predicted range. Though applying the model to the same group of patients from which it was derived is likely to give an overly favorable impression of its fit, these results do support the validity of this model in predicting CNS relapse. Table 3. Predicted Ve rsus Observed Probabili ties of CNS Disease Bas ed on Regression Model Predicted Total No. No. CNS Relapses PercentCNS Relapses Probability Observed Observed Observed <12% %-29% >30%

7 RISK FACTORS IN CNS LYMPHOMAS 1255 Table 4. Probability of CNS Relapse Related to Risk Factors I. Low-risk group (0%- 10% probability CNS relapse) A. No unfavorable factors B. Only 1 of the following unfavorable factors 1. Age <35 2. Extranodal (other than marrow) disease 3. Bad histology (DUL or DPDL) II. Intermediate-risk group (1 1 %-29% probability) A. Prior chemotherapy B. Bone marrow disease C. Any of the following combinations 1. Bad histology + age <35 2. Bad histology + extranodal disease 3. Age <35 + extranodal disease III. Very high-risk group (>30% probability) A. More than 2 unfavorable characteristics B. Bone marrow + age <35 C. Bone marrow + extranodal (other than marrow disease) D. Bad histology + marrow disease E. Prior chemotherapy + any other unfavorable characteristics Ihe model could be similarly applied to prospectively estimate probabilities of CNS relapse for a group of new patients. If a patient had only one of the live unfavorable characteristics as determined by the model, his predicted probability of CNS relapse would be as follows: DUL or DPDL histology, 7%; age under 35, 10%; extranodal involvement, 10%; marrow involvement, I 8%; prior chemotherapy, 20%. Probabilities associated with all possible combinations of unfavorable factors have also been calculated, and these are listed in Table 4 grouped into low, intermediate, and high risk of CNS relapse. Given information on prognostic characteristics for a particular patient, it would be possible to assign him to one of three risk groups by referring to this table. DISCUSSION The overall frequency of CNS relapse in our study is 1 1%. This is within the range reported by most investigators when all histologic types are included. Evidently, the most important group in this study are those patients who achieve CR and, thus, are potentially curable but fail treatment because of CNS recurrence. In our series, there were 12 patients in this category. Ihus, the overall incidence of CNS lymphoma as the first site of relapse for patients in CR (including all histologic types) was 6.5% (12/185). Our major concern, however, is the subgroup of patients with diffuse histology in whom we observed a 16% frequency of CNS relapse in contrast to only 3% for nodular types. Bunn et al. have described a 29% incidence of CNS relapse for the DHL and DUL subtypes.5 This figure includes patients in CR, PR, and progressive disease. In our series, we observed CNS relapses in 20/144 (13.8%) of patients with DHL and DUL (Table I). The unusually high frequency of CNS recurrences observed by Bunn is probably accounted for by the high incidence of bone marrow involvement in their patients. Thirty-eight percent (20/52) of their patients with DHL and DUL, as contrasted to only 15% (19/127) (Table 2) of our patients with diffuse histologies, had bone marrow involvement. Bone marrow involvement

8 1256 LITAM ET AL. constitutes a high-risk factor for subsequent development of CNS lymphoma. Herman et al. have recently published the Southwest Oncology Group experience with CNS lymphoma in 50/ I 039 patients.2 Their incidence of CNS recurrences is low (5%) compared to other published series. However, information in regards to the pretreatment characteristics of their patient population is not given. It is conceivable that their lower incidence of CNS relapse could be explained by a difference in the risk factors, such as bone marrow involvement and age. The CNS was the first site of relapse in I I /98 ( I I.2%) of our patients with diffuse histology who achieved a complete remission. This frequency again is lower than that observed by Bunn et al. who describe a 20% incidence (5/25) in their patients who achieved a CR.5 The majority ofour patients who relapsed in the CNS while in CR also developed systemic disease (8/ 12 or 66.6%) at some point, usually early in their clinical course. A similar observation has been made by Bunn et al.5 The fact that most of the CNS relapses were followed shortly after by systemic relapse suggests two possibilities: ( I ) the CNS behaves as a sanctuary for lymphoma cells that eventually become resistant to chemotherapy due to chronic exposure to low doses of drugs that do not adequately penetrate the blood-brain barrier; these resistant cells then seed the rest of the body; or (2) the CNS, due to its very delicate nature, manifests evidence of disease earlier than other involved sites, such as kidney and liver. The rapidity with which systemic relapses developed after CNS relapse is in favor of this second possibility. All of the patients who relapsed both in the CNS and systemically, except one, were still on systemic chemotherapy at the time they developed systemic disease. Ihis suggests that CNS lymphoma results from a clone of cells resistant to the systemic chemotherapy agents being administered. In only one patient in our series has the CNS behaved as a definite sanctuary. He remains alive 12+ mo after CNS relapse, free of any evidence of disease either systemically or in the CNS. Ihe other three patients who developed exclusive CNS relapses died so early after CNS recurrence (3 days, 3 mo, and 4 mo) that it is impossible to determine whether they would have developed systemic disease had they survived longer. Consequently, we recommend that whenever a patient develops CNS lymphoma this should also be interpreted as an early sign of systemic relapse. These patients should, in addition to receiving local therapy for their CNS disease, have their systemic chemotherapy regimen changed before systemic disease manifests itself overtly. In our study, patients with nodular cell types had a very low frequency of CNS relapse (3%) (Table I ). The same observation has also been made by Herman et al. who found a 1% incidence ofcns recurrences in patients with nodular histologies.2 This group, therefore, need not be considered for CNS prophylaxis. On the other hand, the group of patients with diffuse histologies should be judged according to their risk factors. According to the results of this study, we recommend CNS prophylaxis in patients with diffuse histology who fall into the intermediate and high-risk groups (Table 4). Patients with DWDLL, however, should be considered separately, since according to our experience and that of others they rarely develop CNS disease even in the presence of bone marrow involvement. Those in the low-risk group who have at least one unfavorable characteristic should be followed closely with periodical spinal taps. If any abnormality, such as an unexplained elevation of the CNS protein content, is detected, prophylaxis should be considered.

9 RISK FACTORS IN CNS LYMPHOMAS 1257 We feel, however, that radiotherapy to the brain plus intrathecal chemotherapy should not be the only prophylactic treatment, in view of the fact that most patients with CNS relapse eventually develop systemic disease as pointed out above. One study utilizing high-dose methotrexate with CF rescue has already shown encouraging preliminary results in reducing the incidence of CNS relapse in patients with DHL. 6 This systemic approach offers the additional advantage of adequate penetration of drug into other areas that conceivably could behave as sanctuaries. The vast majority of patients with CNS lymphoma (25/31 or 80%) developed their CNS relapse after 5 mo or more from beginning of treatment. Ihus, CNS prophylaxis should be initiated soon after diagnosis is made and preferably within the first 3 mo of systemic therapy. REFERENCES I. Law IP, Dick FR, Blom J, Bergevin PR: Involvement of central nervous system in non- Hodgkin s lymphoma. Cancer 36: , Herman TS, Hammond N, Jones SE, Butler ii, Byrne GE, McKelvey E: Involvement of the central nervous system by non-hodgkin s lymphoma. Cancer 43: , Rosenberg SA, Diamond HD, Jaslowitz B, Craver LF: Lymphosarcoma: A review of I 269 cases. Medicine 40:3 1-84, Williams HM, Diamond HD, Craver LF, Parsons H: Neurological Complications of Lymphomas and Leukemias. Springfield, Ill, Charles C Thomas, I Bunn PA Jr. Schein PS, Banks PM, DeVita VT Jr: Central nervous system complications in patients with diffuse histiocytic and undifferentiated lymphoma: Leukemia revisited. Blood 47:3-10, Luce JK, Gamble if, Wilson HE, Monto RW, Isaacs BL, Palmer RL, Coltman CA Jr. Hewlett is, Gehan EA, Frei E III: Combined cyclophospha mide, vincristine and prednisone therapy of malignant lymphoma. Cancer 28: , Luce JK, Gehan EA, Gamble if, Wilson HE, Monto RW, Talley RW: High rate and long duration of complete remission in malignant lymphoma treated with combined oral cyclophosphamide, prednisone and vincristine. Proc Am Assoc Cancer Res 16:129, 1975 (abstr) 8. McKelvey EM, Gottlieb JA, Wilson HE, Haut A, Talley RW, Stephens R, Lane M, Gamble JF, Jones SE, Grozea PN, Gutterman J, Coltman C Jr, Moon TE: Hydroxyldaunomycin (adriamycin) combination chemotherapy in malignant lymphoma. Cancer 38: , Rodriguez V. Cabanillas F, Burgess M, McKelvey EM, Valdivieso M, Bodey GP, Freireich EJ: Combination chemotherapy ( C HOP- Bleo ) in advanced (non-hodgkin) malignant lymphoma. Blood 49: , Cabanillas F, Rodriguez V, Hersh E, Mavligit G, Bodey GP, Middleman EL: Chemoimmunotherapy of advanced non-hodgkin s lymphoma with CHOP-Bleo-Levamisole. Proc Am Soc Clin Oncol 18:328, 1977 (abstr) I 1. Rappaport H, Winter Wi, Hines EB: Follicular lymphoma: A Re-evaluation of its position in the scheme of malignant lymphoma based on a survey of 253 cases. Cancer 9: , Carbonne PP. Kaplan HS, Musshoff K, Smithers DW, Tubiana M: Report of the committee on Hodgkin s disease staging classification. Cancer Res3l:1860, Kaplan EL, Meier P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 53: , Gehan E: A generalized Wilcoxon test for comparing arbitrarily singly-censored samples. Biometrika 52: , Cox DR: The Analysis of Binary Data. London, Methuen, 1970, pp Skarin A, Canellos G, Rosenthal K, Moloney W, Frei E Ill: Histology non-hodgkin s lymphoma with high dose methotrexate and CF rescue, bleomycin, adriamycin, cyclophosphamide, oncovin and decadron. Proc Am Soc Clin Oncol 19:400, 1978 (Abstr)

10 : Central nervous system relapse in malignant lymphomas: risk factors and implications for prophylaxis JP Litam, F Cabanillas, TL Smith, GP Bodey and EJ Freireich Updated information and services can be found at: Articles on similar topics can be found in the following Blood collections Information about reproducing this article in parts or in its entirety may be found online at: Information about ordering reprints may be found online at: Information about subscriptions and ASH membership may be found online at: Blood (print ISSN , online ISSN ), is published weekly by the American Society of Hematology, 2021 L St, NW, Suite 900, Washington DC Copyright 2011 by The American Society of Hematology; all rights reserved.