Predictive and prognostic markers for epidermal growth factor receptor inhibitor therapy in non-small cell lung cancer

Abstract

Epidermal growth factor receptor (EGFR) related therapies – mainly tyrosine kinase inhibitors (TKIs) such as erlotinib and gefitinib, but also monoclonal antibodies targeting EGFR, for example, cetuximab – have been investigated in numerous settings in non-small cell lung cancer (NSCLC) and in different combinations. The overall clinical benefit of EGFR TKI therapy is roughly 10–30%, with higher benefit in nonsmoker Asiatic women with EGFR-mutated adenocarcinoma. Currently, there are several biomarkers that are able to direct and predict the yield of EGFR-related therapies in NSCLC. These include EGFR mutation status, EGFR protein expression, EGFR gene copy number and a serum proteomic marker (Veristrat®, Biodesix; CO). The usage of such biomarkers is important from many aspects. First, it helps clinicians to make the right treatment decisions and second, it leads to a wiser usage of financial resources. This review will focus on EGFR-related biomarkers for their prognostic power and their ability to predict clinical benefit from EGFR-related therapy.

Introduction

The epidermal growth factor receptor (EGFR) is a 170 kDa transmembrane protein. Its phosphorylation results in downstream pathway activation, which leads to cellular activities such as increased cell proliferation, angiogenesis and apoptosis inhibition. EGFR is widely expressed in several malignancies – lung, breast, colon, esophageal and others – and is among the most extensively studied pathways. The developed approaches to block this pathway are by small molecules that interfere with the tyrosine kinase intracellular domain such as erlotinib (Tarceva®) and gefitinib (Iressa®) or monoclonal antibodies directed at the extracellular receptor domain, for example, cetuximab (Erbitux®). Gefitinib was the first EGFR tyrosine kinase inhibitor (TKI) approved by the United States FDA in 2003 for the treatment of non-small cell lung cancer (NSCLC), for second- or third-line therapy, based on the data from the IDEAL 1 and IDEAL 2 studies [Fukuoka et al. 2003; Kris et al. 2003]. Both studies involved around 200 patients who had failed prior therapies and were randomized to receive either gefitinib at 250 mg/day or 500 mg/day. Both doses produced long-lasting tumor responses and improved symptoms. In 2004, a confirmatory study – Iressa Survival Evaluation in Lung Cancer (ISEL) – a phase III survival trial that compared gefitinib with placebo, failed to demonstrate improvement in survival [Thatcher et al. 2005] and gefitinib was subsequently removed from the market in the United States.

Erlotinib is currently the only FDA-approved EGFR-TKI for the treatment of NSCLC in the United States. In the National Cancer Institute (NCI) study, BR-21, erlotinib was compared with placebo for the treatment of advanced NSCLC after failure of first- or second-line chemotherapy. There were 731 patients randomized in a 2:1 ratio to either erlotinib 150 mg/day or placebo. Erlotinib improved not only objective response rate (ORR; 8.9 versus 1%, p < 0.001) and progression-free survival benefit (PFS; 2.2 versus 1.8 months, hazard ratio [HR] 0.61, p < 0.001) but also overall survival (OS) 7.9 versus 3.7 months (HR 0.7, confidence interval [CI] 0.58–0.85, p < 0.001) [Shepherd et al. 2005].

Cetuximab is the most clinically developed monoclonal antibody against EGFR. Cetuximab interacts with the EGFR inducing its internalization and blocking the downstream cellular signaling. Recently, a phase III clinical trial, FLEX (First-Line ErbituX in Lung cancer) showed that adding cetuximab to standard chemotherapy can prolong survival in patients with advanced NSCLC [Pirker et al. 2009]. In this study, 1,125 patients with EGFR-expressing tumors were randomized to chemotherapy (cisplatin/vinorelbine) with or without cetuximab. Patients in the cetuximab arm had a superior survival (primary endpoint) of 11.3 versus 10.1 months (p < 0.044) over the group that received only chemotherapy. ORR was also improved in the study arm (36 versus 29%). However, the PFS was similar in the two arms. In another study, BMS-099, another chemotherapy regimen (carboplatin/paclitaxel) with and without cetuximab in unselected patients were compared and no difference was observed in PFS (primary endpoint) [Lynch et al. 2008]. The results of the FLEX study support the use of cetuximab in combination with chemotherapy in the first-line treatment of patients with expression of EGFR.

Before describing the measures that may predict tumor response or OS benefit with EGFR-TKI therapy, it is important to define prediction and prognosis. Prediction is the power of a measure to predict tumor response to a certain therapy while prognosis is the association between a measure and OS, independent of therapy. Those measures are not always similar, as tumor response does not necessarily affect survival, and vice versa. The measures that will be discussed are EGFR protein expression, EGFR mutations, EGFR gene copy number and serum proteomics.

EGFR-related biomarkers for prognosis

There are conflicting data about the prognostic importance of EGFR protein levels in NSCLC. A meta-analysis of several studies failed to show a consistent correlation between EGFR-expression levels and survival [Meert et al. 2002]. Most studies showed no prognostic effect of EGFR expression or a slight negative effect. Studies of EGFR gene copy numbers assessed by fluorescent in situ hybridization (FISH) have also failed to show a consistent association between increased EGFR gene copy number and prognosis, although two studies have reported a worse survival with EGFR gene amplification [Sasaki et al. 2008; Hirsch et al. 2003]. The results for activating EGFR gene mutations are strikingly different. Nearly all studies reported that patients with these mutations have a superior outcome compared to those without these mutations, irrespective of stage and treatment [Sequist et al. 2007; Jackman et al. 2006; Riely et al. 2006; Mitsudomi et al. 2005].

Prediction of outcome following EGFR-targeted therapy

Several clinical and pathologic factors are known to predict tumor response or OS benefit of EGFR-TKI therapy. A better response to therapy is related to East Asian ethnicity, female sex, never-smoking status, and/or adenocarcinoma histology. In addition, there are several molecular biomarkers, which, retrospectively, have been shown to discriminate the clinical benefit of EGFR-TKIs and EGFR-targeting monoclonal antibodies in the therapy of NSCLC [Hirsch et al. 2008a]. The recent studies, IPASS [Mok et al. 2008] and OSI-774-203 [Hirsch et al. 2008b], suggest that molecular biomarkers are more important than clinical features in selecting NSCLC patients for first-line therapy with EGFR-TKIs. The observation was that in clinically selected patients, survival was superior with EGFR-TKIs (compared to chemotherapy) in patients with EGFR mutations, whereas survival was inferior with EGFR-TKIs in the clinically selected patients without EGFR mutations.

EGFR protein expression

Using an H-score method, Cappuzzo and colleagues [Cappuzzo et al. 2005] found that patients with a high score for immunohistochemical detection of the EGFR (score > 200, IHC-positive patients) had significantly higher response and disease control rates, and significantly longer time-to-progression (TTP) and survival, than patients with lower scores (score < 200) when treated with gefitinib. In the randomized phase III BR.21 trial comparing erlotinib to placebo, IHC-positive patients treated with erlotinib had a significantly superior survival compared with placebo-treated patients (HR = 0.68, p = 0.02) [Tsao et al. 2005]. In the ISEL trial, gefitinib produced a reduction in the HR for survival in IHC-positive patients but this was not significant (HR = 0.77, p = 0.13) when compared with placebo in IHC-positive patients; however, the interaction coefficient (comparison of HRs to IHC positive and negative groups) was statistically significant [Hirsch et al. 2006].

There are fewer studies evaluating the relationship between EGFR protein expression and the efficacy of cetuximab. The FLEX study, that compared chemotherapy alone (cisplatin and vinorelbine) to chemotherapy plus cetuximab, included only patients whose tumor expressed EGFR by IHC (∼85%) [Pirker et al. 2009]. The study showed a statistically significant survival advantage (HR 0.871, p = 0.04), however, there was no association between the degree of EGFR expression and improved outcome. The BMS-099 study compared a different chemotherapy regimen (carboplatin plus taxane), to chemotherapy plus cetuximab [Lynch et al. 2008]. In this study, all patients were included, irrespective of EGFR expression. There was no significant effect of cetuximab on PFS, although patients receiving cetuximab had a slightly superior PFS (HR 0.89, p = not significant). It is not clear if the small differences in PFS between the FLEX and BMS-099 studies could be related to selection by IHC. The inability of EGFR protein expression to serve as a predictive biomarker in NSCLC is similar to what is seen with the use of cetuximab in metastatic colon cancer and thus invites the exploration of other biomarkers for EGFR-targeted antibodies [Cunningham et al. 2004].

EGFR mutations

Somatic mutations in the kinase domain of EGFR in lung carcinoma exist in approximately 10% of specimens in the United States and 30–50% in Asia [Sequist et al. 2007]. About 90% of EGFR-activating mutations are clustered in exon 19 and 21 [Sharma et al. 2007]. Patients with these mutations have a greater response rate to EGFR-TKIs (approximately 60–80%) than unselected patients (approximately 10–20%) [Riely et al. 2006]. Clinically, there seems to be differences in outcome based on the type of EGFR mutations. Patients with del 19 mutations demonstrate a higher response rate and longer survival with EGFR-TKI therapy than point mutations in exon 21 [Hirsch et al. 2007; Jackman et al. 2006; Riely et al. 2006; Mitsudomi et al. 2005]. Activating mutations in exons 18–21 of EGFR were initially identified in NSCLC patients with a clinical response to gefitinib [Lynch et al. 2004; Paez et al. 2004]. Deletions in exon 19, substitution mutations in exon 21 (L858R), or less common mutations (e.g. G719X, L861Q) cluster in the kinase domain around the ATP-binding site. These mutations lead to preferential phosphorylation of AKT and STAT3/5 as opposed to ERK1/2 and cell lines harboring these mutations show increased sensitivity to gefitinib [Lynch et al. 2004; Paez et al. 2004; Sordella et al. 2004]. T790M mutations (exon 20) were later identified in patients with existing activating mutations of EGFR who developed (acquired) resistance to gefitinib or erlotinib [Pao et al. 2005a]. This mutation restores the lowered affinity of ATP seen in the activating mutations to wild-type levels, thus limiting the effectiveness of drugs such as gefitinib or erlotinib [Yun et al. 2008].

Several prospective phase II and III studies have studied EGFR-TKIs as first-line therapy in NSCLC patients with EGFR mutations [Mok et al. 2008; Sequist et al. 2008; Sugio et al. 2009; West et al. 2008; Cappuzzo et al. 2007; Jackman et al. 2006; Okamoto et al. 2006; Paz-Ares et al. 2006]. These studies reported high response rates and longer PFS and OS than would be expected from chemotherapy alone, based upon historical controls. In the recent IPASS trial [Mok et al. 2008; Fukuoka et al. 2009], 1,217 chemotherapy-naïve patients with advanced NSCLC were clinically selected: Asian ethnicity, adenocarcinoma histology and never or light smokers. Six hundred and nine patients were treated by gefitinib (250 mg/day) as first-line therapy and 608 patients with carboplatin (AUC 5 or 6) and paclitaxel (200 mg/m²) weekly. Preliminary analysis showed that, overall, PFS but not survival were different between the two arms (HR = 0.74, p < 0.0001). Of particular interest was that PFS was significantly longer for gefitinib than chemotherapy in patients with EGFR-mutated tumors (HR = 0.48, p < 0.0001), but significantly longer for chemotherapy than gefitinib in patients whose tumors lacked EGFR mutations (HR = 2.85, p < 0.0001, Figure 1). Similar data were seen in a smaller randomized phase II study by Hirsch and colleagues where EGFR-positive patients (IHC or FISH testing) were randomized to either erlotinib or erlotinib plus chemotherapy [Hirsch et al. 2008b]. In the OSI-774-203 protocol, patients expressing EGFR by IHC or FISH were randomized to receive chemotherapy (carboplatin/paclitaxel) intercalated with erlotinib or erlotinib alone in cycles of 21 days. Preliminary results showed a median PFS of 2.7 months for the erlotinib arm versus 4.6 months for the combination arm. The median OS was 16.7 versus 11.9 months favoring the erlotinib arm, although statistical significance was not achieved. However, in patients with EGFR mutations, the median PFS was 18.2 months with erlotinib compared to 4.9 months with the alternation of chemotherapy and erlotinib. In contrast, the chemotherapy arm had a better median PFS in patients without an EGFR mutation [Camidge et al. 2008]. These data strongly suggest that molecular markers are more important than clinical features in selecting patients. Toxicity and quality-of-life analysis favored gefitinib over chemotherapy. Thus, an EGFR-TKI may be preferred over chemotherapy in chemotherapy-naïve patients with EGFR mutations. OS was not significantly different for both groups. The lack of survival benefit in IPASS can be explained by the crossover to EGFR-TKI as second/third-line therapy.

Figure 1.

Progression-free-survival (PFS) for gefitinib versus chemotherapy as first-line therapy in EGFR positive and negative patients. Edited from the IPASS study [Mok et al. 2008].

There is no evidence that EGFR mutations predict for superior outcome following cetuximab therapy [O’Byrne et al. 2009].

EGFR gene copy number

The EGFR gene is located at the p14-p12 region of human chromosome 7 and can be detected by FISH. True gene amplification or high polysomy is defined according to the Colorado EGFR scoring system [Cappuzzo et al. 2005]. We evaluated the predictive value of FISH status in several phase II studies of EGFR-TKIs in the second- and third-line setting. FISH-positive patients had superior response rates and longer survival than FISH-negative patients in these studies [Goss et al. 2009; Hirsch et al. 2007; Cappuzzo et al. 2005; Hirsch et al. 2005].

The predictive value of FISH positivity was then assessed in prospective randomized studies of erlotinib and gefitinib in the second- or third-line setting. In the BR-21 study, the FISH-positive patients who were randomized to erlotinib had a significantly superior overall survival compared with FISH-positive patients randomized to placebo (HR = 0.43, p = 0.01) whereas in the FISH-negative patients, there were no significant differences in survival between those receiving erlotinib and placebo (HR = 0.93) [Zhu et al. 2008]. In the ISEL study, FISH-positive patients receiving gefitinib also had a superior survival compared with FISH-positive patients receiving placebo (HR = 0.61, p = 0.06) [Hirsch et al. 2006. There are fewer data about the predictive value of FISH positivity in the first-line setting. FISH positivity predicted a superior outcome following gefitinib in patients with bronchioloalveolar lung cancer in a Southwest Oncology Group (SWOG) study (Hirsch et al. 2005). Patients with FISH-positive tumors had a superior outcome in the first-line IPASS study but nearly all FISH-positive patients also had EGFR mutations [Fukuoka et al. 2009; Mok et al. 2008]. In the TRIBUTE trial (a first-line phase III trial evaluating carboplatin and paclitaxel with or without erlotinib), sub-analysis by EGFR FISH showed that OS did not differ between FISH-positive and FISH-negative patients in either the chemotherapy plus erlotinib arm or the chemotherapy plus placebo arm [Hirsch et al. 2008b]. In FISH-positive patients, median TTP was 6.3 months in the erlotinib arm versus 5.8 months in the placebo arm (HR, 0.59; 95% CI, 0.35–0.99; p = 0.0430); in FISH-negative patients, median TTP was 4.6 months versus 6.0 months (HR, 1.42; 95% CI, 0.95–2.14; p = 0.0895; treatment interaction test, p = 0.007). After 6 months of treatment, a notable separation of the TTP curves in favor of erlotinib emerged, indicating a role to erlotinib as maintenance therapy, which has later been verified in the SATURN trial. Objective response rates were 11.6 versus 29.8% in FISH-positive patients (chemotherapy plus erlotinib arm versus chemotherapy plus placebo arm; p = 0.0495) and 21.8 versus 25.4%, respectively, for FISH-negative patients (p = 0.6954).

Contradictory findings were reported in randomized studies comparing EGFR-TKIs to chemotherapy in the second- or third-line settings. In the INTEREST trial that compared gefitinib to docetaxel, neither EGFR mutations nor FISH predicted for superior outcome with gefitinib – most probably due to better outcomes associated to the markers in both treatment arms [Kim et al. 2008]. There was a trend for better PFS in both EGFR mutations and FISH-positive patients treated with gefitinib. Thus, while there is consistent data regarding the predictive value of EGFR mutations in the first line, the data in the second line are less certain.

Since EGFR-activating mutations do not appear to predict response to cetuximab [Tsuchihashi et al. 2005] it is important to develop biomarkers in order to select patients who are more likely to receive benefit from this targeted agent. In a recent analysis from SWOG comparing sequential versus concurrent cetuximab in addition to carboplatin and paclitaxel chemotherapy (S0342), FISH-positive patients demonstrated longer median survival (15 versus 7 months), median PFS (6 versus 3 months) and objective response rate (ORR, 45 versus 26%) [Hirsch et al. 2008]. These results suggest that FISH status might be a predictive biomarker for response to cetuximab therapy, although they were not conclusive since patients in both arms of the study received cetuximab. However, similar results were not observed in the recently reported BMS-099 [Khambata-Ford et al. 2009] and the FLEX [O’Byrne et al. 2009] studies where there was no association between FISH status and outcome with cetuximab. The SWOG group has initiated a prospective validation of EGFR FISH in a large prospective phase III trial (0819), which is also investigating chemotherapy with or without cetuximab as first-line therapy. EGFR FISH is a co-primary endpoint. Hopefully, this large study will clarify the role of FISH and other EGFR biomarkers as predictive markers for cetuximab therapy.

Hirsch and colleagues examined six biomarkers (EGFR and HER2 gene copy number, EGFR IHC, pAKT IHC, EGFR mutation, and KRAS mutation) in 204 gefitinib-treated NSCLC patients [Hirsch et al. 2007]. Multivariate analysis showed that survival was affected independently by never smoking (HR = 0.37, p < 0.001), performance status of zero to one (HR = 0.28, p < 0.001), increased EGFR gene copy number (HR = 0.54, p = 0.006), and high EGFR protein expression (HR = 0.59, p = 0.007). Double positivity for EGFR IHC and high EGFR gene copy number had an OR of 41% and a one year survival rate of 77%, while double negativity to both EGFR IHC and FISH did not benefit from gefitinib therapy. The predictive power of EGFR FISH and other biomarkers will be prospectively tested in a phase III trial, the MARVEL study (Marker Validation for Erlotinib in Lung Cancer, N0723).

KRAS mutation

Much discussion has been devoted to the role of KRAS mutation and prognosis in NSCLC. While KRAS mutation has been demonstrated to be a negative predictor in EGFR inhibition in patients with colorectal cancer (CRC), its role in NSCLC patients is still under debate. In NSCLC, KRAS mutation has been demonstrated to be associated with poor prognosis and is thus a negative prognostic factor, which needs to be taken into account when the predictive performance is assessed.

KRAS is an important downstream mediator of EGFR signaling and harbors an activating mutation in codon 12 or 13 (exon 2) in approximately 10–30% of NSCLC cases. EGFR and KRAS-activating mutations are almost always mutually exclusive. Activating mutations in KRAS are associated with lower response rates to EGFR-TKIs and numerically worse survival [Miller et al. 2008; Zhu et al. 2008; Pao et al. 2005]. A recent analysis of 59 NSCLC patients treated by EGFR-TKIs by Massarelli and colleagues showed that KRAS gene mutation is associated with a poor response to TKIs and overcomes the potentially favorable role of increased EGFR gene copy number in NSCLC tumors [Massarelli et al. 2007]. Testing of samples for KRAS mutations in addition to molecular testing of EGFR (mutation, FISH) and serum proteomics will likely improve our ability to predict response to EGFR targeted therapy.

Serum proteomics

Serum proteomics is an emerging science, analysing the serum for specific proteins by mass spectrometry (MS). Numerous studies have provided evidence that this technology can be used to uncover proteomic expression patterns linked to a disease state. Taguchi and colleagues [Taguchi et al. 2007] have tested the serum from 139 NSCLC patients trying to classify a proteomic profile to predict a clinical benefit from EGFR-TKI therapy. Based on their classification, they identified patients who showed improved outcomes after EGFR TKI treatment. In one cohort, median survival of patients in the predicted ‘good’ and ‘poor’ groups was 207 and 92 days, respectively (HR of death in the good versus poor groups = 0.50, 95% CI = 0.24–0.78). In the other cohort, median survivals were 306 versus 107 days (HR = 0.41, 95% CI = 0.17–0.63). The classifier did not predict outcomes in patients who did not receive EGFR TKI treatment. Finally, the proteomic classifier has later been commercialized under the name Veristrat® (Biodesix; CO, USA). This field should be further investigated.

Conclusion

There are several biomarkers that may lead to better EGFR-TKI selection for lung cancer. Tailoring therapy might lead to a more effective response and improved survival. To summarize, the main predictive biomarkers are as follows:

• EGFR mutation predicts a better response to EGFR TKIs;

• a high EGFR gene copy number predicts a better response to EGFR TKIs;

• KRAS mutation predicts a low response to EGFR TKIs.

Further studies should be conducted in order to be able to use these biomarkers in the routine clinical algorithm.

Conflict of interest statement

F.H: Consultant/Advisory Boards: AstraZeneca, Roche, Lilly, Pfizer, Boehringer-Ingelheim, Merck Serono, Ventana-Roche, GlaxoSmithKline, BMS/Imclone; Research Funding: OSI, Genentech, AstraZeneca, Merck (USA), Syndax, Ventana-Roche; Patent: EGFR FISH as a predictive marker for EGFR Inhibitors. K.Y; M.W; N.P: nothing to disclose. Supported by SPORE (FH) and IASLC (NP).