Brain Metastasis after Prophylactic Cranial Irradiation in Patients with Small Cell Lung Cancer


brain metastases, small cell lung cancer, prophylactic cranial irradiation, chemotherapy, relapse rate, detection, management


  • Outline

  • Introduction 200

  • Brain Metastasis after Prophylactic Cranial Irradiation 201

    • How Often does Brain Relapse Occur after PCI? 201

    • Does the Timing of PCI Affect the Rate of Brain Relapse? 201

    • Does the Timing of Chest RT Affect the Rate of Brain Relapse after PCI? 204

    • Is the Rate of Brain Relapse Affected by the Dose of PCI? 205

    • When do they Occur after PCI? 205

    • How to Detect them? 206

  • Treatment Options 206

    • Stereotactic Radiotherapy 206

    • Whole Brain Re-Irradiation 207

    • Chemotherapy 208

  • Outcome and Conclusion 209

  • References


Small cell lung cancer (SCLC) accounts for about 15–20% of all lung cancers. It is both chemo- and radiosensitive. The brain is a common site of metastases. Unfortunately, because of the blood–brain barrier, chemotherapy may not be able to achieve a sufficient tumoricidal dose. Some patients achieve a complete response (CR) to treatment in the chest, only to develop brain metastases (BM) later. In 1974, Newman and Hansen reported that central nervous system (CNS) metastases were more common in SCLC than in other types of lung cancer, and postulated that CNS prophylaxis might decrease the frequency of CNS relapse (Hansen et al., 1973; ). Attempts to diminish the frequency of CNS relapse with chemotherapeutic agents, including those that cross the blood–brain barrier, have been unsuccessful even with modern chemotherapy. Prophylactic cranial irradiation (PCI), initially employed in leukemic patients, has been extended to SCLC. In the literature, researchers have tested the efficacy of PCI in patients with limited stage SCLC who achieved CR after combined chemotherapy and thoracic radiation. In a well-known study from Institut Gustave-Roussy in France, the 5-year cumulative rate of BM as an isolated first site of relapse was 37% in the control group and 20% in the PCI group ( p <0.001) ( ). The overall 5-year rates of BM were 59% and 43%, respectively [relative risk (RR) 0.50; p <0.001]. The 5-year overall survival rates were 15% in the control group and 18% in the PCI group (RR 0.84; p =0.06).

recently challenged the traditional idea that only patients with limited stage SCLC with CR benefit from PCI. In their report, PCI is also beneficial in extensive stage SCLC with any chemotherapy response. Irradiated patients had a lower risk of symptomatic BM (hazard ratio, 0.27; 95% confidence interval [CI], 0.16–0.44; p <0.001). The cumulative risk of BM within 1 year was 14.6% in the irradiation group (95% CI, 8.3–20.9) and 40.4% in the control group (95% CI, 32.1–48.6).

PCI is better than therapeutic cranial irradiation ( ), since the mean duration of survival is brief (approximately 4.5 months) after BM are detected. In a clinical trial with 370 patients, 50 relapsed initially in the brain and 20 in the liver ( ). Those who relapsed in the brain suffered a greater deterioration in performance status, and spent a greater proportion of their remaining life in hospital than did patients whose initial relapse was in the liver. This difference was most marked in patients who died soon after relapse. Radiotherapy (20 Gy in five fractions over one week or 30 Gy in 10 fractions over 2 weeks) and dexamethasone were not very effective treatments for brain relapse though subjective responses were common.

Another study also demonstrated that the palliative treatment of overt BM is relatively unsuccessful. Thirty-nine of 225 patients with SCLC developed BM after the initiation of chemotherapy ( ). Treatment with high-dose dexamethasone in all 39 patients and cranial irradiation in 32 patients resulted in a complete neurologic recovery in only eight of 39 patients (20%). Twenty-one of 39 patients (53%) failed to derive lasting benefit from their palliative treatment. Thirteen of 24 patients with limited disease with cranial relapse had no clinical evidence of other distant metastases prior to death and, in these patients, the CNS disease was an important cause of morbidity.

The outcome of BM after PCI will be even worse as the general condition of patients and responsiveness to further therapy would be worse after previous treatment. Factors affecting the relapse rate, detection and management of BM after PCI will be the focus of this chapter.

Brain Metastasis after Prophylactic Cranial Irradiation

How Often does Brain Relapse Occur after PCI?

The incidence varies from 6% to 33.3%, as seen in the following studies. An overview in 1988 that included nine earlier trials found fewer patients who received PCI had clinically detected BM (6% irradiated versus 22% non-irradiated patients) ( ). In a meta-analysis of seven randomized trials, 33.3% versus 58.6% of patients in the PCI group and control group had BM respectively ( ). Tai et al. found the brain relapse rate after PCI was 19% (24/128) patients with CR and 28% (11/40) patients with response less than CR ( ; Table 18.1 ). The authors pointed out that, in the past, most randomized studies were on complete responders after chemoradiation. Some studies did not give PCI to partial responders, while others did. Partial responders represent a heterogeneous group of patients with responses ranging from minimal to almost complete response. Partial responders do benefit from PCI with longer progression-free intervals and decreased rates of BM.

Table 18.1

Summary of Studies of Prophylactic Cranial Irradiation (PCI) in Small Cell Lung Cancer

Study Treatment Sequence Overall Survival Other Results
Rosen, prospective 1983 136 LD, 196 ED patients: PCI+
76 group 1: PCI 20 Gy/10f on day 1 of induction chemoRT 18–20%
80 group 2: PCI 24 Gy/8f at week 12 or 24 for both CR and PR PCI−
176 group 3: no PCI, with less intensive chemo 5% (2 y), p <0.005
Murray, NCIC 1993 308 LD patients: Initial ChRT 40%, High rate of BM at 2 y – initial ChRT 15.8%,
155 initial ChRT, 153 late ChRT (40 Gy/15f). PCI 25 Gy/10f to patients without progressive disease after chemo and ChRT late ChRT 34% (2 y), p =0.006 late ChRT 23.8% (2 y), p =0.12
Work, retrospective 1996 133 patients: 118 without BM at diagnosis MS Rates of BM
Group I – 23 CR with PCI 16.1 m 21.7%
Group II – 23 CR without PCI 13.8 m 26.1%
Group III – 72 not achieve CR, and no PCI 8.4 m (n.s.) 22.2% (n.s.)
Work, Aurhus Lung Cancer Group 1997 199 LD patients Initial ChRT 20%, BM – Initial ChRT 19%,
99 initial ChRT, 100 late ChRT. Initial PCI always given to initial ChRT patients. After 1984, initial PCI was given also to late ChRT group late ChRT 19% (2 y), p =0.4 late ChRT 13% (2 y), p =0.24
Used split course ChRT (20–22.5 Gy/11f separated by 21 days)
Auperin, meta-analysis 1999, 2000 847 LD, 140 ED patients in 7 randomized trials RR of
Time between start of induction PCI+ PCI− Death BM
therapy & randomization n = n =
<4 m 84 77 0.92 0.27
4–6 m 127 152 0.79 0.50
>6 m 102 91 1.01 0.69
p =0.39, 0.01
Schild, phase 2 NCCTG 2007 76 LD patients: 6 cycles of EP. PCI 25 Gy/10f during cycle 3 to responding patients. Cycles 4 and 5 included concurrent chemo and chest (30 Gy/20f BID, a 2-week break, and another 30 Gy/20f BID) 24% (5 y) Toxicity – 97%≥Gr 3, 80%≥Gr 4
Sas-Korczynska, phase 2 2010 41 LD patients: early PCI 30 Gy/15f, immediately after end of ChRT 54 Gy/27–30f, prior to last chemo (EP) cycles 25.5% (4 y) 3/41 (7.3%) vs 9/45 (20%) developed BM, p =0.009
45 late PCI 30 Gy/15f, after chemoRT
Ramlov, retrospective 2012 118 patients received PCI: 74 LD, 44 ED. Medium follow up 16.6 m MS: 5/21 of BM were within limbic system
21 (17.8%) had brain relapse, at 4–27 m all patients 16 m, LD 24 m,
ED 12 m
Tai, retrospective population database 2013 289 LD patients treated with sequential or concurrent chemoradiation: 177 CR, 88 IR MS: CR−PCI+21.6 (5.0–256.0) m Incidence of BM and median time to symptoms: CR−PCI+19% BM, at 21 m; IR−PCI+28% at 19 m
IR−PCI+14.3 (4.0–102.6) m

BM: brain metastases; chemo: chemotherapy; ChRT: chest radiation therapy; CR: complete response; ED: extensive disease; EP: etoposide and cisplatin; f: fractions; Gr: grade; Gy: Gray (radiotherapy absorbed dose); IR: incomplete response; LD: limited disease; m: months; MS: median survival; n : number of patients; NCCTG: North Central Cancer Treatment Group; NCIC: National Cancer Institute of Canada; n.s.: non-significant; PCI+: received prophylactic cranial irradiation; PCI−: not received prophylactic cranial irradiation; RR: relative risk; SWOG: Southwest Oncology Group; y: year.

Another retrospective study included 133 unselected patients with histologically proven SCLC from 1985 to 1990 ( ). The primary therapy consisted of combined radiochemotherapy or only chemotherapy. Sixteen patients were treated only by irradiation. The overall incidence of CNS metastases for all 133 patients was 33.1%. The incidence of new BM in the group of CR patients given PCI was 21.7%, with an average time to development of BM after primary diagnosis of 15.4 months.

Apart from radiotherapy or chemotherapy, there is a small subgroup of SCLC patients who are treated with surgery as the sole definitive local treatment modality. The controversy is whether all such patients or only those with more advanced stages need PCI. reported 13/29 patients without PCI developed BM at 19 months (13/29), while 0/17 patients who received PCI had BM at 4 years.

The incidence of BM would be slightly higher if the hippocampal sparing technique was used for PCI, as only 2.3% BM involves the hippocampus ( ). The authors identified 697 intracranial metastases in 107 patients after reviewing contrast-enhanced computerized tomography (CT) and/or magnetic resonance (MR) image sets for each patient. Patients had different histology (e.g. non-SCLC, SCLC, breast cancer, and others). There were 36 limbic metastases (5.2% of all metastases) identified in 22 patients who had a median of 16.5 metastases/patient. Sixteen metastases (2.29%) involved the hippocampus, and 20 (2.86%) involved the rest of the limbic circuit. Another contemporary study is from Denmark ( ). From 2007 to 2010 a total of 118 consecutive patients underwent PCI (25 Gy in 10 fractions). Five of the 21 patients had brain relapses after PCI within the limbic system. There is a Radiation Therapy Oncology Group (RTOG)-0933 study on the hippocampal sparing technique which has just completed accrual. Results are pending.

Does the Timing of PCI Affect the Rate of Brain Relapse?

The rate of BM relapse reported in the literature appears to depend on the timing of PCI. Proponents of early PCI state: (1) performed radiobiological estimation and felt that PCI started within 60 days of chemotherapy is more effective since we are dealing with smaller micrometastases in the brain. A smaller number of patients developed BM if PCI starts at the third chemotherapy cycle than by the completion of chemoradiation. (2) Radiotherapy helps to increase the permeability of the blood–brain barrier ( ). The remaining chemotherapy then can treat the brain. (3) Patients develop BM when chemoradiation is all completed – 13/40 (32.5%) were found to have brain relapse immediately before PCI ( ). reported the National Cancer Institute of Canada (NCIC) study results with a higher rate of BM at 2 years despite PCI, compared to other studies using early chest RT and PCI (see Table 18.1 ). The authors proposed that a difference in the brain failure rate may explain different long-term outcomes.

However, early PCI concurrent with chemotherapy is usually not delivered because of the following reasons: (1) neurocognitive functions deteriorated in a large prospective CALGB study, in which PCI was given concurrently with chemotherapy ( ). (2) Partial responders may not benefit from PCI. When PCI is given too early, oncologists do not have time to assess any response and may waste resources. (3) If CR had not been achieved in the chest, reseeding to the brain after PCI may occur. (4) PCI would delay the delivery of chemotherapy and reduce the dose-intensity of subsequent chemotherapy. The toxicity of PCI given between chemotherapy cycles is more than when given after all chemotherapy is finished. There will be a decrease in quality of life during the intensive treatment period. The general condition may decline after PCI and thoracic RT, if given early in the chemotherapy schedule.

found that patients who received PCI (Group I, early PCI and group II, late PCI, in Table 18.1 ) had significantly longer times to CNS relapse than the control patients of Group III with no PCI ( p <0.001 and 0.026, respectively), and patients in Group I tended to have prolonged time to CNS relapse compared with patients in Group II ( p =0.07). After adjustment was made for the most influential prognostic factors, stage of disease, and liver involvement, the overall significant differences remained among the three groups ( p =0.049). However, there was no longer a significant difference between the two groups of patients who received PCI at different times (Group I and Group II, p =0.23). In all groups, CNS relapse continued to occur over time until 24 to 30 months, when there was a suggestion of a plateau in the survival curve.

The more recent studies use modern chemotherapy and RT techniques. An example is the Polish study by . It had a median follow up of 19 (range: 4–135) months. None of their 86 patients with PCI developed neurotoxicity, despite 41 receiving PCI during the chemoradiotherapy period (see Table 18.1 ). In summary, the current Canadian standard of practice is to finish the induction chest radiotherapy and chemotherapy before starting PCI.

Does the Timing of Chest RT Affect the Rate of Brain Relapse after PCI?

reported a relatively large series which used sequential chemoradiation before 2001: radical chest radiotherapy, 16×2.5 Gy, concurrently with PCI of 15×2 Gy after CR ( n =111). After 2001, 40 patients received four to five cycles of chemotherapy concurrently with RT of 25×1.8 Gy, and PCI was applied to patients with CR. Median survivals of the two groups were 14.0 and 21.8 months, and 5-year overall survivals 10.5 and 26.9%, respectively. Frequencies of BM after PCI were 16.4% and 8.7%, respectively. Therefore, concurrent chemoradiation resulted in longer median survival and higher overall survival than sequential chemoradiation.

Is the Rate of Brain Relapse Affected by the Dose of PCI?

The randomized United Kingdom Coordinating Committee for Cancer Research (UKCCCR) and the European Organization for Research and Treatment of Cancer (EORTC) trial comparing no PCI, PCI with 24 Gy (in 2-Gy fractions) and PCI with 36 Gy (in 2-Gy fractions) suggested a possible dose-response effect ( ). The decrease in the incidence of BM with 24 Gy of PCI versus no PCI was not significant (hazard ratio [HR] 0.71). In contrast, there was a significant reduction in BM with 36 Gy of PCI compared to no PCI (HR 0.16).

A retrospective analysis of PCI doses demonstrated an approximately linear dose-response relationship for reduced BM with increasing dose ranging from 0 to 35 Gy (in 2-Gy fractions) ( ). Then a randomized trial was launched between September 1999 and December 2005. It recruited 720 patients with limited-stage SCLC in CR after chemotherapy and thoracic radiotherapy from 157 centers in 22 countries ( ). Patients were randomly assigned to a standard ( n =360, 25 Gy in 10 daily fractions of 2.5 Gy) or higher PCI total dose ( n =360, 36 Gy) delivered using either conventional (18 daily fractions of 2 Gy) or accelerated hyperfractionated (24 fractions in 16 days with two daily sessions of 1.5 Gy separated by a minimum interval of 6 hours) radiotherapy. There was no significant difference in the 2-year incidence of BM between the standard PCI dose group and the higher-dose group, at 29% and 23%. Therefore, the recommended radiation dose for PCI in most cancer centers is now 25 Gy in 10 daily fractions given over 2 weeks.

When do they Occur after PCI?

A German study reported that 30/125 limited stage SCLC patients (24%) developed BM after chemoradiation, of whom there are 10 CR, five partial response (PR), and 15 non-response (NR) of primary tumor after chemoradiotherapy ( ). Five of these 30 (17%) patients had PCI before. BM-free survival time for the entire cohort was 298 days (95% CI: 218–377): 567 days (95% CI: 322–811) in CR, 298 days (95% CI: 244–351) in PR and 252 days (95% CI: 217–286) in NR ( p <0.0001). PCI prolonged BM-free survival time in CR patients: 640 days (95% CI: 483–796) with PCI versus 482 days (95% CI: 111–926) without PCI ( p =0.047) versus 273 days (95% CI: 243–302) for PR and non-responders. The duration of BM-free survival was shown to correlate with long-term outcome ( p <0.0001).

The timeline for BM is interesting. Among the 332 patients with SCLC treated on National Cancer Institute protocols between 1970 and 1980, there were 33 cases of CNS failure in patients with CR, of whom two developed in the first 6 months after therapy, but all occurred within 30 months ( ). The most striking differences were seen after 30 months, when 70–75% of patients who received PCI, but only 50% of patients who did not receive PCI, were free from CNS metastases.

observed that brain relapse after PCI occurred earlier than expected from published clinical data of tumor doubling times. Between August 1981 and December 1983, 30 patients with SCLC treated by induction chemotherapy, followed by high-dose cyclophosphamide (7 mg/m 2 ), were selected for PCI on the basis of clinical, radiologic, and bronchoscopic CR. Pretreatment CT brain scans were normal in all patients, and 20 Gy mid-plane dose in five daily fractions were delivered by lateral fields to whole brain. Progressive focal neurologic signs of BM developed in 7/30 (23%) patients 3–11 months after PCI, confirmed on CT scanning in four patients. Relapse at other sites, predominantly the thorax, occurred in all seven patients at intervals of 1–6 months prior to CNS relapse. It is likely that the usual dose of PCI was inadequate to eradicate microscopic tumor deposits in the brain. Unfortunately, higher doses cannot be used due to concerns of toxicity, as dementia will affect the quality of life and patients can become institutionalized.

How to Detect them?

Currently, brain imaging is often not part of the routine in many countries for SCLC follow up post-therapy. BM are generally found when there are clinical symptoms and then scans are ordered to confirm clinical suspicion. It is impossible to estimate the cost-effective analysis of routine brain CT or MRI as most of these patients are elderly and even if BM are detected, treatment would not result in a very long quality-adjusted survival time.

Only gold members can continue reading. Log In or Register to continue

Feb 5, 2019 | Posted by in NEUROLOGY | Comments Off on Brain Metastasis after Prophylactic Cranial Irradiation in Patients with Small Cell Lung Cancer
Premium Wordpress Themes by UFO Themes