Central Nervous System Involvement and the Role of Prophylactic Cranial Irradiation in Small Cell Lung Cancer

Central Nervous System Involvement and the Role of Prophylactic Cranial Irradiation in Small Cell Lung Cancer C.G. ALEXOPOULOS, M. VASLAMATZIS, E. PATILA, L. TARANTO Department of Medical Oncology, Evangelismos Hospital, Athens, Greece Key Words. Small cell lung cancer (SCLC) Prophylactic cranial irradiation (PCI) Brain CT-Scan Central nervous system (CNS) metastases ABSTRACT This paper studies the frequency and manifestations of central nervous system (CNS) involvement and assesses the role of prophylactic cranial irradiation in small cell lung cancer (SCLC). All patients with confirmed diagnosis, admitted to our department within the last 15 years, were included. Patients were staged as having limited or extensive disease. Irradiation (40 Gy/20f) was offered to all complete responders immediately after polychemotherapy. There were 200 patients (176 men and 24 women, median age 58), 68 with limited and 132 with extensive disease. Twenty (10%) presented with CNS involvement, 14 (7%) developed it during chemotherapy and 47 (23.5%) during follow-up. In total, 81 (40.5%) developed CNS involvement, and in 57 (28.5%) it was the main manifestation. There was no relation to disease extent or type of response to therapy. The most frequent site of metastases was brain (33%), followed by leptomeninges (6%), spinal cord (1.5%), and pituitary (1.5%). Of 79 complete responders, 51 (65%) received prophylactic cranial irradiation (PCI) and 28 (35%) did not. Frequency of CNS involvement was not significantly different (49% and 39%, respectively). Actuarial INTRODUCTION Central nervous system (CNS) metastases are a frequent complication of small cell lung cancer (SCLC) and occur at diagnosis or during the subsequent course of disease. Several reports have been published in the past commenting on incidence, distribution, management, and prognosis of brain metastases in SCLC [1-5]. With only one exception [5], those studies have been published at a time when diagnostic and mainly therapeutic approaches of patients with SCLC were very different from today. On the other hand, despite early suggestions that prophylactic cranial irradiation (PCI) should be part of the treatment of SCLC [6], its exact role is still evolving. We probabilities of developing CNS involvement were also not different. Nevertheless, 91% of complete responders without PCI relapsed only to CNS involvement, versus 48% with prophylactic irradiation. Cranioprophylaxis administration was followed by an improvement in overall survival, which was highly significant in limited disease. The actuarial survival of complete responders at two and four years was 46% and 26% with cranioprophylaxis versus 18% and 9% without, respectively. CNS involvement in SCLC not only is a frequent complication, but also its frequency increases with lengthening survival. The necessity of routine use of brain CT scan during staging and follow-up is questioned in view of the present data. Administration of cranioprophylaxis did not reduce the frequency of CNS involvement in our series apparently because while it significantly delays CNS involvement, it does not abolish it. Nevertheless, survival of complete responders was prolonged with cranioprophylaxis and very significantly so in limited disease. This last finding, although clear cut, must await confirmation from randomized trials. The Oncologist 1997;2:153-159 considered it appropriate, therefore, to present the findings of a study we conducted in the Department of Medical Oncology at Evangelismos Hospital, the largest general hospital in Greece. AIM OF THE STUDY The aim was twofold: to study the frequency and manifestations of CNS involvement and to assess the role of PCI in patients with SCLC. PATIENTS AND METHODOLOGY The population of the study consists of all patients with a confirmed diagnosis of SCLC admitted to our department Correspondence: C.G. Alexopoulos, M.D., Department of Medical Oncology, Evangelismos Hospital, 45, Ipsilantou Str., Athens 106 76, Greece. Telephone: 30-1-7243482; Fax: 30-1-7243482. Accepted for publication February 18, 1997. AlphaMed Press 1083-7159/97/$5.00/0 The Oncologist 1997;2:153-159

154 Role of Cranioprophylaxis in Small Cell Lung Cancer within the last 15 years. The diagnosis was based on one or more of the following: histology of the biopsy material taken during bronchoscopy, biopsy of a peripheral node/tissue, cytology of fine needle aspiration, sputum cytology, and/or bronchial brushing and lavage. Staging procedures included: detailed clinical examination, full blood count and erythrocyte sedimentation rate (ESR), liver biochemistry, renal function tests, chest X-rays, CT scan of thorax and abdomen, bone scan, and skeletal survey as indicated. The policy regarding the use of brain CT scan during staging and follow-up varied during the study. Patients were staged as having limited (LD) or extensive disease (ED). Fairly aggressive polychemotherapy was used. Although minor variations between chemotherapeutic regimens did exist, the basic chemotherapy combination included cyclophosphamide, doxorubicin, and etoposide ± methotrexate. Thirtyseven patients received three cycles of cisplatin plus etoposide alternating with three cycles of the basic regimen (Table 1). World Health Organization response criteria were used for assessment and characterization of response. According to our therapeutic protocol, PCI was offered to all patients with a documented complete response (CR) immediately after six courses of chemotherapy. Accordingly, this is not a randomized study and complete responders did not receive PCI, either because they denied it or because of residency outside Athens. The irradiation schedule was uniform throughout the study and consisted of whole-brain irradiation at 40 Gy in 20 fractions using a Cobalt-60 machine. Table 1. Chemotherapeutic regimens used throughout the study Chemotherapeutic combination Schedule Cycles n Patients n Vincristine 2 mg i.v. stat d.1 Doxorubicin 40 mg/m 2 i.v. stat d.1 Etoposide 100 mg/m 2 IVI/2 h d.1,3,5 q 3 wks 6 115 Cyclophosphamide 1,000 mg/m 2 i.v. stat d.2 Methotrexate 200 mg/m 2 IVI/6 h d.2 Folinic acid rescue for 48 h Cyclophosphamide 1,200 mg/m 2 i.v. stat d.1 Doxorubicin 45 mg/m 2 i.v. stat d.1 q 3 wks 6 48 Etoposide 120 mg/m 2 IVI/6 h d.1,2,3 Cisplatin 100 mg/m 2 i.v. stat d.1 Etoposide 100 mg/m 2 IVI/6 h d.1-5 q 6 wks 3 37 (alternating with regimen n 1 3) For statistical analysis of the results, X 2 test, Kaplan-Meier method, and log rank test were used. RESULTS The characteristics of the 200 patients studied are shown in Table 2. Median follow-up was 11 months (1-120 months). Eighty-seven patients (43.5%) were followed for more than 12 months. CNS Involvement In total, 81 patients (40.5%) developed CNS involvement sometime during the course of their disease and in 57 (28.5%) it was the main disease manifestation. Among these 81 patients, seven (9%) were asymptomatic at diagnosis. Table 3A summarizes the findings concerning the timing of CNS involvement, while Table 3B presents comparative data from the literature. Figure 1 gives the actuarial probability of developing CNS metastases and the actual cumulative number of cases with CNS involvement in all patients. An actuarial probability of 57% at 24 months and of 63% at 48 months was found. The actual cumulative numbers of cases, at the same points of time, were 78 and 81, respectively. Table 2. Characteristics of the study population Number of patients 200 Men (%) 176 (81) Women (%) 24 (19) Median age (range) 58 (38-76) Limited disease (%) 68 (34) Extensive disease (%) 132 (66) Table 3A. Time of CNS involvement Patients n (%) Main manifestation (%) During staging 20 (10) 14 (7) During chemotherapy 14 (7) 8 (4) During follow-up 47 (23.5) 35 (17.5) Total number 81 (40.5) 57 (28.5) Table 3B. Comparative data from the literature concerning timing of CNS involvement Our study Pedersen [9] Nugent [13] Autopsy studies [9, 12] At presentation 10% 10% 14% During chemotherapy 7% 20% 22.5% During follow-up 23.5% Total number 40.5% 30%-40% 36% 65%

Alexopoulos, Vaslamatzis, Patila et al. 155 The findings concerning CNS involvement during staging and follow-up were then separately analyzed according to our policy in the use of brain CT scan. The results are shown in Table 4. During staging, comparison of the findings between 1980 and 1984 when brain CT scan was optional, and between 1985 and 1996 when it was obligatory, demonstrated no significant difference (p > 0.1). During follow-up, no significant difference was found in the period 1985-1991 when brain CT scan was performed systematically every six months and 1992-1996 when it was performed only if clinically indicated (p > 0.1). The development of CNS metastases in relation to the extent of disease and type of response to chemotherapy is presented in Table 5. No correlation was found with either parameter of disease. The location of CNS metastases was as follows: 66 (33%) were parenchymal lesions, 12 (6%) meningitis carcinomatosa, three (1.5%) spinal cord compression and three (1.5%) diabetes insipidus, in two of whom it was the only manifestation of CNS involvement (Table 6). PCI The development of CNS metastases in relation to PCI administration, in the 79 patients who achieved CR, is shown in Table 7A. There was no significant difference in Figure 1. Actuarial curve of developing CNS involvement and actual cumulative number of cases (histogram) in the 200 patients. the frequency of CNS involvement between PCI receivers and nonreceivers (p > 0.1). On the contrary, the frequency of developing CNS metastases as the sole site of relapse was significantly different between the two groups (p < 0.05). Comparative data from the literature are given in Table 7B. Actuarial development of CNS involvement by PCI receivers and nonreceivers is illustrated in Figure 2. The two curves are not different (p > 0.1). Finally, the actuarial survival of complete responders who received PCI was compared with that of complete responders who did not (Fig. 3) and a highly significant difference was found (p < 0.001) in favor of PCI receivers. A further comparison of the survival curves, separately for LD and ED (Figs. 4 and 5), demonstrated a significant difference (p < 0.001 ) only for LD. DISCUSSION CNS Involvement Although CNS involvement in SCLC is a well-known complication [1, 2, 4, 7, 8], we think our findings add useful information. Overall incidence of CNS involvement found was 40.5%, comparable with the 30%-40% calculated by Pedersen [9] and that reported by others [7, 10-12]. Table 5. CNS involvement in relation to stage and type of response to chemotherapy Patients n CNS involvement p value (% of total) (%) Limited disease 70 (35) 28 (40) Extensive disease 130 (65) 53 (41) p = NS Complete response 79 (39.5) 35 (44) Partial response 84 (42) 26 (31) p = NS Table 6. Location of CNS metastases and comparative data from the literature Site of metastasis Present study Nugent [13] Autopsy studies [12] Parenchymal lesions 33% 72% 28%-55% Meningitis carcinomatosa 6% 1.3% 3%-28% Pituitary metastasis 1.5% 12%-21% Spinal cord compression 1.5% 6.5% 7%-13% Table 4. CNS involvement during staging and follow-up according to the policy on the use of brain CT scan Time period Patients n Brain CT scan CNS involvement During 1980-1984 45 optional 5 (11%) staging 1985-1996 155 obligatory 14 (9%) During 1985-1991 99 every 6 months 29 (29%) follow-up 1992-1996 56 only if indicated 14 (25%) Table 7A. Incidence of CNS involvement according to PCI administration in the 79 patients who achieved complete remission Patients n (%) CNS relapse (%) CNS only (%) With PCI 51 (65) 25 (49) 12 (48) p > 0.1 p < 0.05 Without PCI 28 (35) 11 (39) 10 (91) Total 79 (100) 36 (45.5) 22 (61)

156 Role of Cranioprophylaxis in Small Cell Lung Cancer Table 7B. Comparative data from the literature concerning the frequency of CNS involvement in complete responders with and without PCI Patients n CNS relapse (%) CNS sole site (%) A B C A B C A B C With PCI 51 24 30 25 (49) 4 (17) 6 (20) 12 (50) NR 0 Without PCI 28 21 60 11 (39) 5 (24) 25 (42) 10 (91) NR 9 (36) Total number 79 45 90 36 (45.5) 9 (20) 31 (34) 22 (61) NR 9 (29) A= present study; B= Cox study [31]; C= Aroney s study [32]; NR= not reported. Figure 2. Actuarial curves of developing CNS involvement in 51 PCI receivers and in 28 nonreceivers. The actuarial curve of developing CNS metastases demonstrated was 57% at 24 months and 63% at 48 months, figures closer to the 50% found in autopsy studies [9, 12]. A tendency to increased probabilities with time has been reported by others [13, 14]. We also found a cumulative number of CNS involvement of 78 at two years and 81 at four years. These findings strongly suggest that the majority of CNS metastases develop during the first 24 months of follow-up and tend to level off after that time. A similar early development of CNS metastases was observed by Maurer and Livingston [14, 15]. During staging, the frequency of CNS involvement did not differ between optional and obligatory use of brain CT scan which raises the question of its necessity as a routine staging procedure. Similarly, during follow-up, no significant difference was found between 1985-1991 when brain CT scan was performed systematically every six months, and between 1992-1996 when it was performed only if clinically indicated. Therefore, in our department, we have adopted the policy of performing brain CT scan during the follow-up only if clinically indicated. Our finding that 10% of patients had CNS involvement at presentation is similar to that reported by others Figure 3. Actuarial survival of 51 CRs who received PCI versus 28 CRs who did not. [9, 13, 16]. Another 7% developed CNS metastases during chemotherapy, 4% as the main manifestation of progression. Pedersen has calculated that CNS metastases development during therapy is of the order of 20% [9], but during chemotherapy in our study strictly defines a five- to six-month period. A further 23.5% developed CNS metastases during follow-up, 17.5% as the main manifestation of disease recurrence. The frequency of CNS involvement found in LD was not different from that found in ED, indicating that ED is not associated with a higher risk of developing CNS metastases. An increased likelihood of CNS involvement at diagnosis in ED was observed by Nugent et al., but was not statistically significant [13]. We were impressed to find no difference in the frequency of CNS metastases between complete and partial responders. In fact, there was a trend toward higher frequency among complete responders. Similar findings have been reported by others [17] and are obviously due to the fact that complete responders, by living longer, are exposed to a higher risk than either partial responders or the general SCLC population [17, 18]. It was not of surprise that cerebral lesions were the most frequent site [12, 13, 19, 20] of CNS metastases, but

Alexopoulos, Vaslamatzis, Patila et al. 157 Figure 4. Actuarial survival of CRs with PCI versus CRs without PCI in the 35 patients with limited disease. the 6% incidence of meningitis carcinomatosa in our series is higher than reported earlier [16, 21]. Of note, in two autopsy series, the frequency of leptomeningeal metastases was 24% [4, 13]. We would also like to emphasize the fact that in all three cases with pituitary metastases (1.5%) we observed, diabetes insipidus constituted the presenting symptom and in two it was the sole site of CNS involvement. The observed frequency of spinal cord compression is among the lowest reported [4, 12, 13]. PCI Although it is now more than twenty years since the first advocation of PCI in SCLC [6], the debate about its current position in the management of SCLC still continues [18, 22, 23]. There are several reports on the subject based on nonrandomized and randomized trials [14, 24-30], but a review of the literature easily leads to the conclusion that more information is required before the role of PCI in SCLC can be defined. The present findings based on a close follow-up of 79 patients with SCLC who achieved a documented CR after polychemotherapy add some valuable information. Although this is not a randomized study, it contains data which are valid for comparison. The frequency of CNS metastases in the complete responders after PCI, in our series, was not significantly lower than that observed among the complete responders without PCI. Likewise, no significant difference was demonstrated in the actuarial probabilities of developing CNS metastases between the two groups. Although our findings are in accordance with two randomized [29, 31] and two nonrandomized studies [27, 32], they contrast with the findings of other investigators who constantly observed lower numbers of CNS relapses in the PCI receivers [14, 27, 32-34]. This discrepancy cannot be due to an insufficient radiation dose Figure 5. Actuarial survival of CRs with PCI versus CRs without PCI in the 44 patients with extensive disease. since we used a dose of 40 Gy/20f. More likely, it is due to a prolonged survival of complete responders in our series; namely, 46% and 26% at two and four years, respectively, compared with ~24% and 12% in most other series [14, 17, 30, 32]. Despite the lack of significant difference in the overall frequency of CNS involvement, the chances of CNS metastases being the sole site of disease relapse were significantly higher without PCI than with PCI. This finding underscores the value of PCI in delaying but not abolishing the occurrence of CNS metastases in patients with CR. Some investigators report very low rates of CNS metastases as the sole site of relapse in complete responders without PCI, and this discrepancy is possibly due to a short duration of CR in their series [35]. Among the 36 complete responders who developed CNS involvement in our series, 61% had no other evidence of recurrence at that time. Undoubtedly, the most interesting finding of our study is the significantly improved survival of complete responders after PCI. Forty-six percent of complete responders with PCI are projected to be alive at two years, while the probability of being alive at the same time interval without PCI is only 18%. While the latter figure is close to the usually reported two-year survival for complete responders [16, 36], a 46% two-year survival is among the highest reported [36, 37]. Our finding of a significantly improved survival with PCI is in sharp contrast with published results from randomized trials [14, 29, 30, 33, 38]. Although it can be argued that the present study was not a randomized one, thus allowing for a possible selection bias, this reason alone could not explain such a big difference in survival. We, therefore, think that the most likely explanation is a difference in disease-free and overall survival between our study and that of others. This probability is supported by: A) the impressively high actuarial survival of complete responders

158 Role of Cranioprophylaxis in Small Cell Lung Cancer with PCI in our series; B) the fact that 9 of 11 published randomized trials which failed to demonstrate an improvement in survival with PCI did not take into consideration the type of response to chemotherapy [18]; C) the very low number of isolated CNS recurrences reported by others, and D) the lack of significant difference in the frequency of CNS metastases between complete responders with and without PCI in our study. Very recently, Rubenstein also reported an improved survival in complete responders with PCI compared with complete responders without PCI. The two-year survival in his series was 46%, identical to ours [36]. Finally, it is worthy of emphasis that the improvement in survival after PCI proved statistically significant only in patients with LD. Improved survival of LD patients has been very recently reported in two other studies [36, 37]. This interesting observation clearly indicates that the stage of disease has to be incorporated into the inclusion criteria of trials investigating the impact of PCI on survival of patients with SCLC. From this point of view, it is of interest that among the four large ongoing randomized multicenter trials of PCI in SCLC being conducted among the membership of IASLC, one recruits patients with limited disease only [18, 39]. REFERENCES 1 Halpert B, Erickson EE, Fields WS. Intracranial involvement from carcinoma of the lung. Arch Pathol 1960;69:93-103. 2 Takita H, Brugarolas A, Marabella P et al. Small cell carcinoma of the lung. Clinicopathological studies. J Thorac Cardiovasc Surg 1973;66:472-477. 3 Auerbach O, Garfinkel L, Parks VR. Histologic type of lung cancer in relation to smoking habits, year of diagnosis and sites of metastases. Chest 1975;65:382-387. 4 Nugent JL, Bunn PA, Ihde DC et al. Diagnosis and treatment of CNS complications of small cell bronchogenic carcinoma (SCBC). Proc AACR and ASCO 1978;19:116a. 5 Sculier JP, Feld R, Evans WK et al. Neurologic disorders in patients with small cell lung cancer. Cancer 1987;60:2275-2283. 6 Hansen HH. Should initial treatment of small carcinoma include systemic chemotherapy and brain irradiation? Cancer Chemother Rep 1973;4:239-241. 7 Martini N, Wittes RE, Hilaris BS et al. Oat cell carcinoma of the lung. Clin Bull 1975;5:144-148. 8 Watson WL, Berg JW. Oat cell lung cancer. Cancer 1962;15:759-767. 9 Pedersen AG. Diagnosis of CNS-metastases from SCLC. In: Hansen HH, ed. Lung Cancer: Basic and Clinical Aspects. Dordrecht: Martinus Nijhoff 1986;153-182. Toxicity No prospective detailed neuropsychological evaluation was included in our study, and subclinical neurological or mild psychological abnormalities cannot be excluded. Nevertheless, no clinically significant shortterm or long-term toxicity was observed despite the fact that a number of PCI receivers are still alive after more than five years. CONCLUSIONS CNS involvement is a frequent manifestation of SCLC, and its frequency is increasing with lengthening survival. A 10% sector of patients has CNS involvement at presentation. Routine use of brain CT scan during staging is of questionable usefulness although its use if clinically indicated during follow-up seems reasonable. ED is not associated with higher incidence of CNS involvement. Complete responders demonstrate a trend in developing CNS involvement more frequently because they live longer. The finding of improved survival of complete responders who received PCI, although clear-cut in our study, must await confirmation from large randomized trials. Until that time, we think it is advisable to offer PCI to all complete responders with limited disease. 10 Watson WL, Deeley TI. The necropsy findings in carcinoma of the bronchus. Br J Chest 1971;65:228-242. 11 Line DH, Deeley TI. The necropsy findings in carcinoma of the bronchus. Br J Dis Chest 1971;65:238-242. 12 Bunn PA, Nugent JL, Matthews MJ. Central nervous system metastases in small cell bronchogenic carcinoma. Semin Oncol 1978;5:314-322. 13 Nugent JL, Bunn PA, Matthews MJ et al. CNS metastases in small cell bronchogenic carcinoma. Increasing frequency and changing pattern with lengthening survival. Cancer 1979;44:1885-1893. 14 Maurer LH, Tulloh M, Weiss RB et al. A randomized combined modality trial in small cell carcinoma of the lung: comparison of combination chemotherapy-radiation therapy versus cyclophosphamide-radiation therapy effects of maintenance chemotherapy and prophylactic whole brain irradiation. Cancer 1980;45:30-39. 15 Livingstone RB, Moore TN, Heilburn L et al. Small cell carcinoma of the lung: combined chemotherapy and radiation. Ann Intern Med 1978;88:194-199. 16 Ihde DC, Pass HI, Glatstein EJ. Small cell lung cancer. In: De Vita VT, Hellman S, Rosenberg SA, eds. Cancer: Principles and Practice of Oncology. 4th edition. Philadelphia: JB Lippincott Co., 1993;1:751-752.

Alexopoulos, Vaslamatzis, Patila et al. 159 17 Mira JG, Chen TT, Livingston RB et al. Outcome of prophylactic and therapeutic cranial irradiation in disseminated small cell lung carcinoma: a Southwest Oncology Group Study. Int J Radiat Oncol Biol Phys 1988;14:861-865. 18 Kristjansen PEG, Hansen HH. Prophylactic cranial irradiation in small cell lung cancer an update. Lung Cancer 1995;12(suppl 3):S23-S40. 19 Posner JB. Management of central nervous system metastases. Semin Oncol 1977;4:81-91. 20 Newman SJ, Hansen HH. Frequency, diagnosis and treatment of brain metastases in 247 consecutive patients with bronchogenic carcinoma. Cancer 1974;33:492-496. 21 Rosen ST, Aisner J, Makuch RW et al. Carcinomatous leptomeningitis in small cell lung cancer: a clinicopathologic review of the National Cancer Institute experience. Medicine 1982;61:43-45. 22 Kristjansen PEG. Should current management of small cell lung cancer include prophylactic cranial irradiation? Lung Cancer 1994;10 (suppl 1):S319-S329. 23 Shaw EG, Eagan RT, Jett JR et al. Prophylactic cranial irradiation in complete responders with small cell lung cancer: analysis of the Mayo Clinic and North Central Cancer Treatment Group data bases. J Clin Oncol 1994;12:2327-2332. 24 Byhardt RW, Libnoch JL, Cox JD et al. Local control of intrathoracic disease with chemotherapy and role of prophylactic cranial irradiation in small cell carcinoma of the lung. Cancer 1981;47:2239-2246. 25 Cox JD, Holoye PY, Byhardt RW et al. The role of thoracic and cranial irradiation for small cell carcinoma of the lung. Int J Radiat Oncol Biol Phys 1982;8:191-196. 26 Moore TN, Livingston R, Heilburn L et al. The effectiveness of prophylactic brain irradiation in small cell carcinoma of the lung. A Southwest Oncology Group Study. Cancer 1978;41:2149-2153. 27 Jackson DV, Richards F, Cooper MR et al. Prophylactic cranial irradiation in small cell carcinoma of the lung. A randomized study. J Am Med Assoc 1977;237:2730-2733. 28 Katsenis AT, Karpasitis N, Giannakakis D et al. Elective brain irradiation in patients with small cell carcinoma of the lung: preliminary report. Lung Cancer International Congress Series 558. Excerpta Med 1982;277-284. 29 Hansen HH, Dombernowsky P, Hirsch FR et al. Prophylactic irradiation in bronchogenic small cell anaplastic carcinoma. Cancer 1980;46:279-284. 30 Beiler DD, Kane RC, Bernath AM et al. Low dose elective brain irradiation in small cell carcinoma of the lung. Int J Radiat Oncol Biol Phys 1979;5:941-945. 31 Cox JD, Petrovich Z, Paig C et al. Prophylactic cranial irradiation in patients with inoperable carcinoma of the lung. Cancer 1978;42:1135-1140. 32 Aroney RS, Aisner J, Wesley MN et al. Value of prophylactic cranial irradiation given at complete remission in small cell lung carcinoma. Cancer Treat Rep 1983;67:675-682. 33 Seydel HG, Creech R, Pagano M et al. Prophylactic versus no brain irradiation in regional small cell lung carcinoma. Am J Clin Oncol 1985;8:218-223. 34 Neijstrom ES, Capizz RI, Rudnick SA. High dose Methotrexate in small cell lung cancer: lack of efficacy in preventing CNS relapse. Cancer 1981;51:1056-1061. 35 Elliot JA, Osterlind K, Hirsch FR et al. Metastatic pattern in small cell lung cancer: correlation of autopsy findings with clinical parameters in 537 patients. J Clin Oncol 1987;5:246-254. 36 Rubenstein JH, Dosoretz DE, Katin MJ et al. Low doses of prophylactic cranial irradiation effective in limited stage small cell carcinoma of the lung. Int J Radiat Oncol Biol Phys 1995;33:329-337. 37 Liengswangwong V, Bonner JA, Shaw EG. Prophylactic cranial irradiation in limited stage small cell lung cancer. Cancer 1995;75:1302-1309. 38 Arriagata R, Le-Chevalier T, Borie F et al. Prophylactic cranial irradiation for patients with small cell lung cancer in complete remission. J Natl Cancer Inst 1995;87:183-190. 39 Gregor A, Cull A, Stephens RJ et al. Prophylactic cranial irradiation (PCI) in small cell lung cancer (SCLC): first analysis of a UKCCCR/EORTC randomized trial (meeting abstract). Br J Cancer 1995;72(suppl 25):13.