Fax: (805) 541-1653
Risk of cerebral metastases and neurological death after pathological complete response to neoadjuvant therapy for locally advanced nonsmall-cell lung cancer
Clinical implications for the subsequent management of the brain
Article first published online: 6 MAR 2007
Copyright © 2007 American Cancer Society
Volume 109, Issue 8, pages 1668–1675, 15 April 2007
How to Cite
Chen, A. M., Jahan, T. M., Jablons, D. M., Garcia, J. and Larson, D. A. (2007), Risk of cerebral metastases and neurological death after pathological complete response to neoadjuvant therapy for locally advanced nonsmall-cell lung cancer. Cancer, 109: 1668–1675. doi: 10.1002/cncr.22565
- Issue published online: 4 APR 2007
- Article first published online: 6 MAR 2007
- Manuscript Accepted: 18 DEC 2006
- Manuscript Revised: 14 DEC 2006
- Manuscript Received: 31 OCT 2006
- brain metastases;
- complete response;
- neoadjuvant therapy;
- locally advanced;
- nonsmall-cell lung cancer
The incidence and pattern of brain metastases was analyzed among patients who achieved a pathological complete response (pCR) after neoadjuvant chemotherapy or chemoradiotherapy for locally advanced nonsmall-cell lung cancer (NSCLC).
Between 1990 and 2004, 211 patients were treated with neoadjuvant therapy before surgical resection for stage III NSCLC. The clinical course of 51 patients who demonstrated a pCR were reviewed. The neoadjuvant regimen consisted of either chemotherapy (29 patients) or chemoradiotherapy (22 patients). Histology was 45% adenocarcinoma, 41% squamous cell, and 14% large cell carcinoma. No patient received prophylactic cranial irradiation (PCI).
Overall survival at 1, 3, and 5 years was 82%, 63%, and 42%, respectively. The most common site of initial recurrence was the brain. Twenty-two (43%) patients developed brain metastasis as the site of first failure, which represented 71% of all isolated recurrences. Ultimately, 28 (55%) patients developed brain metastases at some point during their clinical course. The 5-year estimates of brain metastasis-free survival for patients with squamous and nonsquamous cancers were 57% and 34%, respectively (P = .02). Median survival from the time of brain metastasis was 10 and 5 months for those with isolated and nonisolated recurrences, respectively.
Patients with a pCR after multimodality therapy for locally advanced NSCLC are at excessively high risk for the subsequent development of brain metastases. Implications for management strategies including PCI and stereotactic radiosurgery (SRS) are discussed. Cancer 2007. © 2007 American Cancer Society.
Approximately 50% of patients with locally advanced nonsmall-cell lung cancer (NSCLC) will develop brain metastasis at some time during the course of their disease.1–3 This is the first site of recurrence in 15% to 40% of these patients, and autopsy studies have demonstrated that a substantial proportion of those who die of intrathoracic or extracranial disease progression from NSCLC also have synchronous brain metastasis.3–6 Several authors have therefore suggested that with improvements in overall survival and local-regional control from effective multimodality regimens incorporating chemotherapy with radiation and/or surgery the incidence and relative importance of brain metastases among patients with locally advanced NSCLC has visibly increased.6–9 This naturally raises the question of whether aggressive follow-up strategies, incorporating serial head imaging with stereotactic radiosurgery (SRS) and/or prophylactic cranial irradiation (PCI) to detect, prevent, and/or treat brain metastasis among asymptomatic patients after completion of local-regional therapy are warranted. Although randomized studies have failed to establish a survival benefit for PCI, subset analysis was not performed and it remains possible that selected patients, particularly those without evidence of disease at the primary or other extracranial sites, may significantly benefit from the prevention of brain metastasis.10–13 Likewise, some studies have suggested that early detection of brain metastasis from NSCLC and prompt treatment with SRS can significantly improve survival, particularly among those without evidence of extracranial disease.14, 15
Whereas the optimal local-regional treatment approach for locally advanced NSCLC is uncertain, the integration of surgery as a component of multimodality therapy may enable the identification of a subset of patients, those with a pathological complete response (pCR) to neoadjuvant chemotherapy or chemoradiotherapy, who may be at high risk for the subsequent development of brain metastases. The purpose of this analysis was to evaluate the outcome of these patients, focusing on the incidence and temporal presentation of intracranial metastases, to assess potential strategies for the management of the brain.
MATERIALS AND METHODS
This study was formally approved by the Committee on Human Research at the University of California, San Francisco (UCSF) before collection of all patient information. Between January 1990 and December 2004, 211 patients underwent definitive surgical resection after neoadjuvant treatment with chemotherapy or chemoradiotherapy for biopsy-proven, clinical stage IIIA and IIIB NSCLC. The medical records of 51 patients who exhibited a pCR, defined as having no evidence of residual tumor at the primary site or regional lymph nodes, at the time of surgery were reviewed to determine the clinical course and subsequent patterns of failure. The neoadjuvant regimen consisted of chemotherapy alone in 29 patients and chemoradiotherapy in 22 patients. No patient had a history of prior malignancy or cranial irradiation.
All patients were retrospectively staged in accordance with 2002 American Joint Committee on Cancer (AJCC) guidelines using disease characteristics at the time of diagnosis. All patients underwent computed tomography (CT) of the chest and upper abdomen. Forty-five (88%) patients underwent mediastinoscopy, and 18 (35%) patients underwent positron emission tomography (PET) as part of the initial staging work-up. Forty-four (86%) patients and 20 (39%) patients had a CT and magnetic resonance imaging (MRI) of the head, respectively, before therapy.
The chemotherapy agents used varied depending on the time period of this study as well as the discretion of the treating physicians. Chemotherapy was platinum-based in all patients. The most common regimens were cisplatin/etoposide (18 patients), carboplatin/taxol (13 patients), and carboplatin/etoposide (5 patients). The median number of chemotherapy cycles administered before surgery for those treated without radiation was 3 (range, 2–8). For those treated with neoadjuvant chemoradiotherapy the median number of chemotherapy cycles was 2 (range, 2–4).
Radiation was delivered using megavoltage equipment with once-daily, conventional fractionation. The median dose was 4500 cGy (range, 4200–7000 cGy). The planning target volume (PTV) was based primarily on the gross tumor extent as determined by CT. Three-dimensional (3D) planning was performed in 46 (90%) patients. For patients treated without 3D planning, target volumes were designed based on plain films taken at the time of fluoroscopy and were generally AP/PA with laterals with dose prescribed to midplane.
All patients were clinically reassessed approximately 2 weeks after completion of neoadjuvant therapy. The decision for surgical resection was typically made by a multidisciplinary tumor board after reviewing pre- and postneoadjuvant therapy CT images with consideration given to the patient's performance status and medical condition. Clinical response after neoadjuvant therapy was partial (greater than 50% reduction in the sum of the products of the perpendicular diameters of all measurable lesions) in 39 (76%) patients, and stable (<50% reduction in the sum of the products of the perpendicular diameters of all measurable lesions) in the remaining 12 (24%) patients. No patient had a clinical complete response (disappearance of all measurable tumor) or progressive disease.
The median time from completion of neoadjuvant therapy to definitive surgery was 28 days (range, 15–60 days). The type of procedure was dependent on the primary site, extent of disease, medical considerations, and the discretion of the surgeon. Pulmonary function tests were performed for all patients before surgery. In general, an attempt was made to maximize local control with preservation of functional outcome. Surgical resection encompassed the tumor extent at the time of operation rather than the preneoadjuvant therapy volume at the time of initial diagnosis. The surgical procedure consisted of lobectomy (30 patients), bilobectomy (14 patients), or pneumonectomy (7 patients). The number of lymph nodes sampled ranged from 4 to 36 (median, 16). Thirty-three (65%) patients received platinum-based chemotherapy postoperatively. None of the patients received postoperative radiation therapy or prophylactic brain irradiation (PCI).
Follow-up imaging consisted of CT of the thorax with or without PET scan. Imaging of the head with CT and/or MRI was performed at the discretion of the physician but in general was not routinely performed in the absence of symptoms. Median follow-up was 37 months (range, 3–85 months) for the entire population and 53 months among surviving patients (range, 12–85 months). None of the patients suffered early death without progression of NSCLC.
The primary endpoint analyzed was brain metastasis-free survival. Both the initial site of recurrence and all subsequent sites of recurrence were documented. Patients who died without evidence of brain metastasis were censored at their time of death in the analysis of brain recurrences. Secondary endpoints analyzed included overall survival and local-regional control. All endpoints were measured from the first day of treatment for all patients until death or date of last contact if still alive. One-, 3-, and 5-year estimates of survival were calculated by the Kaplan-Meier method, with comparisons between subsets performed with 2-sided log-rank tests.16 All tests were 2-tailed, with a P-value <.05 considered significant.
Table 1 outlines the clinical and disease characteristics for the 51 patients with a pCR after definitive treatment for locally advanced NSCLC. The median age was 55 years (range, 38–73 years). Distribution of initial T-stage was: 11 T1, 21 T2, 14 T3, 5 T4, with the median tumor size measuring 4.5 cm (range, 1.6–7.9 cm). Histology was: 23 adenocarcinoma, 21 squamous cell carcinoma, and 7 large cell.
|Characteristic||Total patients (%)|
|Initial primary size, cm|
Sixteen of the 51 patients with a pCR were alive at the time of analysis. Median survival for the entire patient population was 43 months. As illustrated in Figure 1, the overall survivals at 1, 3, and 5 years were 82%, 63%, and 42%, respectively. Thirty-one (61%) patients developed recurrent disease. Sites of initial failure included brain (22 patients), local-regional (5 patients), bone (3 patients), and liver (1 patient). Among the 22 patients with brain metastases as the site of initial recurrence, 2 subsequently experienced local-regional recurrences and an additional 2 developed bone metastases. The 5-year estimate of local-regional control was 86%.
A total of 28 patients (55%) developed brain metastases at some time during the course of follow-up. As illustrated in Figure 2, the 1-, 3-, and 5-year estimates of brain metastases-free survival for the entire patient population were 63%, 49%, and 43%, respectively. Figure 3 demonstrates brain metastases-free survival according to histology. Nineteen of 30 patients with nonsquamous NSCLC developed brain metastases compared with 9 of 21 patients with squamous cell histology. The 5-year estimates of brain metastases-free survival were 34% and 57%, respectively (P = .02). There was no significant difference in brain metastasis-free survival according to gender, age, initial T-stage, or neoadjuvant modality (P > .05, for all).
The median time to the development of brain metastases was 9 months (range, 3–47 months) for isolated recurrences and 13 months (range, 3–47 months) for all brain recurrences. Twenty-four of 28 patients were symptomatic from their brain metastases at the time of diagnosis. The number of brain metastases at diagnosis ranged from 1 to greater than 50 (median, 4). Details regarding the 28 patients who developed brain metastases are outlined in Table 2.
|Characteristic||Total patients (%)|
|Focal weakness||10 (36)|
|Mental status change||10 (36)|
|Speech difficulty||6 (21)|
|Visual disturbance||4 (14)|
|Number of brain metastases|
|Largest brain metastases, cm|
|Whole brain radiotherapy||27 (96)|
|Stereotactic radiosurgery||7 (25)|
|Surgical resection||4 (14)|
Twenty-one of the 28 patients with brain metastases from NSCLC had died at the time of analysis, including 17 of 22 with brain metastases as the site of first recurrence. Median survival from the time of brain metastasis was 10 months for isolated recurrences and 5 months for nonisolated cases. Treatment strategies included whole brain radiotherapy (27 patients), radiosurgery (7 patients), and surgical resection (4 patients). Some patients were treated with more than 1 modality for their brain metastases. Cause of death was assessable for 19 of 21 patients who died after developing brain metastases, including 17 of 17 patients with brain metastases as the site of initial recurrence. Overall, 88% (15 of 17) of patients with brain metastases as the site of initial recurrence died as a result of cranial disease, and at least 76% (16 of 21) of all patients who developed brain metastases died from neurological sequelae.
Among the 21 patients who died of brain metastases, 15 (71%) had no evidence of disease at extracranial sites.
The results of the present study not only confirm previous reports that patients with locally advanced NSCLC are at high risk for the development of brain metastases but also identifies a specific subset of patients who may derive particular benefit from aggressive management strategies to address this issue after completing local-regional therapy.1–6 Patients who achieve a pCR after neoadjuvant chemotherapy or chemoradiotherapy for locally advanced NSCLC have a risk of death from brain metastases that exceeds potential mortality due to extracranial disease progression. Similar to others, we found that this risk was more elevated in patients with nonsquamous cell types compared with those with squamous histology.17–19 Because the majority of patients who developed brain metastases in the present series died from their cranial disease and even more had neurological symptoms at their onset, we contend that vigilant surveillance incorporating serial brain MRI with SRS as needed or, alternatively, the judicious use of PCI may improve overall survival and quality of life among those who achieve a pCR after neoadjuvant therapy for locally advanced NSCLC.
It has been consistently reported that patients with a pCR after neoadjuvant therapy for locally advanced NSCLC have an improved prognosis.20–22 Our results are in accord with previous reports suggesting that the relative significance of brain metastases becomes more appreciable with improvements in local-regional control and survival for those with NSCLC.6–9 Tang et al.23 also showed that patients whose lungs showed a complete response to treatment had a significantly higher rate of developing brain recurrences than nonresponders (78% versus 8%). It has long been argued that as control of extracranial disease improves for patients with NSCLC, prevention and treatment of brain metastases assumes an increasingly important role because the blood-brain barrier provides sanctuary from the effects of cytotoxic chemotherapy. Cox et al.7 reviewed 4 prospective Radiation Therapy Oncology Group (RTOG) trials and concluded that the addition of chemotherapy to definitive radiation dramatically increased the incidence of brain metastases in patients treated for locally advanced NSCLC.
The most important observation from this study is the striking impact that cerebral metastases had on morbidity and mortality, which we believe can be attributed to the lack of follow-up screening and/or prevention strategies for brain metastases after completion of local-regional therapy. In particular, it is likely that earlier detection of cranial lesions with routine MRI, followed by SRS in selected patients, has the potential to significantly improve outcome in this setting.24–27 For patients who are not candidates for surveillance due to medical or social reasons, PCI may be a feasible alternative.
Although 3 randomized trials of PCI for locally advanced NSCLC have failed to demonstrate improved survival, irradiation significantly reduced the rate of developing brain metastasis.10–12 However, these results have been criticized because of patient heterogeneity and for possible imbalances in stratification according to established prognostic variables. It is unknown to what extent the development of local-regional failure or extracranial disease progression may have obscured any potential benefits that PCI may have conferred. Moreover, methods of local-regional treatment differed greatly from current multimodality approaches and only a small minority underwent surgery as a component of definitive therapy. For instance, the vast majority of patients enrolled on trials conducted by the Veterans Administration and the RTOG were considered unresectable and treated with radiation alone, leading some to suggest that death from extracranial processes may have overwhelmed any apparent benefit from PCI. A recent analysis of prospective data from the Southwest Oncology Group (SWOG) found that 39% of all brain metastasis from locally advanced NSCLC occurred after 6 months from starting treatment.17
Other series have also suggested that PCI may be beneficial in select patients with locally advanced NSCLC in which extracranial disease is controlled. Mamon et al.2 reported on 177 patients treated with neoadjuvant therapy and surgery. Whereas the local-regional control rate was an impressive 86%, it was notable that 40% of patients developed brain metastases at some point during follow-up. Additional analysis reveled that 34% of patients recurred in the brain as the first site of failure, which was associated with a 4-fold increase in death. Stuschke et al.1 compared the outcomes of 47 patients treated with PCI to 28 patients who did not receive PCI after neoadjuvant chemotherapy followed by chemoradiotherapy and surgery. The authors found that PCI significantly reduced the rate of brain metastases as the first site of recurrence from 30% to 8% at 4 years and that of overall brain recurrence from 54% to 13%. Notably, among patients with a complete or partial response to neoadjuvant chemotherapy, PCI reduced the rate of death by 50%, suggesting that patients displaying a favorable response to therapy may derive the most benefit from PCI. The weakness of these studies, similar to the present series, was that head imaging was neither routinely performed at diagnosis nor during follow-up in the absence of neurological symptoms. Furthermore, the majority of patients underwent imaging using CT, which is known to be much less sensitive than MRI for detecting early brain metastases.
It is important to recognize that the overwhelming majority of patients in the present series who developed central nervous system metastasis died from their brain involvement. Only 4 of 22 patients who developed brain metastases as the site of first recurrence subsequently developed extracranial disease, suggesting that the aggressive screening and treatment of brain metastases, as well as possibly prevention, has the real potential to improve survival. Wronski et al.28 similarly analyzed 231 patients with brain metastases from NSCLC, many of who did not have control of the primary site, and showed that neurological sequelae contributed to cause of death in approximately two-thirds of cases. Our finding that most patients with brain metastases from NSCLC presented with multiple parenchymal lesions at the time of recurrence, in the absence of screening, has also been reported by others. In the series by Stuske et al.1 only 1 of 8 patients who developed isolated brain metastasis after not receiving PCI presented with a single metastasis at initial recurrence, with most having 4 or more lesions. Nussbaum et al.29 also reported that less than 50% of patients have fewer than 4 metastasis at the time of initial brain recurrence.
Based on our current experience, the sensitivity of MRI to detect brain metastases among asymptomatic patients, and the effectiveness of SRS in treating the majority of these lesions, we recommend serial brain MRI every 3 months for patients who achieve a pCR after neoadjuvant therapy for locally advanced NSCLC. Given its minimally invasive nature and its tolerability with respect to neurotoxicity, SRS appears preferable to PCI, although prospective data are lacking. Until the results of ongoing studies attempting to correlate molecular biomarkers and gene products with clinical outcome in patients treated with neoadjuvant therapy for locally advanced NSCLC become applicable, it appears feasible to use pCR as a means of identifying patients who have a low enough risk of extracranial disease recurrence to benefit from vigilant surveillance or PCI.30–32 The fact that 7 of 51 of patients experienced local-regional failure and an additional 6 of 51 had extracranial distant failure after pCR, however, suggests that pCR is neither a surrogate for complete remission nor an entirely accurate predictor of disease response to therapy.
The primary limitation of this study was that due to its retrospective nature—we were unable to evaluate the outcomes of all 211 patients treated surgically after neoadjuvant therapy for locally advanced NSCLC regardless of pathological response status and thus are unable to determine whether the incidence of brain metastasis represented an increase relative to patients who did not attain pCR. Indeed, it is likely that certain subgroups among those who did not have pCR to neoadjuvant therapy may also derive benefit from aggressive surveillance or PCI. Similarly, due to the heterogeneity of our patient population with respect to pretreatment staging, neoadjuvant treatment (chemotherapy vs chemoradiotherapy), chemotherapeutic regimens, and posttreatment follow-up, we acknowledge that our data need to be validated through multi-institutional trials in which patient evaluation and treatment is more standardized. However, based on the results of a recently published review of 422 prospectively treated patients from the SWOG database, in which all subjects were required to have brain imaging before treatment for stage III NSCLC, it is unlikely that our conclusion would have been significantly altered.17 For instance, in the SWOG review, 27% of patients developed brain metastasis at any point after definitive treatment. Although data reporting the rates of brain metastasis and neurological sequelae according to response at the primary lung site are lacking, and therefore preclude any formal comparison among those with a pCR and non-pCR, our results clearly demonstrate that surgically treated patients who attain a pCR after neoadjuvant therapy for locally advanced NSCLC are at high risk for developing cerebral metastasis and should be strongly considered for strategies for the subsequent management of the brain.
Our findings, subject to the limitations discussed above, suggest that it may be justifiable to aggressively screen for brain metastasis, or alternatively to consider PCI after multimodality treatment for locally advanced NSCLC when the primary tumor is controlled and there is no evidence of systemic disease, because these patients, especially if they have tumors of nonsquamous histology, are at ostensibly high risk for the development of brain metastases. This is particularly relevant because with the advent of newer chemotherapy agents and more efficacious neoadjuvant regimens, the rate of pCR will continue to increase. Additional studies, including the ongoing trial by the RTOG that notably includes any patient with clinical evidence of local-regional control after therapy, could further refine patient selection for PCI in the future.
- 10Prophylactic cranial irradiation for lung cancer patients at high risk for development of cerebral metastasis: results of a prospective randomized trial conducted by the Radiation Therapy Oncology Group. Int J Radiat Oncol Biol Phys. 1991; 21: 637–643., , , et al.