SEARCH

SEARCH BY CITATION

Keywords:

  • nonsmall cell lung carcinoma;
  • gefitinib;
  • brain metastasis;
  • leptomeningeal metastasis;
  • long-term follow-up;
  • disease recurrence

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

BACKGROUND

Gefitinib is an epidermal growth factor receptor tyrosine kinase inhibitor that induces an early and dramatic response in 10% of patients with advanced nonsmall cell lung carcinoma (NSCLC). Long- term outcome and patterns of disease recurrence after response have not been described.

METHODS

The authors evaluated 139 patients with NSCLC treated with gefitinib at Memorial Sloan-Kettering Cancer Center (New York, NY) between 1998 and 2002. They focused on patterns of disease recurrence, risk of brain metastases (BM) and leptomeningeal metastasis (LM), and long-term outcome after initial response to gefitinib.

RESULTS

Of the 139 patients treated with gefitinib, 21 (15%) achieved a partial response. The median age of the responders was 64 years (range, 38–87 years), the median Karnofsky performance score was 80 (range, 60–90), and 4 of the patients were men. All responders had adenocarcinoma. The central nervous system (CNS) was the initial site of disease recurrence in 7 (33%) patients (BM in 5 and LM in 2). In 9 (43%) patients, the initial site of disease recurrence was the lung and in 1 it was the liver and bone. Four (57%) of the patients with disease recurrence in the CNS had lung disease under control. BM also developed in 2 patients who had initial disease recurrence in the lungs. The actuarial 5-year incidence of CNS metastases was 60%. The median overall survival periods were 15 months and 23 months for patients with and without CNS metastases, respectively (P = 0.24).

CONCLUSIONS

The CNS was a frequent site of disease recurrence in patients with NSCLC after an initial response to gefitinib, regardless of disease control in the lungs. Patients should be carefully monitored for neurologic symptoms. Intrinsic resistance of metastatic clones, incomplete CNS penetrance of the drug, and longer survival are possible explanations for this high incidence. Cancer 2005. © 2005 American Cancer Society.

Although chemotherapy improves survival and quality of life in patients with advanced-stage nonsmall cell lung carcinoma (NSCLC), complete responses (CR) rarely occur.1 Inhibition of the epidermal growth factor receptor (EGFR) tyrosine kinase has emerged as a therapeutic option in patients with NSCLC. Gefitinib is an anilinoquinazoline that inhibits the EGFR tyrosine kinase and benefits nearly all individuals with mutations in exons 19 or 21 of the EGFR tyrosine kinase domain receptor and many others.2–4 Two Phase II trials demonstrated an objective radiographic response in 11–18% of unselected patients with advanced-stage NSCLC, accompanied by symptomatic relief and survival benefit in ≤ 40% of patients.5, 6 With prolonged follow-up, there have been anecdotal reports of an increased risk of brain (BM) and leptomeningeal metastases (LM) among patients who achieved disease control with gefitinib. The purpose of the current study is to evaluate the patterns of disease recurrence and long-term outcomes in a cohort of patients who initially experienced partial responses (PR) to gefitinib, with particular attention to the risk of central nervous system (CNS) metastases.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

We reviewed the medical records and imaging findings of patients with NSCLC treated with gefitinib at Memorial Sloan-Kettering Cancer Center (MSKCC) between 1998 and 2002. This cohort of patients was included in a retrospective study on response to gefitinib, previously reported.7 The medical records and imaging findings of 139 patients were reviewed, including 3 patients treated in a Phase I trial,8 6 patients who received treatment during the IDEAL5 study, and 122 patients who received treatment via the Expanded Access Program that provided gefitinib for patients with advanced-stage NSCLC before approval of the drug by the Food and Drug Administration. The pathology and radiology findings for all patients were reviewed at MSKCC. Patients with a radiographic CR or PR were identified and assessed for duration of response, site of initial disease recurrence, and development of CNS metastases (parenchymal BM and LM), both as initial site of disease recurrence and later in the course of disease. Survival was calculated using the Kaplan–Meier product-limit method from the start of treatment with gefitinib to date of death or last follow-up. Survival of patients with CNS metastases was compared with patients without CNS recurrence. Actuarial incidence of CNS metastasis was estimated by censoring patients at death or last follow-up. Confidence intervals were calculated at the 95% level (95% CI). Follow-up extended through September 1, 2004.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Patient Characteristics

An objective radiographic PR occurred in 21 of the 139 patients treated with gefitinib (15%; 95% CI, 9–21%). No patient had a CR. The median age of the responders was 64 years (range, 38–87 years), the median Karnofsky performance score was 80 (range, 60–90), and there were 4 men and 17 women. Eight patients had a history of smoking, whereas 13 did not. All patients had adenocarcinoma, and bronchoalveolar features were present in 9 of them. All patients had Stage IIIB or IV disease. Ten patients received ≥ 3 previous chemotherapy regimens, 6 patients received 2, 4 patients received 1, and 1 patient had not received any previous chemotherapy. Eight patients had previous bone metastases and four patients had previous parenchymal BMs treated with whole-brain radiotherapy (WBRT). Two of these patients had also been treated with radiosurgery. EGFR tyrosine kinase domain mutations in exons 19 and 21 were tested in 6 patients, and 4 of them had positive results. Two other patients only had exon 21 tested and no mutations were found.

Patterns of Disease Progression

To date, 17 patients have disease progression after a median follow-up of 27 months (range, 20–67 months). The median progression-free survival for the entire cohort of responding patients was 12 months (range, 3–67+) (Fig. 1). Three patients continue to receive therapy with gefitinib after ≥ 24, ≥ 32, and ≥ 67 months. One patient discontinued gefitinib because of toxicity and progression of comorbidities.

thumbnail image

Figure 1. Progression-free survival

Download figure to PowerPoint

The patterns of initial disease recurrence after response to gefitinib are summarized in Table 1. The first site of disease recurrence was the lung in 9 (43%) patients, the CNS in 7 (33%) patients, and both liver and bone in 1 patient. Parenchymal BMs occurred in 5 patients and LMs occurred in 2 patients with CNS recurrence. All were diagnosed after presenting with neurologic symptoms. For 4 (57%) of these patients, the CNS was the sole site of disease progression, including the 2 patients with LMs. The three remaining patients had concomitant disease progression in the lung. CNS metastases also developed in another 2 patients after discontinuation of gefitinib because of initial disease progression in the lung. The median time to development of CNS metastases was 13 months (range, 3–29 months). The actuarial 5-year incidence of BMs is 60% (Fig. 2). The median duration of response in the lungs for patients who did not develop CNS metastases was 10 months (range, 3–67+ months). The median overall survival period for the entire population of responders was 16 months (range, 8–67+ months). The median overall survival for patients with CNS metastases was 15 months, compared with 23 months for patients without CNS metastases (P = 0.24) (Fig. 3).

Table 1. Patterns of Initial Failure after Response to Gefitinib, by Location Involved
CharacteristicsNo. of patientsPercentage (95% CI)
  • 95% CI: 95% confidence interval; CNS: central nervous system.

  • *a

    Patients who developed central nervous system disease in the setting of good systemic control.

Lungs1257 (34–78)
CNS733 (15–57)
CNS only*a419 (5–42)
Other distant metastases10.5 (0–24)
No disease progression419 (5–42)
thumbnail image

Figure 2. Actuarial incidence of central nervous system metastases.

Download figure to PowerPoint

thumbnail image

Figure 3. Overall survival in patients by the development of central nervous system (CNS) metastases after the start of gefitinib therapy. Dashed line: no CNS metastases; straight line: CNS metastases present.

Download figure to PowerPoint

All seven patients who developed parenchymal BMs had multiple brain lesions and were treated with WBRT. None had a history of previous BMs. Three of these patients whose systemic disease remained controlled resumed gefitinib after WBRT. For two patients, disease progressed rapidly in the brain. The third patient was alive at 33 months of follow-up. The two patients with LM had a previous history of parenchymal BMs treated with WBRT. They were treated with temozolomide but died 3 and 4 weeks later, respectively, despite systemic disease control. The other 2 patients with parenchymal BMs before the initiation of gefitinib had disease progression in the lung after 9 and 10 months, respectively, but CNS metastases remained well controlled.

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Our patients exhibited a high prevalence of CNS metastases, both as initial disease recurrence (33%) and later in the course of disease (a 5-year actuarial incidence of 60%). Moreover, the majority of patients who had initial disease recurrence in the CNS had achieved good control of other disease sites. The development of CNS metastases may have had a negative impact on survival (a median of 15 months vs. 23 months for patients with and without CNS metastasis, respectively), although the difference was not statistically significant.

Patterns of disease recurrence in patients with Stage IV lung carcinoma have rarely been included in reports of new systemic treatments. Recent Phase III trials that demonstrated the benefits of chemotherapy in patients with advanced-stage lung carcinoma have not differentiated between CNS and non-CNS sites of disease recurrence. Conversely, the incidence of BMs has been widely reported in trials of patients with locally advanced NSCLC. Trials of chemotherapy after surgery or radiotherapy in patients with stage IIIA/IIIB NSCLC have found a 20–35% incidence of BMs after a median survival period of 21–36 months.9–13 We found a similar incidence and survival in our population composed almost exclusively of heavily pretreated patients with Stage IV disease. This suggests that improved local control may be associated with a high incidence of subsequent CNS metastases in patients with NSCLC, regardless of initial staging.

The explanation for the high incidence of CNS metastases found in our study is likely multifactorial. The main mechanism of CNS seeding is hematogenous dissemination. Lung carcinoma has an anatomic propensity to disseminate metastatic clones to the CNS. Although tumor cells originating from other locations have to bypass the lungs to reach the arterial circulation, lung carcinoma cells have direct access to the pulmonary venous system and, therefore, to the arterial circulation. The likelihood of these cells reaching the CNS is further increased by the preferential flow directed to the brain, which receives approximately 20% of the cardiac output.14 The relatively extended survival time provided by improved local control in our study may have magnified the effect of these and other events involved in the metastatic process. An increased incidence of CNS metastases has been observed in a variety of tumor types in association with prolonged survival.15 Our patients might be at particular risk if we consider that all of them potentially had active systemic tumors that could release metastatic clones throughout the course of their treatment.

Once reaching their target, tumor cells that find an appropriate environment that matches their biologic properties will develop into a full metastasis (the “seed and soil” hypothesis), explaining the predilection of certain histologic types for particular metastatic locations.14 The majority of studies reporting on the incidence of BMs in NSCLC have shown a higher prevalence of metastases in patients with adenocarcinoma histologies, which was the histologic diagnosis for all of our patients.13, 16, 17 Moreover, a higher prevalence of BMs has been reported in women, which is also the predominant gender (80%) in our study.12 Therefore, our population of patients preselected by their response to gefitinib has a high prevalence of known risk factors for developing CNS metastases. This raises the intriguing question as to whether a higher potential for developing BMs is an additional feature of NSCLC likely to respond to gefitinib, along with nonsmoking history, adenocarcinoma histology, gender, and the associated EGFR mutations. Unfortunately, none of our patients who developed BMs were candidates for surgical resection, which precluded a comparative molecular analysis of both primary and metastatic tissue specimens. Such comparison would also provide some insight into whether chemoresistance of CNS metastases is due to intrinsic properties of metastatic clones or to the presence of the blood-brain barrier. In fact, incomplete blood-brain barrier penetration has been the most accepted explanation for the high prevalence of BMs in patients with NSCLC who achieve good local control with current chemotherapy regimens, because these regimens have reduced the incidence of metastases in locations other than the brain, which argues against chemoresistance of the metastatic cells.10 To our knowledge, little is known regarding the penetrance of gefitinib through the blood-brain barrier in humans. Studies in mice have suggested that this drug does not cross an intact blood-brain barrier despite its low molecular weight,18 although a recent Phase II trial has demonstrated that gefitinib may have activity in preexisting BMs.19 This discrepancy may be partially explained by the finding that the barrier is disrupted to some extent in full-grown metastases, allowing better penetration of the drug, whereas microscopic metastases develop behind an intact blood-brain barrier and are not adequately treated by ongoing chemotherapy. Patients in our study did not develop metastases in other sites, which suggests that incomplete penetrance of the blood-brain barrier may be an important factor in the development of BMs in the setting of good locoregional control with gefitinib. However, this must be interpreted carefully because complete restaging was not performed in all patients with disease progression and non-CNS metastases may have been missed.

Our study is limited by the small sample of patients and by its retrospective design. However, if confirmed by larger prospective studies, these findings will have several implications. In addition to careful monitoring of patients regarding neurologic symptoms, the role of prophylactic WBRT may merit investigation. Moreover, efforts should be made to better understand the metastatic process in the CNS. If the role of the blood-brain barrier is confirmed, gefitinib and other candidates for targeted therapy should be studied further in terms of penetrance in the CNS. This might be of particular relevance for designing new Phase I trials of targeted therapy, because the paradigm has been shifting from establishing the maximum tolerated dose to establishing the optimal biologic dose, which may not achieve adequate CNS concentrations.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The authors thank Dr. Lisa DeAngelis for reviewing the current article.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES