SEARCH

SEARCH BY CITATION

Keywords:

  • nonsmall cell lung cancer;
  • adenocarcinomas;
  • EGFR;
  • KRAS;
  • survival;
  • prognostic factors

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

BACKGROUND:

Lung adenocarcinomas can be distinguished by identifying mutated driver oncogenes, including epidermal growth factor receptor (EGFR) and KRAS. Mutations in EGFR are associated with both improved survival as well as response to treatment with erlotinib and gefitinib. However, the prognostic significance of KRAS has not been evaluated in large numbers of patients and remains controversial. For the current report, the authors examined the association of EGFR and KRAS mutations with survival among patients with advanced lung adenocarcinomas.

METHODS:

Data were analyzed from patients with advanced lung adenocarcinomas who had known EGFR and KRAS mutation status evaluated between 2002 and 2009. The collected clinical variables included age, sex, Karnofsky performance status, smoking history, and treatment history. Overall survival from the diagnosis of advanced disease was analyzed using Kaplan-Meier and Cox proportional hazard methods.

RESULTS:

In total, 1036 patients were evaluated, including 610 women (59%) and 344 never-smokers (33%). The median patient age was 65 years (range, 25-92 years), and the majority of patients (81%) had a Karnofsky performance status ≥80%. In multivariate analysis, EGFR mutations were associated with longer overall survival (hazard ratio, 0.6; P < .001), and KRAS mutations were associated with shorter survival (hazard ratio, 1.21; P = .048).

CONCLUSIONS:

KRAS mutations predicted shorter survival for patients with advanced lung adenocarcinomas. The presence of EGFR and KRAS mutations define distinct subsets of patients with lung adenocarcinomas and should be determined in patients when they are diagnosed with advanced disease. Clinical trial reports should include EGFR and KRAS mutation status along with other prognostic factors. Cancer 2013. © 2012 American Cancer Society.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

Recent therapeutic advances have highlighted the molecular heterogeneity underlying oncogene-driven lung adenocarcinomas. For example, patients with mutations in the epidermal growth factor receptor (EGFR) make up approximately 10% to 15% of patients in the United States who are diagnosed with advanced lung adenocarcinoma. These patients have high rates of radiographic response when they receive treatment with erlotinib or gefitinib, tyrosine kinase inhibitors (TKIs) that target EGFR; and they have longer progression-free survival compared with patients who have EGFR wild-type tumors.1-4

Kirsten rat sarcoma 2 viral oncogene homolog (KRAS) mutations are present in a larger number of patients: approximately 30% of patients with advanced lung adenocarcinoma in the United States. Efforts to develop effective therapies that inhibit lung cancers with KRAS mutations have been largely unsuccessful, and the prognostic significance of KRAS mutations remains in question. Several small studies5-9 and 1 meta-analysis10 evaluated the prognostic effects of KRAS mutations with conflicting conclusions. However, none of those studies adequately accounted for other prognostic factors, and few of the included patients received the modern chemotherapy regimens now considered standard. We hypothesized that, among patients with stage IV lung adenocarcinomas, EGFR and KRAS mutations would identify patients with different outcomes. Here, we present the largest analysis to date examining a population of patients with advanced lung adenocarcinomas and known EGFR and KRAS mutation status. We report clinical characteristics, treatment histories, and mutation analysis from 1036 patients and investigate their association with survival.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

Study Population

All patients who were evaluated at Memorial Sloan-Kettering Cancer Center (MSKCC) in the Thoracic Oncology Service clinics between 2002 and 2009 were analyzed. Only those patients with stage IV lung adenocarcinoma (American Joint Committee on Cancer Cancer Staging Manual, seventh edition) whose tumors had undergone routine analysis for EGFR and KRAS mutations were included. Patients with early stage lung adenocarcinoma who subsequently developed advanced disease were not included in this analysis. Molecular analysis of all tumors for EGFR began in 2004, so tumors from 2002 and 2003 underwent analyses retrospectively. We obtained permission from the MSKCC Institutional Review Board and Privacy Board for the retrospective chart review.

Mutational Analysis

Genomic DNA was extracted from tumor specimens. EGFR mutations were assessed by polymerase chain reaction (PCR)-based methods that detect exon 19 deletions and exon 21 leucine-to-arginine codon 858 (L858R) amino acid substitutions or by mass spectrometry-based genotyping (Sequenom, Inc., San Diego, Calif), as described previously.11, 12 Testing for these 2 major EGFR mutations identifies >90% of patients with sensitizing mutations. KRAS mutations in codons 12 and 13 were assessed by direct sequencing of exon 2 or by mass spectrometry-based genotyping (Sequenom, Inc.).12, 13 Testing for KRAS exon 2 mutations identifies >95% of patients who have lung cancer with KRAS mutations.

Data Collection

We collected clinical variables for all patients from medical records, including age, sex, and Karnofsky performance status (KPS). We obtained smoking status (never, former, or current cigarette use) using self-reported smoking questionnaires. Never-smokers were defined as those patients who smoked less than 100 cigarettes in a lifetime. Current smokers were those who were smoking at the time of diagnosis or who quit less than 1 year before diagnosis. Patient treatment histories were recorded, including receipt of EGFR-TKIs, platinum-based chemotherapy, or bevacizumab. Patients with unknown treatment histories were excluded from treatment analyses.

Statistical Analysis

Survival was calculated from the date of diagnosis of metastatic disease until the date of death. At the last available follow-up, all patients who were still living were censored. Patient demographics, clinical characteristics, and treatment histories were compared using the chi-square test. Overall survival (OS) was estimated using the Kaplan-Meier method and was compared across groups using the log-rank test (for univariate analysis) and Cox proportion hazard methods (for multivariate analysis).

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

Clinical Characteristics

Among 1036 patients who had advanced lung adenocarcinomas evaluated, 275 tumors (27%) harbored EGFR mutations, 241 tumors (23%) had KRAS mutations, and 520 tumors (50%) were wild-type for both KRAS and EGFR (designated KRAS/EGFR wild-type). Clinical characteristics were similar across the 3 subgroups (Table 1). The median age was 65 years (range, 25-92 years), and median KPS was 80%. There was variation in sex across mutation subgroups, and women accounted for 65% of patients who had EGFR mutations (179 of 275 patients; 95% confidence interval [CI], 59%-71%), 60% of patients who had KRAS mutations (144 of 241 patients; 95% CI, 53%-66%), and 55% of patients without KRAS or EGFR mutations (287 of 520 patients; 95% CI, 51%-60%).

Table 1. Baseline Characteristics of Patients With Advanced Lung Adenocarcinomas
 No. of Patients (%)
CharacteristicTotal, n = 1036KRAS-Positive, n = 241EGFR-Positive, n = 275KRAS/EGFR Wild-Type, n = 520
  1. Abbreviations: EGFR, epidermal growth factor receptor; KPS, Karnofsky performance status; KRAS, Kirsten rat sarcoma 2 viral oncogene homolog.

Women610 (59)144 (60)179 (65)287 (55)
Age ≥70 y333 (32)83 (34)80 (29)170 (33)
Age: Median [range]65 [25-92]66 [33-87]64 [26-89]65 [25-92]
KPS: Median80%80%80%80%
Smoking history    
 Never-smoker344 (33)16 (7)166 (60)162 (31)
 Current smoker197 (19)73 (30)14 (5)110 (21)
 Former smoker495 (48)152 (63)95 (35)248 (48)

Smoking history varied by mutation status. The majority (60%) of patients with EGFR mutations were never-smokers (166 of 275 patients; 95% CI, 54%-66%), whereas only 7% of patients with KRAS mutations (16 of 241 patients; 95% CI, 4%-11%), and 31% of patients without KRAS or EGFR mutations (162 of 520 patients; 95% CI, 27%-35%) had never smoked cigarettes. Conversely, 93% of patients with KRAS mutations were current or former smokers (225 of 241 patients; 95% CI, 89%-96%) compared with 69% of patients without KRAS or EGFR mutations (358 of 520 patients; 95% CI, 65%-73%). Forty percent of patients with EGFR mutations (109 of 275 patients; 95% CI, 34%-46%) had smoked cigarettes.

Treatment History

We reviewed all chemotherapy and targeted agents that patients had received and, in particular, determined whether patients had been received EGFR-TKIs (erlotinib or gefitinib), platinum-based doublet chemotherapy, or antiangiogenesis antibody bevacizumab (Table 2). Patients whose treatment histories were not available in the medical record were excluded in this analysis. Patients who had tumors with EGFR mutations were significantly more likely to have received EGFR-TKIs compared with patients without EGFR mutations (97% vs 30%; P < .001). However, patients with KRAS mutations were as likely as patients without KRAS or EGFR mutations to have received platinum-based chemotherapies (70% vs 65%, respectively; P = .13). Patients with KRAS mutations and patients wild-type KRAS and EGFR received bevacizumab with similar frequency (41% vs 37%, respectively; P = .26).

Table 2. Summary of Treatment History by Genotype
 No. of Patients (%)P-value
TreatmentKRAS-Positive, n = 241EGFR-Positive, n = 275KRAS/EGFR Wild-Type n = 520EGFR-Positive vs Wild-Type EGFRKRAS-Positive vs Wild-Type KRAS
  1. Abbreviations: EGFR, epidermal growth factor receptor; KRAS, Kirsten rat sarcoma 2 viral oncogene homolog; TKI, tyrosine kinase inhibitor.

Platinum     
 Yes152 (63)132 (48)327 (63) .13
 Unknown25 (10)29 (11)57 (11)  
Bevacizumab     
 Yes88 (37)65 (23)190 (37) .26
 Unknown27 (11)39 (14)64 (12)  
EGFR-TKI     
 Yes44 (18)233 (85)148 (28)< .001 
 Unknown28 (12)34 (12)84 (16)  

Survival Analysis of Patients According to Mutation Status

The median follow-up among the 326 patients who were still living with advanced lung adenocarcinoma was 29 months (range, 0.2-108 months). We analyzed OS according to genotype (Fig. 1). The median OS for all patients was 25 months (95% CI, 23-27 months). For patients who had tumors with KRAS mutations, the median OS was 16 months (95% CI, 13-19 months) compared with patients who had tumors with EGFR mutations (median OS, 34 months; 95% CI, 32-39 months) and all patients who had tumors with wild-type KRAS and EGFR (median OS, 23 months; 95% CI, 20-26 months).

thumbnail image

Figure 1. Overall survival (OS) is illustrated according to genotype in patients with advanced lung adenocarcinoma. (A) This Kaplan-Meier plot illustrates the survival of all patients with advanced lung adenocarcinoma. CI indicates confidence interval. (B) These Kaplan-Meier plots illustrate the survival of patients with EGFR-positive (+) disease and KRAS-positive disease. (C) These Kaplan-Meier plots illustrate the survival of patients with KRAS-positive disease and KRAS/EGFR wild-type (wt) disease.

Download figure to PowerPoint

We evaluated each clinical variable to determine its impact on survival outcomes (Table 3). Sex, KPS, and smoking history all were associated with OS in univariate analysis. Women lived longer than men (hazard ratio [HR], 0.8; P < .001). Patients with better performance status at diagnosis (KPS, 80% or 90%) lived longer than patients with poor baseline KPS (HR, 0.5; P < .001). Patients who never smoked cigarettes lived longer than patients who smoked (current or former smoker: HR, 0.9; P = .042). There also was a trend toward improved survival for younger patients (aged ≤70 years) compared with patients who were aged ≥70 years at diagnosis.

Table 3. Clinical Variables and EGFR/KRAS Mutations Associated With Overall Survival: Univariate Analysis
VariableNo. of Patients (%)Median Survival (Range), Reported in MonthsHR [95% CI]P
  1. Abbreviations: CI, confidence interval; EGFR, epidermal growth factor receptor; HR, hazard ratio; KPS, Karnofsky performance status; KRAS, Kirsten rat sarcoma 2 viral oncogene homolog.

Women vs men610 (59) vs 426 (41)28.4 (25.3-31.4) vs 19.3 (17.4-23.3)0.78 [0.67-0.9] vs 1.00< .001
Age <70 y vs ≥70 y703 (68) vs 333 (32)26.4 (23.6-28.9) vs 22.9 (19.3-26.6)0.87 [0.74-1.0] vs 1.00.068
KPS 80%-90% vs ≤70%842 (81) vs 194 (19)27.5 (25.3-31.0) vs 13.2 (10.1-17.2)0.53 [0.44-0.63] vs 1.00< .001
Never-smoker vs current/former344 (33) vs 692 (67)30.4 (25.3-32.6) vs 22.1 (19.0-25.5)0.85 [0.73-0.99] vs 1.00.042
EGFR-positive vs EGFR wild-type 275 (27) vs 761 (73)33.7 (31.6-39.0) vs 20.7 (18.5-23.0)0.63 [0.53- 0.74] vs 1.00< .001
KRAS-positive vs EGFR-positive241 (23) vs 275 (27)16.3 (13.0-19.2) vs 33.7 (31.6-39.0)1.73 [1.41-2.14] vs 1.00< .001
KRAS-positive vs EGFR/KRAS wild-type241(23) vs 520 (50)16.3 (13.0-19.2) vs 22.8 (20.4-26.4)1.13 [0.94-1.35] vs 1.00.17

Mutation status varied with OS in univariate analysis (Table 3). Patients who had tumors with EGFR mutations tumors lived longer than patients without EGFR mutations (HR, 0.6; P < .001). Patients who had tumors with KRAS mutations had significantly worse survival compared with patients who had EGFR mutations (HR, 1.7; P < .001) and a trend toward shorter survival compared with patients who had tumors with wild-type KRAS and EGFR (HR, 1.13; P = .17).

In multivariate analysis, being a women (HR = 0.8, P < .001) and having a higher performance status (HR, 0.5; P < .001) retained an association with favorable survival outcomes. Likewise, the presence of an EGFR mutation retained a strong association with favorable survival compared with the presence of wild-type EGFR (HR, 0.6; P < .001). After adjusting for age, sex, performance status, and the presence of KRAS and EGFR mutations, the presence of a KRAS mutation became more strongly associated with shorter survival compared with the presence of wild-type KRAS/EGFR (HR, 1.21; P = .048) (Table 4).

Table 4. Clinical Variables and EGFR/KRAS Mutations Associated With Overall Survival: Multivariate Analysis
VariableHR (95% CI)P
  • Abbreviations: CI, confidence interval; EGFR, epidermal growth factor receptor; HR, hazard ratio; KPS, Karnofsky performance status; KRAS, Kirsten rat sarcoma 2 viral oncogene homolog.

  • a

    The effect of age is constant until age 70 years and is linear thereafter. Therefore, an HR of 1.15 corresponds to a 15% increase in the risk of death per 5 years of age for patients aged ≥70 years.

Women vs men0.78 (0.67-0.90) vs 1.00< .001
Age <70 y vs ≥70 ya1.15 (1.6-1.25) vs 1.00.001
KPS 80%-90% vs ≤70%0.54 (0.45-0.64) vs 1.00< .001
Current/former smoker vs never-smoker1.00 (0.85-1.18) vs 1.00> .99
EGFR-positive vs wild-type EGFR0.62 (0.52-0.74) vs 1.00< .001
KRAS-positive vs EGFR-positive1.85 (1.47-2.32) vs 1.00< .001
KRAS-positive vs wild-type KRAS/EGFR1.21 (1.01-1.46) vs 1.00.048

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

With the identification of EGFR mutations and anaplastic lymphoma kinase (ALK) rearrangements, and with a renewed interest in KRAS mutations, our understanding of the heterogeneity of lung adenocarcinomas has changed dramatically since 2004. Here, we present outcomes of patients with advanced lung adenocarcinomas and known KRAS and EGFR mutation status. In our analyses, we confirm the known prognostic effects of sex, age, performance status, smoking history, and EGFR mutations.14-16 In this, the largest collection of patients to date with advanced lung adenocarcinomas and KRAS mutations from 1 institution, we also demonstrate that the presence of a KRAS mutation is an independent factor associated with shorter survival (HR, 1.21; P = .048). The negative prognostic effect of KRAS mutations persists in our analyses despite the finding that patients who had tumors with KRAS mutations were equally likely as patients who had tumors with wild-type KRAS and EGFR to have received platinum-based chemotherapy and bevacizumab. We conclude that patients with KRAS mutations have shorter survival because of inherent differences in the tumor biology of KRAS-driven lung cancer rather than differences in the receipt of life-lengthening therapies (platinum-based chemotherapy17 and bevacizumab18).

Although the presence of KRAS mutations in nonsmall cell lung cancers was well established almost 2 decades before the discovery of EGFR,19 there is no analogous targeted therapy available for patients with KRAS mutations. Likewise, conclusions about the prognostic or predictive value of KRAS mutations remain uncertain. Slebos et al first reported the prognostic importance of KRAS mutations in lung cancer in 1990, when they studied tumors from 69 patients who had undergone complete lung tumor resection. After adjustments for sex, smoking, and tumor stage, the presence of a KRAS mutation was the single most important factor identifying patients with shorter disease-free survival (P = .038), OS (P = .002), and death from cancer (P < .001).5 Mitsudomi et al subsequently reported that KRAS mutations also were associated with shortened survival in 45 patients with advanced lung cancer (P = .0103).6

Further investigations into the prognostic significance of KRAS mutations in nonsmall cell lung cancer have been limited to additional small studies. Mascaux et al compiled 28 studies (3620 patients) to determine the prognostic value of KRAS in a meta-analysis.10 Among studies that evaluated only lung adenocarcinomas, the presence of a KRAS mutation was associated with worse survival (HR, 1.59; 95% CI, 1.26-2.02). This univariate analysis of aggregate data combined patients with all stages, combinations of curative and palliative treatment modalities, and a variety of techniques used to identify KRAS mutations in tumor tissue, all of which limited the strength and generalizability of its conclusions. Because other prognostic variables were not available for many of the studies that were included in the meta-analysis, the authors were unable to perform a multivariate analysis to exclude the effects of such important variables as sex, stage, smoking history, and performance status.

In contrast to conventional wisdom and prior studies,16 our multivariate analysis did not identify smoking status as an independent variable associated with prognosis. Our results are explained best by a recent publication in this journal that explored the correlation between patient smoking history, driver mutations (EGFR, KRAS, and ALK), and prognosis.20 Paik et al observed distinct frequencies of EGFR, KRAS, and ALK mutations among never-smokers and current/former smokers with advanced lung adenocarcinoma. Although the prognosis differed between never-smokers and current/smokers and between patients separated by genotype in both groups, there were no differences in survival between never-smokers and current/former smokers who harbored a given mutation in multivariate analysis. Paik et al concluded that distinct mutation profiles drive prognostic differences observed among never-smokers and current/former smokers. Therefore, we would expect the impact of smoking on survival to diminish in significance in a multivariate analysis that also accounts for EGFR and KRAS mutations, such as in our current study.

One limitation of our study was that the patients who underwent mutation testing during this time frame represented a select population with distinct clinical features—they were younger, had a better performance status, and were more likely to be women than the average individual with lung cancer. The frequency of patients with EGFR mutations also was greater than typically reported in United States populations, as would be expected with the described clinical characteristics. Because of this enrichment of favorable clinical characteristics, the survival outcomes for the total population (median OS, 25 months) were longer than those generally reported in unselected patients with advanced lung adenocarcinomas. Although our data describe a “healthier” cross-section of the total lung cancer population, we demonstrate here that patients whose tumors harbor KRAS mutations have comparatively worse survival, suggesting that the disparity between mutation subgroups may be even more pronounced in a broader population of patients with lung adenocarcinoma.

Furthermore, we believe enrichment for favorable clinical characteristics among our patients also explains why the association between KRAS mutations and survival was not significant in univariate analysis but, in multivariate analysis, had a stronger association with shorter survival. That is, after adjusting for the favorable effects of younger age, better performance status, and being a woman, the true association between the presence of KRAS mutations and shorter survival became apparent.

Our study also was limited by the absence of ALK rearrangements from the molecular analysis. All patients who were included in this analysis were diagnosed and had molecular analysis before 2009, when ALK fluorescent in situ hybridization analysis became a common practice for all patients without EGFR and KRAS mutations and well before the 2011 US Food and Drug Administration approval of crizotinib and the companion diagnostic test for ALK rearrangements. Thus, we estimate that 2% to 10% of patients who had tumors with wild-type KRAS/EGFR identified in our analysis would have had ALK rearrangements.21, 22 Although the identification of these patients may have affected the survival analysis for patients who had tumors with wild-type KRAS/EGFR, it would not have had an impact on our findings in patients who had tumors with KRAS mutations. Compared with patients who do not have ALK rearrangements, patients who do have ALK rearrangements reportedly have similar response rates to platinum-based chemotherapy and no difference in overall survival.21 Other known molecular subsets of lung adenocarcinoma, such as tumors with mutations in v-raf murine sarcoma viral oncogene homolog B1 (BRAF),23 human epidermal growth factor receptor 2 (HER2), and neuroblastoma v-ras oncogene homolog (NRAS), represented even smaller proportions of patients than those with the EML4-ALK fusion transcript and were even less likely to be confounding factors in this analyses.

In conclusion, we report here that the presence of a KRAS mutation is a poor prognostic factor for patients with lung adenocarcinomas. Because patients with KRAS mutations have a distinct clinical course that results in shorter survival, they should be evaluated separately in clinical trials. We recommend including KRAS testing in upfront mutation analyses along with testing for EGFR mutations and EML4-ALK to prospectively identify these patients in the clinic.

FUNDING SOURCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

No specific funding was disclosed.

CONFLICT OF INTEREST DISCLOSURES

Dr. Johnson is a consultant to Genentech, receives research funding from Novartis, and her spouse employed as Governmental Affairs lobbyist for Astellas/OSI. Dr. Pao is a consultant to Molecular MD, AstraZeneca, Bristol-Myers Squibb, Symphony Evolution, and Clovis Oncology; receives research funding from Enzon, Xcovery, AstraZeneca, and Symphogen; and holds the rights to EGFR T790M testing, which he has licensed to Molecular MD. Dr. Kris is a consultant to Pfizer, Inc., Boehringer Ingelheim, and Genentech. Dr. Riely is a consultant to Chugai, Tragara, Ariad, Daiichi, and Novartis an receives research funding from Pfizer, GSK, Chugai, and Novartis.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES