Clinical factors affecting acquired resistance to gefitinib in previously treated Japanese patients with advanced nonsmall cell lung cancer

Abstract

BACKGROUND.

The risk factors for the development of acquired resistance in nonsmall cell lung cancer (NSCLC) patients responding to gefitinib are unclear. The current study assessed clinicopathologic factors affecting acquired resistance to gefitinib in previously treated patients with advanced NSCLC.

METHODS.

Between 2000 and 2004, 197 consecutive Japanese patients with advanced NSCLC underwent treatment with gefitinib. Of these patients, 56 who had continued gefitinib treatment without disease progression for at least 6 months were included in the study.

RESULTS.

At a median follow-up time of 21.6 months (range, 7.7–59.7 months), the median time to disease progression was 19.5 months, with progression-free survival rates of 68.5% at 1 year, 33.6% at 2 years, and 21.2% at 3 years. In a multivariate analysis using a Cox regression model, baseline brain metastasis was the strongest prognostic factor affecting acquired resistance to gefitinib (hazards ratio, 2.14; 95% confidence interval, 1.10- 4.17 [P = .025]). In addition, a decreased baseline hemoglobin level (P = .074) and the administration of >1 chemotherapy regimen before gefitinib treatment (P = .069) tended to be significant negative prognostic factors.

CONCLUSIONS.

In patients undergoing treatment with gefitinib, the presence of brain metastasis was strongly associated with the emergence of acquired resistance in the current series of NSCLC patients. The finding requires confirmation in a large cohort of patients with advanced NSCLC, including a non-Japanese/Asian population. Cancer 2006. © 2006 American Cancer Society.

Lung cancer became a leading cause of cancer death in industrialized countries in the 20th century.1 Nonsmall-cell lung cancer (NSCLC) accounts for approximately 80% of all cases of lung cancer. A metaanalysis reported in 1995 demonstrated that cisplatin-based chemotherapy produced a modest but significant survival benefit over best supportive care in patients with advanced NSCLC.2 In the 1990s, new agents including irinotecan, docetaxel, paclitaxel, vinorelbine, and gemcitabine became available for the treatment of NSCLC. Randomized trials using platinum in combination with new agents have revealed significant survival advantages over older combinations,3, 4 although many trials have in fact failed to reveal a survival advantage of such regimens over older combinations.5, 6 In addition, pivotal trials comparing platinum-based doublet chemotherapy in combination with new agents demonstrated no superiority of any such combinations in terms of response rate, survival, or quality of life.7, 8

Recently, molecular-targeted agents have been introduced to the treatment of NSCLC. Gefitinib is an orally active epidermal growth factor receptor (EGFR) tyrosine kinase inhibitor, which demonstrates activity in previously treated patients with NSCLC. In 2 large phase II trials of single-agent gefitinib in previously treated patients with NSCLC, the objective response rate was 12% to 18%, with higher response rates observed in females, those who had never smoked, patients with adenocarcinoma, and Japanese patients.9, 10 A subgroup analysis of the Iressa Survival Evaluation in Lung Cancer trial, comparing gefitinib monotherapy with best supportive care alone in patients with advanced NSCLC who had failed ≥1 regimens of chemotherapy, suggested that patients of Asian origin obtain a survival benefit from gefitinib.11

Recent translational research has identified interesting potential predictive biomarkers of EGFR tyrosine kinase inhibitors.12, 13 Somatic mutations within the EGFR tyrosine kinase domain have been associated with marked responses to gefitinib,12–18 with high frequencies of such mutations observed in females, those who had never smoked, patients with adenocarcinoma, and those of East Asian ethnicity.12, 14–16, 19 However, the correlation between EGFR mutation status and survival in patients receiving gefitinib remains controversial,14–19 and the detection of mutations in EGFR is still not commonly performed clinically. In addition, the association between EGFR expression or gene amplification and survival is less clear.14–16, 18

Despite the initial success of gefitinib in the treatment of NSCLC, a substantial proportion of patients ultimately exhibit disease progression. The mechanism underlying such resistance to gefitinib is still unknown, although it was recently reported that the presence of a second mutation, resulting in substitution of methionine for threonine at position 790 of EGFR, was identified in an EGFR-mutant, gefitinib-responsive recurrent NSCLC tumor specimen.20, 21 In addition, to our knowledge, no information has been obtained regarding clinical factors affecting acquired resistance to gefitinib in patients with NSCLC who have received gefitinib treatment for a long period of time. We report herein the results of our retrospective study assessing risk factors for the development of acquired resistance to gefitinib in previously treated patients with advanced NSCLC who had continued gefitinib treatment without disease progression for at least 6 months. In this study, acquired resistance to gefitinib was defined as a lack of disease progression during at least a 6-month period of treatment with gefitinib.

MATERIALS AND METHODS

Patients

This retrospective study was performed using a database for 56 patients with NSCLC who underwent treatment with gefitinib. The patients comprised a portion of 197 consecutive Japanese patients with NSCLC who had initiated treatment with gefitinib at the National Hospital Organization Shikoku Cancer Center or Okayama University Hospital between December 2000 and December 2004 and met the following criteria: 1) cytologically or histologically proven NSCLC; 2) previous chemotherapy including its use in an adjuvant setting; and 3) continuation of this treatment without disease progression for at least 6 months.

The selected patients included 22 males and 34 females whose median age was 62.5 years (range, 28–77 years) at the initiation of treatment. The Eastern Cooperative Oncology Group (ECOG) performance status (PS) was 0 in 15 patients, 1 in 31 patients, 2 in 8 patients, and 4 in 2 patients. Thirty-two patients had never smoked and 24 were former or current smokers. The majority of patients had adenocarcinoma, and only 4 had nonadenocarcinomatous histology (squamous cell carcinoma and nonsmall cell carcinoma not otherwise specified in 2 patients each). All patients underwent a series of examinations for restaging before the initiation of gefitinib treatment, which included computed tomography (CT) scans of the chest and upper abdomen, magnetic resonance imaging (MRI) of the brain, and radionuclide bone scan. International staging system [1997] Stage IIIA disease was found in 1 patient, Stage IIIB disease in 8 patients, and Stage IV disease in 47 patients. In addition, metastases to the brain, lung, bone, liver, adrenal, and pleura (dissemination and/or malignant effusion) were found in 26 patients, 37 patients, 23 patients, 2 patients, 3 patients, and 12 patients, respectively, in this study population. For the group of 44 patients who had target lesions defined by the Response Evaluation Criteria in Solid Tumors (RECIST),22 the median maximal tumor dimension was 3.2 cm (range, 1.2–11.0 cm). The median interval between the diagnosis of NSCLC and the initiation of treatment with gefitinib was 14.7 months (range, 1.1–89.4 months). During that period, 19 patients had undergone surgical resection and 5 had received thoracic radiotherapy. The numbers of chemotherapy regimens administered were 1 in 31 patients, 2 in 18, 3 in 6, and 4 in 1 patient. The best response to previous chemotherapy was a partial response (PR) in 17 patients, stable disease (SD) in 32 patients, and progressive disease (PD) in 7 patients. To further assess the medical condition of patients, baseline hemoglobin concentration, serum lactate dehydrogenate (LDH), albumin, and carcinoembryonic antigen (CEA) levels were included in this analysis; a decreased hemoglobin concentration (below the lower limit of normal in each institution) was found in 17 patients (30.3%), an increased LDH level (above the upper limit of normal) was noted in 8 patients (14.2%), a decreased albumin level (below the lower limit of normal) was found in 11 patients (19.6%), and an increased CEA level (higher than the median value of 21.7 ng/mL in this study population) was in found 28 patients (50.0%). The other routine laboratory test results such as white blood cell and platelet counts, as well as serum calcium, creatinine, and total bilirubin levels, were not included in this study because they were almost always within normal limits.

Response and Toxicity Evaluation

The RECIST guidelines were used for the evaluation of antitumor activity. The target lesion was defined as a tumor measuring ≥1 cm in longest dimension on helical CT scans. A complete response (CR) was defined as the disappearance of all clinically detectable tumor lesions lasting for at least 4 weeks. A PR was defined as at least 30% decrease in the sum of the longest dimensions of the target lesions for ≥4 weeks with no appearance of new lesions. PD indicated an at least 20% increase in the sum of the longest dimensions of the target lesions or the emergence of new lesions. SD was defined as insufficient decrease in tumor lesion to qualify for a PR and insufficient increase to qualify for PD. Disease control was also defined as the best tumor response of CR, PR, or SD that was confirmed and sustained for ≥4 weeks. Toxicity was graded according to version 2.0 of the National Cancer Institute Common Toxicity Criteria (Japan Clinical Oncology Group version).23

Statistical Analysis

Statistical analyses were performed using the SPSS software (version 11.0J; SPSS Inc., Chicago, IL). Progression-free survival time was defined as the period from the initiation of treatment with gefitinib to PD or last follow-up evaluation. Patients who died in the absence of PD were not censored at the time of death. Survival curves were calculated using the method of Kaplan and Meier and differences in survival distribution between 2 categorized groups were assessed using a log-rank test. To estimate the prognostic significance of covariates for survival, a Cox regression model was employed using a backward stepwise technique. Removal testing was based on the probability of likelihood ratio statistics based on maximum likelihood estimates. P values <.05 in 2-tailed analyses were considered significant.

RESULTS

Of the 56 patients with NSCLC evaluated in the current study, 44 had target lesions defined by the RECIST guidelines and 12 did not. Of the 44 patients with target lesions, 3 achieved a CR and 31 a PR, resulting in an overall response rate of 77.3% (95% confidence interval [95% CI], 64.9–89.7%). The remaining 10 patients had long-term SD. In addition, in 12 patients without target lesions, PD was not observed during a period of treatment of at least 6 months.

Risk Factors for the Development of Acquired Resistance to Gefitinib

At a median follow-up time of 21.6 months (range, 7.7–59.7 months), 38 of 56 patients evaluated experienced disease recurrence. The median time to disease progression was 19.5 months (95% CI, 15.4–23.6 months), with progression-free survival rates of 68.5% at 1 year, 33.6% at 2 years, and 21.2% at 3 years (Fig. 1). Progression-free survivals are summarized by pretreatment factors in Table 1. There were no differences noted in survival by either general pretreatment factors such as age, gender, or ECOG PS or exploratory factors such as maximum tumor dimension or best response to previous chemotherapy. In addition, as shown in Figure 2, there was no difference in survival noted based on response to gefitinib. However, patients with metastases to the brain (11.5 months vs. 23.6 months; P = .008), bone (13.8 months vs. 23.6 months; P = .025), liver (8.7 months vs. 19.5 months; P = .046), and adrenal gland (10.0 months vs. 19.5 months; P = .008) had significantly worse progression-free survival compared with patients without metastases to these sites. In addition, patients with a decreased baseline hemoglobin level (13.8 months vs. 23.1 months; P = .021) and decreased serum albumin level (13.8 months vs. 23.1 months; P = .017), as well as those who had received >1 cycle of chemotherapy (14.7 months vs. 23.6 months; P = .026) had significantly worse progression-free survival.

Figure 1. Progression-free survival curve from the time of initiation of treatment with gefitinib. The median time to disease progression was 19.5 months, with progression-free survival rates of 68.5% at 1 year, 33.6% at 2 years, and 21.2% at 3 years.

Figure 2. Progression-free survival curves for the 34 responders (solid line) and 10 patients with stable disease (dotted line). The median time to disease progression was 19.5 months in each group.

Table 1. Pretreatment Factors and Progression-Free Survival

Pretreatment factors	No. of patients	Median time to progression, months	P
ECOG indicates Eastern Cooperative Oncology Group; LLN, lower limit of normal; LDH, lactate dehydrogenase; ULN, upper limit of normal; CEA, carcinoembryonic antigen; PR, partial response; SD, stable disease; PD, progressive disease; NSCLC, nonsmall-cell lung cancer. Pleural metastasis indicates pleural dissemination and/or malignant effusion.
Age, y
<63	28	18.7	.888
≥63	28	19.5
Gender
Male	22	16.7	.706
Female	34	20.6
ECOG performance status
0–1	46	19.5	.903
2–4	10	13.8
Smoking history
Absence	32	20.6	.912
Presence	24	18.6
Histology
Adenocarcinoma	52	18.7	.952
Nonadenocarcinoma	4	19.5
Disease stage
Stage III	9	19.5	.934
Stage IV	47	19.7
Maximum tumor dimension
Nontarget lesion	12	9.2	.111
<3.2 cm	20	23.6
≥3.2 cm	24	15.8
Brain metastasis
Absence	30	23.6	.008
Presence	26	11.5
Pulmonary metastasis
Absence	19	16.7	.789
Presence	37	19.5
Bone metastasis
Absence	33	23.6	.025
Presence	23	13.8
Liver metastasis
Absence	54	19.5	.046
Presence	2	8.7
Adrenal metastasis
Absence	53	19.5	.008
Presence	3	10.0
Pleural metastasis
Absence	44	23.1	.105
Presence	12	16.7
Hemoglobin level
<LLN	17	13.8	.021
≥LLN	39	23.1
Serum LDH level
≤ULN	48	19.5	.945
>ULN	8	18.7
Serum albumin level
<LLN	11	13.8	.017
≥LLN	45	23.1
Serum CEA level
<21.7 ng/mL	28	23.6	.082
≥21.7 ng/mL	28	15.8
Surgical resection
No	37	18.7	.400
Yes	19	23.1
No. of chemotherapy regimens
1	31	23.6	.026
≥2	25	14.7
Best response to previous chemotherapy
PR	17	23.1	.124
SD/PD	39	16.7
Interval between the diagnosis of NSCLC and the initiation of treatment with gefitinib
<14.7 mo	28	18.6	.304
≥14.7 mo	28	19.5

Factors influencing progression-free survival were further assessed using a Cox regression model. All the parameters associated with a significant difference in survival on univariate analysis were included and analyzed using a backward stepwise technique. The finally selected model (χ² (3) = 14.36, P = .002) demonstrated that the presence of brain metastasis was the strongest prognostic factor in our series of NSCLC patients (hazards ratio, 2.14; 95% CI, 1.10–4.17 [P = .025]) (Table 2). In addition, a decreased baseline hemoglobin level and the administration of >1 chemotherapy regimen were found to be marginally significant factors (P = .074 and P = .069, respectively).

Table 2. Multivariate Analysis in Relation to Progression-Free Survival

Variables	Hazards ratio	95% confidence interval	P
Brain metastasis	2.14	1.10–4.17	.025
Hemoglobin level	1.91	0.94–3.87	.074
No. of chemotherapy regimens	1.85	0.95–3.61	.069

With respect to pattern of treatment failure of gefitinib, 30 of 38 patients (78.9%) who experienced disease recurrence had initial failure at the same sites as before the initiation of gefitinib treatment. In addition, 20 of 26 patients (76.9%) with baseline brain metastasis had experienced disease recurrence. However, initial failure in the brain was observed in only 7 patients. Of 26 patients with baseline brain metastasis, 15 received whole brain irradiation (6 patients), radiosurgery (8 patients), or both (1 patient) before gefitinib treatment, and 11 patients had not received any local treatment of the brain. The incidence of initial brain failure was 26.7% (n = 4 patients) in the previous local radiotherapy group, and was similar to that noted in the observation alone group (27.3%; n = 3 patients).

Treatment Compliance

Over a period of treatment with gefitinib of at least 6 months, 26 of 56 patients (46.4%) experienced an interruption of treatment (Table 3), the main reasons for which included Grade 2 or 3 liver dysfunction in 12 patients (21.4%) and Grade 2 or 3 skin toxicity in 9 patients (16.1%). In most patients, treatment with gefitinib was resumed after a rest period of <1 month. However, a reduction in the dose of gefitinib was thereafter required for only approximately 25% of patients; every second-day or third-day administration for 3 patients and 2 patients, respectively; and 2-week administration with a rest period of 2 weeks for 9 patients. There were no differences noted in progression-free survival by interruption of treatment (20.6 months [presence] vs. 18.6 months [absence]; P = .612) or reduction in dose (23.6 months vs. 18.6 months; P = .143) of gefitinib in this study population.

Table 3. Compliance in Treatment with Gefitinib

	No. (%) of patients
Treatment interruption	26 (46.4)
Reasons
Liver dysfunction	12 (21.4)
Skin toxicity	9 (16.1)
Diarrhea	1 (1.8)
Others	4 (7.1)
Duration, mo
<1	17 (30.4)
<2	6 (10.7)
≥2	3 (5.4)
Dose reduction
250 mg/every second day	14 (25.0)
250 mg/every third day	3 (5.4)
2 weeks administration/	2 (3.6)
2 weeks rest	9 (16.1)

DISCUSSION

In our retrospective study assessing risk factors for the development of acquired resistance to gefitinib in previously treated patients with advanced NSCLC who benefited from gefitinib treatment for at least 6 months, baseline brain metastasis was found to be an independent clinical predictor of the emergence of acquired resistance. In addition, a decreased baseline hemoglobin level and the administration of >1 chemotherapy regimen were found to be factors potentially affecting the development of acquired resistance. Although the patient selection criteria used in this study (e.g., continuation of gefitinib treatment without disease progression for at least 6 months) were somewhat arbitrary, they appear to be sufficient for the extraction of patients who benefited from gefitinib treatment from an entire population of advanced NSCLC patients, because previous studies have reported that median progression-free survival time ranged from 2.7 months to 3.0 months in patients with advanced NSCLC receiving gefitinib treatment,9, 11 and because subgroup analysis in 1 of these studies reported that median survival time was 5.5 months in patients of Asian ethnicity receiving placebo.11 In addition, our selection criteria included patients who exhibited long-term SD as a result of gefitinib treatment.

Despite the lack of conclusive evidence level, gefitinib is generally considered to be active against brain metastases in patients with advanced NSCLC. The objective response rate and median duration of response in patients with NSCLC and brain metastasis were reported to be 10% and 13.5 months, and 43% and 7.7 months, respectively, in 1 prospective and 1 retrospective series with small sample sizes.24, 25 Moreover, disease control at the site of brain metastasis was observed in 27% and 100% of patients, respectively. However, in the current study, the median progression-free survival was quite poor in patients with brain metastasis compared with those without metastasis, although only approximately 25% of the patients with brain metastasis experienced initial treatment failure in the brain. It is unclear why the majority of patients with baseline brain metastasis tended to develop acquired resistance to gefitinib with a lower frequency of brain failure. The finding that acquired resistance to gefitinib is more likely to develop outside of the central nervous system in patients with advanced NSCLC with brain metastasis should therefore be carefully reconfirmed, because it is possible that it was the result of our small sample size. In addition, it is possible that patients with extrathoracic lesions such as bone, liver, and adrenal metastases had a poor prognosis, although the numbers of such patients in the current study were too small to enable meaningful analysis. It appears that any extrathoracic site of disease portended inferior survival.

Anemia is generally considered an independent prognostic factor for patients with advanced NSCLC. Our finding of decreased baseline hemoglobin level is consistent with the results of our previous prognostic factor analysis using a recursive partitioning and amalgamation technique,26 as well as those reported by the Southwest Oncology Group.27 A potential association between a decreased baseline hemoglobin level and the development of acquired resistance to gefitinib was found in this study. Similar associations have been demonstrated both in patients with chronic myelogenous leukemia28, 29 and those with gastrointestinal stromal tumor30 undergoing treatment with imatinib, a low moleculer weight tyrosine kinase inhibitor of Bcr-Abl, Kit, and platelet-derived growth factor receptor. Although to our knowledge no findings have been obtained concerning the relation between hemoglobin level and the pharmacokinetics (PK) of gefitinib, a population PK study of imatinib suggested that a low hemoglobin level was associated with a low volume of distribution.31 This finding led Van Glabbeke et al.30 to speculate that the decreased distribution of imatinib to tumor due to a decreased baseline hemoglobin level might be correlated with the development of acquired resistance in patients with gastrointestinal stromal tumors. Therefore, there is a need for an examination of the association between PK profiles of gefitinib and hemoglobin level.

Finally, in the current study, the administration of >1 chemotherapy regimen was found to potentially affect the development of acquired resistance to gefitinib. A similar finding was obtained in a postmarketing trial comparing gefitinib with placebo in previously treated patients with advanced NSCLC.11 These findings indicate that the early integration of gefitinib treatment might improve efficacy throughout the course of treatment of advanced NSCLC. However, first-line treatment with gefitinib should be avoided at the current time, because a recent phase II trial of first-line, single-agent treatment with gefitinib reported an unacceptable frequency of lethal interstitial lung injury in Japanese patients.32 In addition, no survival benefit was found in 2 large phase III trials comparing standard platinum-based chemotherapy plus gefitinib with standard chemotherapy alone in chemotherapy-naive patients with advanced NSCLC.33, 34

In the current study, the presence of baseline brain metastasis was found to be the strongest clinical predictor of emergence of acquired resistance to gefitinib in previously treated Japanese patients with advanced NSCLC who benefited from gefitinib treatment for at least 6 months. In addition, a decreased baseline hemoglobin level and the administration of >1 chemotherapy regimen appeared potentially to affect the development of acquired resistance. These findings need to be confirmed in a large cohort of patients with advanced NSCLC, including a non-Japanese/Asian population. In addition, EGFR mutation and amplification status in NSCLC tumors for which gefitinib is effective need to be examined at the time of failure of gefitinib treatment to determine the mechanism of acquired resistance both clinically and biologically.