Is Prophylactic Cranial Irradiation Indicated in Small-Cell Lung Cancer? By James F. Fleck, Lawrence H. Einhorn, Richard C. Lauer, Stephen M. Schultz, and Michael E. Miller Although prophylactic cranial irradiation (PCI) is frequently used in the treatment of patients with limitedextent small-cell lung cancer (SCLC), its role remains controversial. One hundred fourteen SCLC patients with limited disease treated at Indiana University were retrospectively reviewed. Fifty-eight of 114 (51%) patients achieved a complete remission (CR) and were analyzed. Thirty-eight of these 58 CR patients received PCI (+PCI) and 20 of 58 CR patients did not receive PCI (-PCI). Twenty-six of 38 patients who received PCI subsequently relapsed. No patient initially relapsed in the CNS, although one patient had a brain metastasis following recurrence in the chest. Eleven of 38 patients who were treated with PCI survived for longer than 30 months and were consid-
SMALL-CELL
lung cancer (SCLC) represents 20% of all cases of lung cancer1 and is highly responsive to chemotherapy and radiotherapy. 2 There is superior median and long-term survival in patients with limited disease (tumor confined to one hemithorax and encompassable within a single radiotherapy portal) compared with extensive disease. 3'4 However, despite the high partial (PR) and complete remission (CR) rates in limited SCLC, the majority of patients still relapse, with over 60% of recurrences occurring in the chest when chemotherapy is given as a single treatment modality.5 CNS metastases are an important cause of morbidity and mortality in patients with SCLC.6 Brain metastases occur in 10% of the patients at initial diagnosis." 9 If prophylactic cranial irradiation (PCI) is not used, over 50% of patients from autopsy series will have CNS metastases. 3 In one series, there was an incidence as high as 80% of CNS metastases in patients who did not receive PCI and survived for 2 years. 10 Several efforts have been made to decrease the incidence of CNS metastases and improve survival of patients with SCLC. The use of chemotherapy agents that cross the blood-brain barrier such as lomustine (CCNU), procarbazine, and high-dose methotrexate has been unsuccessful. 10 ' Therefore, most studies have used PCI with the expectation of reducing the incidence of CNS
ered long-term survivors. Seven of these 11 patients (63%)developed clinically significant neurological tox-
icity. Sixteen of 20 patients who did not receive PCI relapsed, but there was only one initial relapse in the CNS. Three additional patients who relapsed in the chest subsequently developed CNS metastasis. All responded to palliative radiation with improvement in their symptoms. The high incidence of CNS toxicity in the long-term survivors and the relatively infrequent incidence of isolated CNS recurrent in patients not subjected to PCI raise serious questions concerning the role, if any, of PCI in limited SCLC. J Clin Oncol 8:209-214. @ 1990 by American Society of Clinical Oncology.
metastases and the hope that such a reduction might translate to improved survival. This study is a retrospective analysis of the role of PCI in limited SCLC patients who achieved a CR with chemotherapy or chemoradiotherapy. MATERIALS AND METHODS One hundred fourteen patients with histologically or cytologically proven SCLC and staged as limited disease were treated at Indiana University from December 1978 to November 1986. The first 85 patients were placed into two consecutive Southeastern Cancer Study Group (SECSG) trials (LUN 328 and LUN 343) evaluating chemotherapy versus chemotherapy plus thoracic irradiation.'12 3 The most recent 29 patients have been treated off protocol with cyclophosphamide plus doxorubicin plus vincristine induction followed by consolidation with cisplatin plus etoposide (VP-16), based upon the positive results of late intensification with cisplatin
From the Department of Medicine, Section of Hematology and Oncology and Section of Biostatistics, Indiana University, Indianapolis,IN. Submitted February28, 1989; accepted June 14, 1989. Supported in partby National CancerInstitute GrantNo. 2R35 CA 39844-03;James F. Fleck, MD, is supportedby the National Research Council of Brazil (CNPq) Grant No. 203422/86 CL. Lawrence H. Einhorn is the Walther American Cancer Society Clinical Professorof Oncology. Address reprint requests to Lawrence H. Einhorn, MD, 926 W Michigan.Indiana University HospitalA-109, Indianapolis, IN 46223. © 1990 by American Society of Clinical Oncology. 0732-183X/90/0802-0002$3.00/0
Journal of Clinical Oncology, Vol 8, No 2 (February), 1990: pp 209-214
Downloaded from ascopubs.org by DFG on December 15, 2016 from 130.133.008.114 Copyright © 2016 American Society of Clinical Oncology. All rights reserved.
209
210
FLECK ET AL
plus VP-16 in SECSG LUN 343." We retrospectively analyzed the effect of PCI on the outcome of 58 patients who achieved a CR during this time period. Of the 58 CR patients, 38 received PCI as part of their treatment and 20 did not. Table 1 summarizes the distribution of the patients in the three consecutive studies. Limited disease was defined using the same criteria for all the three groups of study. Patients whose disease was more extensive than the involvement of the ipsilateral intrathoracic structures and/or ipsilateral scalene or supraclavicular lymph nodes were ineligible for retrospective analysis. Patients with ipsilateral pleural effusion or those who had a resection of their primary tumor at the time of diagnosis were considered eligible for study. Staging was performed based upon clinical examination, radiography of the chest, SMA- 12 chemistry profile, radioisotope or computed tomographic (CT) scanning of the brain and liver, bone scan, and bone marrow aspiration and biopsy. Standard criteria were used in the definition of CR.
Treatment Study no. 1. The first SECSG protocol (LUN 328) was activated in September 1978 and closed in February 1982. The protocol randomized patients to schedule A (cyclophosphamide, doxorubicin, and vincristine [CAV] plus PCI) or schedule B (CAV, chest irradiation, and PCI) for induction therapy. CAV in both arms consisted of cyclophosphamide 1,000 mg/m2 intravenously (IV), doxorubicin 40 mg/m2 IV, and vincristine 1 mg/m2 IV administered every 3 weeks for a total of six cycles. PCI was given to all patients during the induction treatment. The brain received 3,600 rad midline tumor dose delivered in a split course of 1,200 rad in five fractions (240 rad/day) given in weeks 5, 8, and 11. Two parallel opposing fields encompassed the whole brain. This was achieved by basing the fields along a line drawn from the upper margin of the eyebrow through the external auditory meatus, including the entire middle fossa of the foramen magnum. Twenty-seven patients who achieved a CR were analyzed, all of whom received PCI as per protocol. These patients, all of whom received PCI, have the longest follow-up. Study no. 2. The second SECSG protocol (LUN 343) was activated in June 1982 and closed in May 1985. Patients were first randomized into one of the two induction treatment arms: schedule A (CAV, chest irradiation, and PCI) or schedule B (CAV and PCI). At the end of the induction, patients in both arms who achieved a CR or PR were subsequently randomized for schedule C (late intensification) or schedule D (observation). The induction chemotherapy in both arms consisted of six cycles of CAV in the same doses previously described for LUN 328. Late intensification chemotherapy consisted of two cycles of VP-16 100 mg/m2 IV Table 1. Patient Distribution Study Number
Number of Patients
CR (%)
+ PCI
- PCI
1 (LUN 328) 2 (LUN 343) 3 (Off protocol)
42 43 29
27 of 42 (64) 18 of 43 (42) 13 of 29 (45)
27 11 0
0 7 13
114
58 of 114 (51)
38
20
Total
(days 1, 2, 3, and 4) and cisplatin 20 mg/m 2 IV (days 1, 2, 3, and 4) given every 4 weeks. PCI was given during thoracic irradiation or with the first course of CAV. The brain received 3,000 rad midline tumor dose delivered in a continuous course of 10 fractions (300 rad) and the fields were the same as described in the previous protocol (LUN 328). PCI initially was given to all patients, but subsequently it was optional. Seven of the 18 CR patients in this study did not receive PCI. Seven of the 11 patients receiving PCI and three of the seven not receiving PCI were consolidated with cisplatin and VP-16. The final patient population consisted of the most recent 29 patients with SCLC and limited disease treated at Indiana University. They were treated off protocol using the chemotherapy previously described with CAV induction followed by late intensification with cisplatin and VP-16. None of these patients received thoracic irradiation or PCI. The median age of patients receiving PCI was 59 years (range, 40 to 71) compared with 60 years for patients not receiving PCI (range, 50 to 76). Forty-two percent and 45% of the patients were female on the PCI and no-PCI arms, respectively. RESULTS
The effects of PCI were analyzed in the 58 patients with limited-disease SCLC who achieved a CR. PCI was administered to 38 of 58 CR patients and 20 of 58 patients did not receive PCI. Twenty-six of 38 CR patients who received PCI subsequently relapsed. Intrathoracic recurrence was the most predominant site and some patients had more than one metastatic foci when they first relapsed. None of these patients had CNS as their first site of relapse. One patient had CNS metastasis 8 months following a relapse in the chest. Sixteen of 20 CR patients who did not receive PCI subsequently relapsed. The predominant site of relapse again was the chest. One patient had CNS as the initial site of relapse. Three patients had CNS metastases following the chest relapse at 2, 2, and 5 months later, respectively. Table 2 shows the sites of relapse in the two groups of patients. As expected, there were fewer overall CNS relapses in the group receiving PCI (one of 38 v four of 20; P = .04, Fisher's exact test). However, under the assumption that the retrospective samples of + PCI and -PCI groups are comparable with respect to other health and demographicrelated characteristics, the log-rank test indicated no difference between the + PCI and - PCI groups for overall survival (P = .80) and diseasefree survival (P = .95). The Kaplan-Meier esti-
Downloaded from ascopubs.org by DFG on December 15, 2016 from 130.133.008.114 Copyright © 2016 American Society of Clinical Oncology. All rights reserved.
211
PROPHYLACTIC CRANIAL IRRADIATION IN SCLC Table 2. Sites of Relapse + PCI
Number of patients Total relapsed Chest Bone Liver Cervical lymph node Multiple, except CNS CNS, first relapse CNS, subsequent relapse
E
- PCI
38 26 12 4 2 2 6 0 1
.--- a
20 16 13 1 0 0 1 1 3
+PCI -PCI
-J
U, LL
0 >._
M
mate of the median survival time was 15 months on both arms and Brookmeyer and Crowley's 95% nonparametric confidence intervals for the median survival times are 12 to 21 months (PCI) and 11 to 23 months (no PCI), respectively. The survival curves based on a minimal follow-up of 24 months further illustrate the similar overall (Fig 1) and disease-free (Fig 2) survival experienced by the two groups. Long-term survival was defined as survival for 30 months or longer from initiation of chemotherapy. There were 11 such patients in the 38 CR patients included in the +PCI group and five in the 20 CR patients in the -PCI group. Seven of the 11 (63%) long-term survivors who received PCI had clinically significant CNS toxicity. In contrast, none of the five patients who did not receive PCI had this sequelae of treatment. 1.0O
-j
e----
+PCI
0-0
-PCI
0.8-
rr
o U0
0.6-
I-_J
m
0.4-
no
0c
a-
0.2-
0.0- A U
5
10
I
I
15
20
25
30
MONTHS Fig 1.
Disease-free survival.
35
40
m CO oC Q_
0 MONTHS Fig 2.
Overall survival.
Table 3 shows the eight long-term survivors treated with PCI who developed CNS toxicity. All seven patients were evaluated for concomitant medical problems; however, only their therapy could be implicated in the etiology of their CNS symptomatology. The predominant clinical manifestations of neurological toxicity observed in all these seven patients were memory deficiency and ataxia. Anatomical changes showing cerebral atrophy and ventricular dilatation were observed in the head CT scans of three of these seven patients (46%). Before starting therapy, their head CT scans were normal. Leukoencephalopathy (characterized by apathy, dysarthria, intermittent aphasia, memory loss, and gait ataxis) was observed in one patient. Serial head CT scans of this patient showed slow progressive cerebral atrophy and ventricular dilatation. Other manifestations of CNS toxicity included cranial nerve VII and XII dysfunction (one patient), optic atrophy, sensory neuropathy and lethary (one patient) as well as dizziness and diplopia (one patient). No acute CNS toxicity was observed. The median onset of CNS abnormalities was 24 months after PCI administration (range, 14 to 54 months). All the patients were less than 65 years old when treated with PCI. The subsequent follow-up of these seven patients with treatment-induced CNS toxicity
Downloaded from ascopubs.org by DFG on December 15, 2016 from 130.133.008.114 Copyright © 2016 American Society of Clinical Oncology. All rights reserved.
212
FLECK ET AL Table 3. Patients With CNS Toxicity Patient
Number
Onset of CNS
Age at PCI Survival (Yr) (Mo)
1 2 3 4 5 6
57 55 48 56 50 52
109+ 97+ 86+ 75+ 65+ 57+
7
61
56+
PCI in cGy (Fraction/Number) 3,600 3,600 3,600 3,600 3,000 3,000
Toxicity (Mo from PCI)
(240/15) (240/15) (240/15) (240/15) (300/10) (300/10)
54 42 47 14 15 18
3,000 (300/10)
24
showed progressive deterioration of symptoms. None of these patients is capable of employment; all are dependent upon nursing home or family care, with significant disability and considerable limitations in their lifestyle. The four patients of the -PCI group who had CNS relapse were treated with symptomatic irradiation using 10 fractions of 300 cGy (3,000 cGy) and dexamethasone 16 to 24 mg orally per day. The most frequent neurologic symptoms during CNS relapse were headache, nausea, and vomiting. The patient with CNS as the first site of relapse had a complete palliation of symptoms and signs after CNS treatment and died 7 months later with disseminated disease. Two of the three patients who developed CNS metastases following relapse in the chest had complete palliation of neurologic complaints living 4 and 3 months, respectively, and dying of progressive systemic disease. The third patient did not respond to palliative treatment and died with both systemic and CNS progressive disease 2 months later. Consequently, the majority of deaths occurring after CNS relapse were due to systemic rather than CNS progressive disease. DISCUSSION PCI is effective in reducing the CNS relapse rate in patients with SCLC; however, no improvement in survival has been demonstrated in several randomized studies.' 4-19 Despite the lack of consistent data showing improvement in survival, PCI is currently recommended in the treatment of most limited- and extensive-disease SCLC patients achieving CR.20 Initially, it was felt that PCI was both safe and effective. Recently, however, the safety of PCI has been questioned, especially when administered in conjunction with chemotherapy. Cranial irradiation might dam-
Type of CNS Toxicity Ataxia + memory deficiency Ataxia + memory deficiency Ataxia + memory deficiency Leukoencephalopathy Ataxia + memory deficiency + cranial nerve disfunction Ataxia + memory deficiency + optic atrophy + lethargy + sensory neuropathy Ataxia + memory deficiency + dizziness + diplopia
Abnormal Head CT Scan No No No Yes Yes Yes No
age the blood-brain barrier and increase the potential CNS toxic effects of systemic chemotherapy, or CNS irradiation itself may be deleterious.21,22 Acute and chronic neurological impairment have been reported in patients with SCLC treated with PCI. 23-26 A syndrome of tremor, ataxia, and memory loss has been described in patients with SCLC receiving combined modality treatment.27 Leukoencephalopathy was shown in patients treated with PCI 28 and methotrexate chemotherapy. 29 Cerebral parenchymal abnormalities have been reported in the CT scan even when no neurological clinical deterioration was detected. 27 The chance of developing CNS disease increased with lengthening survival, reaching 80% of cumulative actuarial probability by 28 months.1' However, this increased probability of CNS relapse was often accompanied by systemic progressive disease and only 4% of the patients had the CNS as the sole clinical evidence of recurrence. Furthermore, in a large autopsy series, only 3% of patients had the CNS as the sole site of recurrent disease.7 Another unknown issue with CNS recurrence is the mechanism of such relapses. It is quite possible that the majority of "isolated" CNS relapses represent seeding from a hitherto undetected intrathoracic relapse. If that hypothesis is correct, then PCI would not be expected to have an impact upon survival. CNS metastases have often been considered a poor prognostic indicator. However, comparison of survival curves of patients with positive and negative brain autopsies showed no difference in survival.7 The prognostic importance of isolated CNS metastases has never been adequately evaluated. Isolated reports have shown survival over 17 months in patients treated for CNS relapse whose systemic disease was under control.' 0
Downloaded from ascopubs.org by DFG on December 15, 2016 from 130.133.008.114 Copyright © 2016 American Society of Clinical Oncology. All rights reserved.
213
PROPHYLACTIC CRANIAL IRRADIATION IN SCLC
If PCI is not effective in improving survival, a benefit of PCI could be the prevention of signs and symptoms arising from CNS relapse. However, the palliation rate of symptomatic cranial irradiation ranges from 47% to 92%.'0,30 Cranial
palliative irradiation produced complete relief of neurologic signs and symptoms in three of our four patients with CNS metastases who did not receive PCI. There was a high incidence of CNS toxicity (63%) observed in PCI-treated long-term survivors (more than 30 months) in this retrospective analysis. The predominant manifestations of CNS toxicity were memory deficiency and ataxis. Cerebral atrophy and ventricular dilatation were observed in 46% of these patients. Leukoencephalopathy was diagnosed in one patient. All patients showed progressive deterioration of the CNS symptomatology. The simultaneous use of PCI and chemotherapy should be avoided, considering the potential interaction and related CNS toxicity. The best dose and fraction of PCI should be identified and compared with the results of palliative CNS irradiation in patients with isolated CNS relapse. Frequent head CT scans and neurologic and neuropsychologic evaluations should be included in the follow-up of both -PCI and +PCI groups evaluating, respectively, the risk of isolated CNS relapse and the potential time-, dose-, or fractionrelated CNS toxicity. There have been seven randomized studies evaluating PCI (Table 4). These studies included a heterogeneous patient population that was comprised of patients who did not achieve CR and of patients with extensive disease. There was a reduction in CNS relapse rate from an average
Table 4. PCI Randomized Trials
Investigator
Aisner et a132 Beiler et al14 5 Cox et al 6 Hirsch et al Jackson et a117 Maurer et al' Seydel et al" Total
CNS Relapse (%) PCI Dose Number of (cGy) Patients +PCI -PCI PValue Survival
3,000 2,400 2,000 4,000 3,000 3,000 3,000
29 54 45 111 29 163 271
0 0 17 9 0 4 5
36 16 24 13 27 18 21
702
6
20
SMALL-CELL
lung cancer (SCLC) represents 20% of all cases of lung cancer1 and is highly responsive to chemotherapy and radiotherapy. 2 There is superior median and long-term survival in patients with limited disease (tumor confined to one hemithorax and encompassable within a single radiotherapy portal) compared with extensive disease. 3'4 However, despite the high partial (PR) and complete remission (CR) rates in limited SCLC, the majority of patients still relapse, with over 60% of recurrences occurring in the chest when chemotherapy is given as a single treatment modality.5 CNS metastases are an important cause of morbidity and mortality in patients with SCLC.6 Brain metastases occur in 10% of the patients at initial diagnosis." 9 If prophylactic cranial irradiation (PCI) is not used, over 50% of patients from autopsy series will have CNS metastases. 3 In one series, there was an incidence as high as 80% of CNS metastases in patients who did not receive PCI and survived for 2 years. 10 Several efforts have been made to decrease the incidence of CNS metastases and improve survival of patients with SCLC. The use of chemotherapy agents that cross the blood-brain barrier such as lomustine (CCNU), procarbazine, and high-dose methotrexate has been unsuccessful. 10 ' Therefore, most studies have used PCI with the expectation of reducing the incidence of CNS
ered long-term survivors. Seven of these 11 patients (63%)developed clinically significant neurological tox-
icity. Sixteen of 20 patients who did not receive PCI relapsed, but there was only one initial relapse in the CNS. Three additional patients who relapsed in the chest subsequently developed CNS metastasis. All responded to palliative radiation with improvement in their symptoms. The high incidence of CNS toxicity in the long-term survivors and the relatively infrequent incidence of isolated CNS recurrent in patients not subjected to PCI raise serious questions concerning the role, if any, of PCI in limited SCLC. J Clin Oncol 8:209-214. @ 1990 by American Society of Clinical Oncology.
metastases and the hope that such a reduction might translate to improved survival. This study is a retrospective analysis of the role of PCI in limited SCLC patients who achieved a CR with chemotherapy or chemoradiotherapy. MATERIALS AND METHODS One hundred fourteen patients with histologically or cytologically proven SCLC and staged as limited disease were treated at Indiana University from December 1978 to November 1986. The first 85 patients were placed into two consecutive Southeastern Cancer Study Group (SECSG) trials (LUN 328 and LUN 343) evaluating chemotherapy versus chemotherapy plus thoracic irradiation.'12 3 The most recent 29 patients have been treated off protocol with cyclophosphamide plus doxorubicin plus vincristine induction followed by consolidation with cisplatin plus etoposide (VP-16), based upon the positive results of late intensification with cisplatin
From the Department of Medicine, Section of Hematology and Oncology and Section of Biostatistics, Indiana University, Indianapolis,IN. Submitted February28, 1989; accepted June 14, 1989. Supported in partby National CancerInstitute GrantNo. 2R35 CA 39844-03;James F. Fleck, MD, is supportedby the National Research Council of Brazil (CNPq) Grant No. 203422/86 CL. Lawrence H. Einhorn is the Walther American Cancer Society Clinical Professorof Oncology. Address reprint requests to Lawrence H. Einhorn, MD, 926 W Michigan.Indiana University HospitalA-109, Indianapolis, IN 46223. © 1990 by American Society of Clinical Oncology. 0732-183X/90/0802-0002$3.00/0
Journal of Clinical Oncology, Vol 8, No 2 (February), 1990: pp 209-214
Downloaded from ascopubs.org by DFG on December 15, 2016 from 130.133.008.114 Copyright © 2016 American Society of Clinical Oncology. All rights reserved.
209
210
FLECK ET AL
plus VP-16 in SECSG LUN 343." We retrospectively analyzed the effect of PCI on the outcome of 58 patients who achieved a CR during this time period. Of the 58 CR patients, 38 received PCI as part of their treatment and 20 did not. Table 1 summarizes the distribution of the patients in the three consecutive studies. Limited disease was defined using the same criteria for all the three groups of study. Patients whose disease was more extensive than the involvement of the ipsilateral intrathoracic structures and/or ipsilateral scalene or supraclavicular lymph nodes were ineligible for retrospective analysis. Patients with ipsilateral pleural effusion or those who had a resection of their primary tumor at the time of diagnosis were considered eligible for study. Staging was performed based upon clinical examination, radiography of the chest, SMA- 12 chemistry profile, radioisotope or computed tomographic (CT) scanning of the brain and liver, bone scan, and bone marrow aspiration and biopsy. Standard criteria were used in the definition of CR.
Treatment Study no. 1. The first SECSG protocol (LUN 328) was activated in September 1978 and closed in February 1982. The protocol randomized patients to schedule A (cyclophosphamide, doxorubicin, and vincristine [CAV] plus PCI) or schedule B (CAV, chest irradiation, and PCI) for induction therapy. CAV in both arms consisted of cyclophosphamide 1,000 mg/m2 intravenously (IV), doxorubicin 40 mg/m2 IV, and vincristine 1 mg/m2 IV administered every 3 weeks for a total of six cycles. PCI was given to all patients during the induction treatment. The brain received 3,600 rad midline tumor dose delivered in a split course of 1,200 rad in five fractions (240 rad/day) given in weeks 5, 8, and 11. Two parallel opposing fields encompassed the whole brain. This was achieved by basing the fields along a line drawn from the upper margin of the eyebrow through the external auditory meatus, including the entire middle fossa of the foramen magnum. Twenty-seven patients who achieved a CR were analyzed, all of whom received PCI as per protocol. These patients, all of whom received PCI, have the longest follow-up. Study no. 2. The second SECSG protocol (LUN 343) was activated in June 1982 and closed in May 1985. Patients were first randomized into one of the two induction treatment arms: schedule A (CAV, chest irradiation, and PCI) or schedule B (CAV and PCI). At the end of the induction, patients in both arms who achieved a CR or PR were subsequently randomized for schedule C (late intensification) or schedule D (observation). The induction chemotherapy in both arms consisted of six cycles of CAV in the same doses previously described for LUN 328. Late intensification chemotherapy consisted of two cycles of VP-16 100 mg/m2 IV Table 1. Patient Distribution Study Number
Number of Patients
CR (%)
+ PCI
- PCI
1 (LUN 328) 2 (LUN 343) 3 (Off protocol)
42 43 29
27 of 42 (64) 18 of 43 (42) 13 of 29 (45)
27 11 0
0 7 13
114
58 of 114 (51)
38
20
Total
(days 1, 2, 3, and 4) and cisplatin 20 mg/m 2 IV (days 1, 2, 3, and 4) given every 4 weeks. PCI was given during thoracic irradiation or with the first course of CAV. The brain received 3,000 rad midline tumor dose delivered in a continuous course of 10 fractions (300 rad) and the fields were the same as described in the previous protocol (LUN 328). PCI initially was given to all patients, but subsequently it was optional. Seven of the 18 CR patients in this study did not receive PCI. Seven of the 11 patients receiving PCI and three of the seven not receiving PCI were consolidated with cisplatin and VP-16. The final patient population consisted of the most recent 29 patients with SCLC and limited disease treated at Indiana University. They were treated off protocol using the chemotherapy previously described with CAV induction followed by late intensification with cisplatin and VP-16. None of these patients received thoracic irradiation or PCI. The median age of patients receiving PCI was 59 years (range, 40 to 71) compared with 60 years for patients not receiving PCI (range, 50 to 76). Forty-two percent and 45% of the patients were female on the PCI and no-PCI arms, respectively. RESULTS
The effects of PCI were analyzed in the 58 patients with limited-disease SCLC who achieved a CR. PCI was administered to 38 of 58 CR patients and 20 of 58 patients did not receive PCI. Twenty-six of 38 CR patients who received PCI subsequently relapsed. Intrathoracic recurrence was the most predominant site and some patients had more than one metastatic foci when they first relapsed. None of these patients had CNS as their first site of relapse. One patient had CNS metastasis 8 months following a relapse in the chest. Sixteen of 20 CR patients who did not receive PCI subsequently relapsed. The predominant site of relapse again was the chest. One patient had CNS as the initial site of relapse. Three patients had CNS metastases following the chest relapse at 2, 2, and 5 months later, respectively. Table 2 shows the sites of relapse in the two groups of patients. As expected, there were fewer overall CNS relapses in the group receiving PCI (one of 38 v four of 20; P = .04, Fisher's exact test). However, under the assumption that the retrospective samples of + PCI and -PCI groups are comparable with respect to other health and demographicrelated characteristics, the log-rank test indicated no difference between the + PCI and - PCI groups for overall survival (P = .80) and diseasefree survival (P = .95). The Kaplan-Meier esti-
Downloaded from ascopubs.org by DFG on December 15, 2016 from 130.133.008.114 Copyright © 2016 American Society of Clinical Oncology. All rights reserved.
211
PROPHYLACTIC CRANIAL IRRADIATION IN SCLC Table 2. Sites of Relapse + PCI
Number of patients Total relapsed Chest Bone Liver Cervical lymph node Multiple, except CNS CNS, first relapse CNS, subsequent relapse
E
- PCI
38 26 12 4 2 2 6 0 1
.--- a
20 16 13 1 0 0 1 1 3
+PCI -PCI
-J
U, LL
0 >._
M
mate of the median survival time was 15 months on both arms and Brookmeyer and Crowley's 95% nonparametric confidence intervals for the median survival times are 12 to 21 months (PCI) and 11 to 23 months (no PCI), respectively. The survival curves based on a minimal follow-up of 24 months further illustrate the similar overall (Fig 1) and disease-free (Fig 2) survival experienced by the two groups. Long-term survival was defined as survival for 30 months or longer from initiation of chemotherapy. There were 11 such patients in the 38 CR patients included in the +PCI group and five in the 20 CR patients in the -PCI group. Seven of the 11 (63%) long-term survivors who received PCI had clinically significant CNS toxicity. In contrast, none of the five patients who did not receive PCI had this sequelae of treatment. 1.0O
-j
e----
+PCI
0-0
-PCI
0.8-
rr
o U0
0.6-
I-_J
m
0.4-
no
0c
a-
0.2-
0.0- A U
5
10
I
I
15
20
25
30
MONTHS Fig 1.
Disease-free survival.
35
40
m CO oC Q_
0 MONTHS Fig 2.
Overall survival.
Table 3 shows the eight long-term survivors treated with PCI who developed CNS toxicity. All seven patients were evaluated for concomitant medical problems; however, only their therapy could be implicated in the etiology of their CNS symptomatology. The predominant clinical manifestations of neurological toxicity observed in all these seven patients were memory deficiency and ataxia. Anatomical changes showing cerebral atrophy and ventricular dilatation were observed in the head CT scans of three of these seven patients (46%). Before starting therapy, their head CT scans were normal. Leukoencephalopathy (characterized by apathy, dysarthria, intermittent aphasia, memory loss, and gait ataxis) was observed in one patient. Serial head CT scans of this patient showed slow progressive cerebral atrophy and ventricular dilatation. Other manifestations of CNS toxicity included cranial nerve VII and XII dysfunction (one patient), optic atrophy, sensory neuropathy and lethary (one patient) as well as dizziness and diplopia (one patient). No acute CNS toxicity was observed. The median onset of CNS abnormalities was 24 months after PCI administration (range, 14 to 54 months). All the patients were less than 65 years old when treated with PCI. The subsequent follow-up of these seven patients with treatment-induced CNS toxicity
Downloaded from ascopubs.org by DFG on December 15, 2016 from 130.133.008.114 Copyright © 2016 American Society of Clinical Oncology. All rights reserved.
212
FLECK ET AL Table 3. Patients With CNS Toxicity Patient
Number
Onset of CNS
Age at PCI Survival (Yr) (Mo)
1 2 3 4 5 6
57 55 48 56 50 52
109+ 97+ 86+ 75+ 65+ 57+
7
61
56+
PCI in cGy (Fraction/Number) 3,600 3,600 3,600 3,600 3,000 3,000
Toxicity (Mo from PCI)
(240/15) (240/15) (240/15) (240/15) (300/10) (300/10)
54 42 47 14 15 18
3,000 (300/10)
24
showed progressive deterioration of symptoms. None of these patients is capable of employment; all are dependent upon nursing home or family care, with significant disability and considerable limitations in their lifestyle. The four patients of the -PCI group who had CNS relapse were treated with symptomatic irradiation using 10 fractions of 300 cGy (3,000 cGy) and dexamethasone 16 to 24 mg orally per day. The most frequent neurologic symptoms during CNS relapse were headache, nausea, and vomiting. The patient with CNS as the first site of relapse had a complete palliation of symptoms and signs after CNS treatment and died 7 months later with disseminated disease. Two of the three patients who developed CNS metastases following relapse in the chest had complete palliation of neurologic complaints living 4 and 3 months, respectively, and dying of progressive systemic disease. The third patient did not respond to palliative treatment and died with both systemic and CNS progressive disease 2 months later. Consequently, the majority of deaths occurring after CNS relapse were due to systemic rather than CNS progressive disease. DISCUSSION PCI is effective in reducing the CNS relapse rate in patients with SCLC; however, no improvement in survival has been demonstrated in several randomized studies.' 4-19 Despite the lack of consistent data showing improvement in survival, PCI is currently recommended in the treatment of most limited- and extensive-disease SCLC patients achieving CR.20 Initially, it was felt that PCI was both safe and effective. Recently, however, the safety of PCI has been questioned, especially when administered in conjunction with chemotherapy. Cranial irradiation might dam-
Type of CNS Toxicity Ataxia + memory deficiency Ataxia + memory deficiency Ataxia + memory deficiency Leukoencephalopathy Ataxia + memory deficiency + cranial nerve disfunction Ataxia + memory deficiency + optic atrophy + lethargy + sensory neuropathy Ataxia + memory deficiency + dizziness + diplopia
Abnormal Head CT Scan No No No Yes Yes Yes No
age the blood-brain barrier and increase the potential CNS toxic effects of systemic chemotherapy, or CNS irradiation itself may be deleterious.21,22 Acute and chronic neurological impairment have been reported in patients with SCLC treated with PCI. 23-26 A syndrome of tremor, ataxia, and memory loss has been described in patients with SCLC receiving combined modality treatment.27 Leukoencephalopathy was shown in patients treated with PCI 28 and methotrexate chemotherapy. 29 Cerebral parenchymal abnormalities have been reported in the CT scan even when no neurological clinical deterioration was detected. 27 The chance of developing CNS disease increased with lengthening survival, reaching 80% of cumulative actuarial probability by 28 months.1' However, this increased probability of CNS relapse was often accompanied by systemic progressive disease and only 4% of the patients had the CNS as the sole clinical evidence of recurrence. Furthermore, in a large autopsy series, only 3% of patients had the CNS as the sole site of recurrent disease.7 Another unknown issue with CNS recurrence is the mechanism of such relapses. It is quite possible that the majority of "isolated" CNS relapses represent seeding from a hitherto undetected intrathoracic relapse. If that hypothesis is correct, then PCI would not be expected to have an impact upon survival. CNS metastases have often been considered a poor prognostic indicator. However, comparison of survival curves of patients with positive and negative brain autopsies showed no difference in survival.7 The prognostic importance of isolated CNS metastases has never been adequately evaluated. Isolated reports have shown survival over 17 months in patients treated for CNS relapse whose systemic disease was under control.' 0
Downloaded from ascopubs.org by DFG on December 15, 2016 from 130.133.008.114 Copyright © 2016 American Society of Clinical Oncology. All rights reserved.
213
PROPHYLACTIC CRANIAL IRRADIATION IN SCLC
If PCI is not effective in improving survival, a benefit of PCI could be the prevention of signs and symptoms arising from CNS relapse. However, the palliation rate of symptomatic cranial irradiation ranges from 47% to 92%.'0,30 Cranial
palliative irradiation produced complete relief of neurologic signs and symptoms in three of our four patients with CNS metastases who did not receive PCI. There was a high incidence of CNS toxicity (63%) observed in PCI-treated long-term survivors (more than 30 months) in this retrospective analysis. The predominant manifestations of CNS toxicity were memory deficiency and ataxis. Cerebral atrophy and ventricular dilatation were observed in 46% of these patients. Leukoencephalopathy was diagnosed in one patient. All patients showed progressive deterioration of the CNS symptomatology. The simultaneous use of PCI and chemotherapy should be avoided, considering the potential interaction and related CNS toxicity. The best dose and fraction of PCI should be identified and compared with the results of palliative CNS irradiation in patients with isolated CNS relapse. Frequent head CT scans and neurologic and neuropsychologic evaluations should be included in the follow-up of both -PCI and +PCI groups evaluating, respectively, the risk of isolated CNS relapse and the potential time-, dose-, or fractionrelated CNS toxicity. There have been seven randomized studies evaluating PCI (Table 4). These studies included a heterogeneous patient population that was comprised of patients who did not achieve CR and of patients with extensive disease. There was a reduction in CNS relapse rate from an average
Table 4. PCI Randomized Trials
Investigator
Aisner et a132 Beiler et al14 5 Cox et al 6 Hirsch et al Jackson et a117 Maurer et al' Seydel et al" Total
CNS Relapse (%) PCI Dose Number of (cGy) Patients +PCI -PCI PValue Survival
3,000 2,400 2,000 4,000 3,000 3,000 3,000
29 54 45 111 29 163 271
0 0 17 9 0 4 5
36 16 24 13 27 18 21
702
6
20
