Purpose: To explore in a panel of patient-derived xenograft models of human non–small cell lung cancer (NSCLC) whether high EGFR expression, was associated with cetuximab activity.

Experimental Design: NSCLC patient-derived xenograft models (n = 45) were implanted subcutaneously into panels of nude mice and randomization cohorts were treated with either cetuximab, cisplatin, cisplatin plus cetuximab, vehicle control, or else were left untreated. Responses according to treatment were assessed at week 3 by analyzing the relative change in tumor volume and an experimental analogue of the Response Evaluation Criteria in Solid Tumors (RECIST) guidelines. An EGFR IHC score was calculated for each patient-derived xenograft model and response was assessed according to EGFR expression level.

Results: When tumors were stratified into high and low EGFR expression groups (IHC score threshold 200; scale 0–300), a stronger antitumor activity was seen in the high EGFR expression group compared with the low EGFR expression group in both the cetuximab monotherapy and cisplatin plus cetuximab combination therapy settings. For tumors treated with cisplatin plus cetuximab, the objective response rate was significantly higher in the high EGFR expression group compared with the low EGFR expression group (68% vs. 29%). Objective response rates were similar in high and low expression groups for tumors treated with cisplatin alone (27% vs. 24%, respectively).

Conclusion: Cetuximab activity in NSCLC patient-derived xenograft models was demonstrated clearly only in tumors that expressed high levels of EGFR, as defined by an IHC score of ≥200. Clin Cancer Res; 20(17); 4478–87. ©2014 AACR.

Translational Relevance

The phase III FLEX study in patients with non–small cell lung cancer (NSCLC) showed that the survival benefit associated with the addition of cetuximab to chemotherapy was limited to patients whose tumors expressed high levels of EGFR, as defined by an IHC score of ≥200 (scale of 0–300). This study investigated in an extensive panel of patient-derived xenograft models of human NSCLC whether cetuximab activity was similarly associated with high tumor EGFR expression when cetuximab was administered alone or in combination with cisplatin. Objective responses to cetuximab monotherapy were seen only in tumors with EGFR expression levels determined to be ≥200. Furthermore, the addition of cetuximab to cisplatin was associated with a significant improvement in response rate in tumors with high EGFR expression but had no apparent effect in tumors with low EGFR expression. This provides further evidence that in the treatment of advanced NSCLC, high tumor EGFR expression can predict cetuximab benefit.

The efficacy of a particular anticancer drug may be associated with the presence or absence of specific genetic driver lesions within the cells of a tumor. Such lesions may be point mutations, chromosomal translocations, or increases in gene copy number (1–6). Perhaps related to such changes in the genome, the level of expression of the target protein of an anticancer drug within tumor cells can also represent a predictive biomarker on which treatment decisions may be based (7, 8). The discovery and clinical application of predictive tumor biomarkers are key enabling factors in the emerging field of personalized medicine, which allow the selection of subgroups of patients most likely to benefit from treatment with a particular anticancer drug.

The addition of the EGFR antibody, cetuximab, to platinum-based chemotherapy was shown to improve response rate and overall survival compared with chemotherapy alone in the first-line treatment of patients with advanced non–small cell lung cancer (NSCLC; ref. 9). However, in the absence of a predictive biomarker, the effect was modest in an unselected patient population. The randomized phase III trial FLEX comparing the combination of cetuximab plus cisplatin and vinorelbin versus cisplatin and vinorelbin was the largest study in this setting (10). An analysis of prospectively collected tumor EGFR IHC data from FLEX study patients explored whether EGFR expression level was predictive for cetuximab activity (11). This led to the identification of a discriminatory threshold IHC score of 200 on a scale of 0–300, which allowed the separation of a patient subgroup benefiting from the addition of cetuximab to chemotherapy (high EGFR expression; IHC score ≥200) from one deriving little or no benefit (low EGFR expression; IHC score <200). For patients in the high EGFR expression group (n = 345, 31%), overall survival was longer in the chemotherapy plus cetuximab arm compared with the chemotherapy alone arm (median 12.0 months vs. 9.6 months; HR = 0.73; P = 0.011) and the response rate was higher (44.4% vs. 28.1%; OR = 2.04; P = 0.002). No corresponding benefit for the addition of cetuximab to chemotherapy was seen for patients in the low EGFR expression group. Treatment interaction tests assessing the difference in the HRs for overall survival and ORs for response suggested a predictive value for EGFR expression level based on the threshold IHC score of 200.

Recently, a similar finding was reported from the RTOG 0617 trial exploring the addition of cetuximab to radiotherapy and standard chemotherapy for the treatment of patients with unresectable stage III NSCLC. Although adding cetuximab to chemoradiotherapy did not improve overall survival in an unselected patient population, patients with tumors showing high EGFR expression level according to a hybrid IHC score were significantly more likely to benefit from the addition of cetuximab than patients with low EGFR expression (12).

It has been suggested that patient-derived xenograft model systems may more faithfully represent the characteristics of primary human tumors than xenografts of established human tumor cell lines (13–16). The aim of the current study was to investigate whether high tumor EGFR expression levels, as defined in the FLEX clinical study by an IHC score of ≥200, were associated with response to therapy in an extensive panel of patient-derived NSCLC xenografts grown in athymic nude mice when cetuximab was administered alone or in combination with cisplatin. The intention was therefore to further explore in an experimental setting the biologic rationale that high EGFR expression may be a predictive biomarker for cetuximab treatment benefit in NSCLC.

Patient-derived NSCLC xenografts and mouse strains

All experiments were carried out on behalf of Merck KGaA by one of the three contract research organizations: Oncotest, Experimental Pharmacology & Oncology, and Xentech.

The patient-derived xenograft models were established from surgically resected tumor specimens from patients with NSCLC (see Supplementary Table S1) by subsequent serial passaging in immune-deficient mice. The passage numbers varied from passage 5 to 33 (see Supplementary Table S1). The following host mice strains were used: Female NMRI nu/nu (Experimental Pharmacology & Oncology and Oncotest) and female athymic nu/nu (Xentech). The xenografts from Experimental Pharmacology & Oncology were described previously (17). All experiments were conducted in accordance with relevant local regulations and guidelines relating to animal welfare.

The only inclusion criterion of a particular patient-derived NSCLC model in the study was the availability of the model during a certain time period at the respective contract research organization. Consequently, the xenografts used in this analysis represent a broad diversity of human NSCLC characteristics and include models derived from primary and metastatic sites and those representing a variety of different histologic subtypes (see Supplementary Table S1).

Experimental design

Each NSCLC xenograft assessed (n = 45) was surgically excised from its murine host and cut into 50 fragments of approximately 2 to 5 mm diameter. These fragments were then implanted subcutaneously using tweezers into the flanks of 50 anaesthetized nude mice (or fewer depending on availability) via small incisions (one tumor fragment per animal). Animals were subsequently monitored daily and when transplanted tumors became palpable or reached a volume of approximately 80 to 200 mm3, they were randomized to five experimental groups of up to 10 mice. Three of the five groups received anticancer drug treatment comprising either cetuximab 30 mg/kg i.p. injection twice weekly, cisplatin 5 mg/kg i.p. injection once weekly, or cisplatin combined with cetuximab, at the monotherapy dose levels. On days on which both cisplatin and cetuximab were administered, cetuximab was given one hour before cisplatin. Mice in the control group received intraperitoneal injections of the cisplatin/cetuximab vehicle (0.9% NaCl) according to the schedule for cetuximab. Response was assessed after 3 weeks of treatment. Animals randomized to the fifth group (biomarker arm) received no treatment and tumors were harvested at a size of 300 to 400 mm3.

Body weights were recorded twice a week. Tumor volume was determined using two-dimensional measurements taken with a caliper on the same day that mice were weighed. Mice were sacrificed if the volume of their tumor reached 2,000 mm3.

Data availability and patient-derived tumor models

Forty-five NSCLC patient-derived xenograft models were included in the data analysis. EGFR IHC data were obtained from 41 tumors. A complete dataset (efficacy data and IHC data) was not available for all tumors, leading to a reduced number evaluable for certain analyses. Tumors for which EGFR expression data were not available were only considered in analyses relating exclusively to the differential effects of therapy. In addition, as certain tumors grew relatively quickly, mice in some instances had to be sacrificed before week 3. Thus, the number of patient-derived xenograft models contributing to the presented data varies between 37 and 45 depending on the analysis.

EGFR expression analysis

Where possible, for each human NSCLC patient-derived xenograft model, EGFR expression was assessed by IHC in tumors from 5 of 10 randomly selected untreated mice from the biomarker arm. Tumors were sampled when they reached a volume of 300 to 400 mm3, formalin fixed, and embedded in paraffin. They were subsequently analyzed at a central pathology laboratory using the EGFR pharmDx Kit (Dako) according to a standard protocol, without knowledge of treatment outcome. A median EGFR IHC score on a scale of 0–300 was calculated from all tumors analyzed per patient-derived xenograft model based on the intensity and frequency of membrane staining of tumor cells, as previously described (11). This score was used to classify the NSCLC patient-derived xenograft models into high (IHC score ≥200) or low (IHC score <200) EGFR expression groups.

Analysis of EGFR kinase domain and KRAS mutations

The EGFR kinase domain mutation status of the patient-derived xenografts was assessed in two, randomly selected tumors from the biomarker arm. DNA was extracted with the Qiagen QIAamp DNA FFPE Tissue Kit according to the manufacturer's instructions and quantified using a Nano Drop spectrophotometer. Detection of EGFR mutations was carried out using the CE-IVD marked Qiagen Therascreen EGFR PCR Kit (QIAGEN) according to the manufacturer's protocol.

KRAS mutation status of tumors was assessed in frame of a CAGE (Cancer Genome) scanner analysis performed by Blackfield. Per patient-derived xenograft, one randomly selected tumor from the biomarker arm was analyzed. DNA was extracted from 10 sections (each 10 μm) derived from fresh-frozen tumor material using the Qiagen Gentra Puregene Kit according to the manufacturer's protocol. DNA concentrations were measured using the Quant-iT PicoGreen dsDNA Assay Kit (Life Technologies). The principle of CAGE is based on the hybrid capture technology (18). Baits, targeting the genomic regions of interest, were designed based on an up-to-date list of cancer relevant genes, observed to be mutated, fused, amplified, or deleted in cancer using Agilent SureDesign. Oligonucleotide composition was adapted on the basis of locus-specific features of the region of interest. Input genomic DNA was sheared and subjected to library preparation according to the manufacturer's protocol. The baits and genomic DNA libraries were mixed to enrich for the genomic regions of interest followed by in-depth sequencing. Sequencing was performed by ATLAS Biolabs GmbH.

Statistical analysis and considerations

Study endpoints included the ratio of relative tumor volumes (compared with treatment start) in cetuximab-treated compared with control tumors and cisplatin plus cetuximab-treated compared with cisplatin-treated tumors (treatment/control, T/C, ratio). Objective response at week 3 was also assessed according to an experimental analogue of the Response Evaluation Criteria in Solid Tumors (RECIST) score (19).

T/C ratios, defined as the ratio of the relative median tumor volume for both treatment arms at time t and start of treatment, were calculated at week 3 according to the formula:

formula

Where data (animals A1–A10) were available for less than 10 animals per group, endpoint calculations were based on as many tumors as were evaluable. T/C ratios <1 (represented graphically as log2 values <0) indicated tumor growth inhibition in the treatment arm compared with the control arm. A NSCLC xenograft was dropped from the analysis if endpoint data were missing entirely for one experimental arm.

Objective response was determined according to a RECIST score calculated from the percentage change in median relative tumor volume at week 3 and the median tumor volume at the start of treatment (defined as 100%) according to the formula:

formula

Objective response was defined as complete (RECIST score −100%) or partial remission (≤−30% to >−100%). Stable disease was defined by a RECIST score of ≤20% to >−30%, and progressive disease by scores of >20%.

A number of different statistical tests and regression models were used in the assessment of treatment effects and associations, including Wilcoxon signed rank test, Pearson correlation, Mann–Whitney test, Fisher exact test, and mixed-effects logistic regression model. Tests were applied according to which endpoints, which EGFR expression categorization, and which treatment arms were to be evaluated. Significance was assigned to test the results that passed a P value threshold of <0.05. Statistical tests were carried out as implemented in the statistics software R (http://cran.r-project.org).

A key objective of the study was to assess the effect of treatment with cetuximab or cisplatin plus cetuximab according to whether tumors expressed high (IHC score ≥200) or low (IHC score <200) levels of EGFR. Outcome was therefore assessed both in high versus low EGFR expression groups as well as according to EGFR IHC score considered as a continuous variable. Where the EGFR IHC score was not available for a xenograft, tumors were excluded from all expression level-related analyses.

Response rates according to EGFR IHC score for cetuximab monotherapy compared with control and cisplatin plus cetuximab compared with cisplatin alone were assessed using the subpopulation treatment effect pattern plot (STEPP) method (20). In a sliding window approach, (tumors ordered by EGFR IHC score; window size of a 25% quartile range of scores) objective response rates and median IHC scores were determined for each window and plotted accordingly.

Patient-derived xenograft bearing animals were randomized to the four different experimental groups: cetuximab, cisplatin, cisplatin plus cetuximab, and vehicle control. In addition, animals were allocated to a biomarker group that did not receive any treatment and that was used for molecular analysis. Treatments started when tumors were well established, at volumes of approximately 80 to 200 mm3.

Cetuximab monotherapy

For the determination of efficacy using T/C ratios for cetuximab monotherapy versus vehicle control, 42 tumors were evaluable (Fig. 1A). Cetuximab administration resulted in a significant reduction of tumor growth in this broad panel of patient-derived xenograft models, demonstrated by T/C ratios <1 for most models (P = 2.95 × 10−7, nonparametric one-sample Wilcoxon signed rank test). When applying the RECIST model for the evaluation of response to cetuximab, an objective response was observed in seven of 42 patient-derived xenografts, corresponding to a response rate of 17%.

Figure 1.

A, T/C ratios (independent of EGFR expression level) across the NSCLC xenograft panel for cetuximab monotherapy/control. Tumors are ordered by log2 of the T/C ratio. Value of 0 indicates no effect of cetuximab administration on tumor volume. B, T/C ratios for cetuximab monotherapy/control according to EGFR expression level on a continuous scale. C, T/C ratios for cetuximab monotherapy/control according to EGFR expression level in low and high EGFR expression groups (top/bottom boundaries of each box plot represent the 25th and 75th percentile; horizontal line within the box denotes the median value).

Figure 1.

A, T/C ratios (independent of EGFR expression level) across the NSCLC xenograft panel for cetuximab monotherapy/control. Tumors are ordered by log2 of the T/C ratio. Value of 0 indicates no effect of cetuximab administration on tumor volume. B, T/C ratios for cetuximab monotherapy/control according to EGFR expression level on a continuous scale. C, T/C ratios for cetuximab monotherapy/control according to EGFR expression level in low and high EGFR expression groups (top/bottom boundaries of each box plot represent the 25th and 75th percentile; horizontal line within the box denotes the median value).

Close modal

The antitumor activity of cetuximab (indicated by low T/C ratios) was particularly pronounced in tumors with high EGFR expression levels (Fig. 1B), which was reflected by a significant correlation of log2 T/C ratio at week 3 and EGFR IHC score (Pearson correlation coefficient r = −0.39; P = 0.016). When an EGFR IHC score threshold of 200 (the same as identified in the FLEX study) was used to stratify the NSCLC xenografts into high (IHC score ≥200; n = 21) and low (IHC score <200; n = 16) expression groups, strong responses to cetuximab versus control were exclusively observed in the group with high EGFR expression. This was illustrated by significantly different log2 T/C ratios at week 3 between the two groups (P = 0.0013, Mann–Whitney test; Fig. 1C).

In line with these results, objective responses to cetuximab treatment measured according to the RECIST analogous endpoint were seen only in tumors with EGFR IHC scores ≥200 (n = 7; P = 0.0437, Mann–Whitney test; Fig. 2A and B). This corresponded to a response rate of 32% in the high EGFR expression group versus 0% in the low EGFR expression group (7/22, 32% vs. 0/16, 0%; P = 0.0144, Fisher exact test). Not all tumors with high EGFR expression showed sensitivity to cetuximab treatment, indicating that additional factors determine the response to anti-EGFR therapy.

Figure 2.

EGFR IHC score and objective response in the cetuximab monotherapy arm. A, tumor EGFR IHC score range according to response category (for derivation of box plot parameters, see Fig. 1). B, response rate in low and high EGFR expression groups.

Figure 2.

EGFR IHC score and objective response in the cetuximab monotherapy arm. A, tumor EGFR IHC score range according to response category (for derivation of box plot parameters, see Fig. 1). B, response rate in low and high EGFR expression groups.

Close modal

Cisplatin plus cetuximab versus cisplatin alone

The phase III FLEX study compared cisplatin-based chemotherapy plus cetuximab with chemotherapy alone. Therefore, in the current study, we assessed the effect of adding cetuximab to cisplatin with T/C ratios calculated using the cisplatin group as the comparator. T/C ratios for cisplatin plus cetuximab versus cisplatin treatment could be calculated for 45 xenograft models (Fig. 3A). The addition of cetuximab to cisplatin led to an overall improvement in the inhibition of tumor growth compared with cisplatin alone (P = 0.00012, nonparametric one-sample Wilcoxon signed rank test).

Figure 3.

A, T/C ratios (independent of EGFR expression level) across the NSCLC xenograft panel for cisplatin plus cetuximab/cisplatin. B, T/C ratios for cisplatin plus cetuximab/cisplatin treatment according to EGFR expression level on a continuous scale. C, T/C ratios for cisplatin plus cetuximab/cisplatin treatment according to EGFR expression level in low and high EGFR expression groups (for derivation of box plot parameters, see Fig. 1).

Figure 3.

A, T/C ratios (independent of EGFR expression level) across the NSCLC xenograft panel for cisplatin plus cetuximab/cisplatin. B, T/C ratios for cisplatin plus cetuximab/cisplatin treatment according to EGFR expression level on a continuous scale. C, T/C ratios for cisplatin plus cetuximab/cisplatin treatment according to EGFR expression level in low and high EGFR expression groups (for derivation of box plot parameters, see Fig. 1).

Close modal

Again, when looking at the association of EGFR expression level with tumor volume in the cisplatin plus cetuximab treatment group and corresponding cisplatin treatment group, a trend for better outcome (lower log2 T/C ratios) at week 3 was apparent in tumors with higher EGFR expression levels (Pearson correlation coefficient r = −0.26; P = 0.12; Fig. 3B). When evaluable tumors were stratified according to the threshold IHC score of 200, log2 T/C ratios were lower (P = 0.0131, Mann–Whitney test) in the high EGFR expression group (n = 22) compared with the low EGFR expression group (n = 17), consistent with a more pronounced inhibition of tumor growth from cisplatin plus cetuximab compared with cisplatin alone in the high EGFR expression group (Fig. 3C).

Adding cetuximab to cisplatin doubled the overall response rate, measured as objective response according to the RECIST model, from 26% (10/39) for cisplatin alone to 51% (20/39) for cisplatin plus cetuximab. This increase in response rate was seen predominantly in the group with high EGFR expression levels (Fig. 4). In tumors with high EGFR expression, the addition of cetuximab to cisplatin markedly increased the response rate from 27% (6/22) to 68% (15/22). In tumors with low EGFR expression, there was no marked difference in response rates between the two treatment arms in the group (cisplatin: 4/17, 24%; cetuximab plus cisplatin: 5/17, 29%.) The association of high EGFR expression with increased response rate was clearly confined to the cisplatin plus cetuximab arm (5/17 vs. 15/22; P = 0.0248, Fisher exact test). In the cisplatin alone arm, there was no significant difference in response rate between the low and high EGFR expression groups (4/17 vs. 6/22; P = 1.00, Fisher exact test).

Figure 4.

Objective response to cisplatin or cisplatin plus cetuximab treatment in the low and high EGFR expression groups.

Figure 4.

Objective response to cisplatin or cisplatin plus cetuximab treatment in the low and high EGFR expression groups.

Close modal

The increase in response rate associated with the addition of cetuximab to cisplatin in NSCLC patient-derived xenograft models with high EGFR expression was most likely due to more than one mechanism. Certain tumors seemed to be highly sensitive to anti-EGFR therapy alone, with the cetuximab treatment making the main contribution to the observed antitumor activity. However, in other tumors, neither cetuximab nor cisplatin alone resulted in the best antitumor activity, but the combination of both agents was needed to achieve the maximal response. Examples for both phenomena are provided in Fig. 5A and B, respectively.

Figure 5.

Antitumor response to cetuximab, cisplatin, and the combination of cisplatin plus cetuximab in two patient-derived NSCLC xenograft models: A, in xenograft IC1TEP (IHC score, 280), the objective response in the cisplatin plus cetuximab group is driven by a high sensitivity to cetuximab. B, in xenograft LXFA 586 (IHC score, 210), only the combination of cisplatin plus cetuximab results in an objective response.

Figure 5.

Antitumor response to cetuximab, cisplatin, and the combination of cisplatin plus cetuximab in two patient-derived NSCLC xenograft models: A, in xenograft IC1TEP (IHC score, 280), the objective response in the cisplatin plus cetuximab group is driven by a high sensitivity to cetuximab. B, in xenograft LXFA 586 (IHC score, 210), only the combination of cisplatin plus cetuximab results in an objective response.

Close modal

Impact of EGFR and KRAS mutation status

None of the tumors was found to harbor an EGFR kinase domain mutation. In 11 of 41 analyzed tumors (27%), a KRAS mutation was detected (see Supplementary Table S1). KRAS mutations were present in tumors with high and low EGFR expression. No association between efficacy of cetuximab (monotherapy or in combination with cisplatin) and KRAS mutation status was found (data not shown). This is in line with clinical results from phase III studies indicating that KRAS mutation status is not a predictive biomarker for cetuximab efficacy in NSCLC (21, 22). The predictive value of EGFR expression for the efficacy of cetuximab in NSCLC seemed to be independent of the KRAS mutation status.

STEPP analysis of response according to IHC score

To further explore whether an EGFR IHC score of 200 was an effective threshold for predicting the activity of cetuximab in the present study, thereby confirming the threshold determined in the FLEX study, response rates according to treatment group were estimated for different EGFR IHC scores in sliding windows according to a STEPP approach. For cetuximab monotherapy compared with control (no active treatment), EGFR IHC scores ≥200 (Fig. 6A) were clearly associated with responses to therapy. In the case of cisplatin plus cetuximab compared with cisplatin alone, the STEPP plot confirmed that a threshold of 200 would effectively differentiate a subgroup of tumors likely to derive the most benefit from the addition of cetuximab to chemotherapy (EGFR IHC score ≥200) from one deriving little or no benefit (Fig. 6B).

Figure 6.

STEPP analysis of objective response rate by IHC score for cetuximab monotherapy compared with no active treatment (A) and cisplatin plus cetuximab compared with cisplatin alone (B). Response rates for each treatment group were calculated in sliding windows across the range of the EGFR IHC score. Each window contained response data from 25% of the analyzed tumors plotted against the median IHC score. OR, objective response.

Figure 6.

STEPP analysis of objective response rate by IHC score for cetuximab monotherapy compared with no active treatment (A) and cisplatin plus cetuximab compared with cisplatin alone (B). Response rates for each treatment group were calculated in sliding windows across the range of the EGFR IHC score. Each window contained response data from 25% of the analyzed tumors plotted against the median IHC score. OR, objective response.

Close modal

Further confirmation that an EGFR IHC score of 200 was an effective threshold in relation to cetuximab activity was provided by an analysis investigating the difference in log2 T/C ratios when the tumors were grouped according to a range of scores from 150, increasing in increments of 10, to 250 (Supplementary Table S2). For cetuximab monotherapy compared with control as well as for cisplatin plus cetuximab compared with cisplatin alone, the most significant T/C differences were seen with threshold scores of 200 and 210. This provides further evidence that a discriminatory EGFR threshold of 200 is effective in relation to the identification of tumors most likely to respond to cetuximab mono- or combination therapy.

Using IHC and response data to select an outcome-based discriminatory IHC score threshold, the FLEX study demonstrated that high tumor EGFR expression could be used to define a subgroup of patients with advanced NSCLC who would derive a meaningful survival benefit from the addition of cetuximab to first-line cisplatin and vinorelbine chemotherapy (11). This observation is in line with the mode of action of cetuximab, which is thought to depend on direct interaction with the EGFR molecule (23).

The current study was designed to explore whether the association between high EGFR expression, as defined in the FLEX analysis by an IHC score of ≥200 on a scale of 0–300, and cetuximab activity could also be demonstrated in vivo in an extensive panel of patient-derived NSCLC xenografts categorized for EGFR expression level. Indeed, this proved to be the case, with objective responses to cetuximab monotherapy seen only in tumors with EGFR expression levels determined to be ≥200. The addition of cetuximab to cisplatin was associated with a marked improvement in response rate in tumors with high EGFR expression (EGFR IHC score ≥200) but had no apparent effect in those tumors with low EGFR expression (EGFR IHC score <200). Interestingly, some patient-derived xenografts with high EGFR expression did not show strong sensitivity to cetuximab or cisplatin alone but did show a strong response when the agents were administered in combination. Response rates in the cisplatin alone treatment groups seemed to be independent of EGFR expression level, being similar in the low and high EGFR expression subgroups.

The in vivo data relating to response from the current analysis are therefore consistent with the response and survival data from the FLEX study in indicating that an EGFR IHC score of ≥200 can predict a treatment benefit in NSCLC associated with the addition of cetuximab to chemotherapy.

In contrast with these clinical and experimental findings in NSCLC, EGFR expression level as determined by IHC does not seem to be of predictive value in relation to the treatment benefit associated with the addition of cetuximab to chemotherapy regimens used in the first-line treatment of metastatic colorectal cancer or recurrent and/or metastatic squamous cell carcinoma of the head and neck (24). In these two settings, the addition of cetuximab to chemotherapy seemed to improve the survival and PFS essentially across the full EGFR IHC score range, with no evidence for a clinically meaningful threshold EGFR IHC score that might identify subgroups of patients with and without benefit. In metastatic colorectal cancer, it has been clearly demonstrated that the KRAS mutational status is a predictive biomarker for cetuximab, with cetuximab benefit limited to those patients who have KRAS wild-type tumors (3, 25, 26). In contrast, in the first-line treatment of advanced NSCLC, KRAS mutational status does not seem to be of predictive value in patients receiving chemotherapy plus cetuximab (21, 22). In line with these observations, no association between the KRAS mutational status and cetuximab efficacy was seen in this study. Thus, although patients with KRAS wild-type metastatic colorectal cancer can derive benefit from cetuximab treatment independently of tumor EGFR expression levels, it is the EGFR expression level that seems to predict cetuximab efficacy in NSCLC, not the KRAS mutation status. These findings provide further evidence that individual biomarkers, which may be predictive for outcome in relation to a particular targeted agent in one tumor type or setting, may not be equally informative for that agent in other tumor types or settings.

To our knowledge, this is one of the first studies where the predictive value of a biomarker has been demonstrated in a clinical phase III trial and also independently, in a large panel of patient-derived xenograft models of the same disease. This is of special interest given the increasing number of studies suggesting the usefulness of patient-derived tumor models in the drug discovery process. In particular, by directly transplanting human tumors into murine hosts, the histopathological features, tumor heterogeneity, genetic profiles, and gene expression patterns can be preserved over several passages of the patient-derived xenograft models in mice (13–16, 27). Interestingly, also the expression of proteins crucial for cancer development like MEK1/2 and p44/p42 MAPK remains stable over different passages as shown by a recent study investigating uveal melanoma patient-derived xenograft models (28). Nevertheless, in the current study, tumors from the biomarker arm coming from the same passage as the treated tumors were used for the determination of biomarkers to prevent any potential bias due to differences in passage number. It has been suggested that patient-derived xenograft models might be able to predict more faithfully the chemoresponsiveness of tumors than xenografts initiated from cancer cell lines (17). Although early studies showed the value of these models in relation to mirroring the response patterns of primary tumors to individual chemotherapeutic agents (17, 29), more recently, a strong correlation has been demonstrated between drug activity in patient-derived xenograft models and clinical antitumor responses to those same drugs in the originally biopsied patients (30).

However, a study investigating whether the predictive value of a molecular biomarker identified in a large phase III study can be confirmed in an extensive panel of patient-derived xenografts has not yet been reported. Although many xenografts used in this study were derived from patients with early-stage tumors, a substantial number was derived from metastasis (see Supplementary Table S1). The majority of the original tumors were treatment naïve. No patient was pretreated with cetuximab. Thus, at least to some extend the original tumors used for the generation of xenografts in this study can be expected to resemble the tumors of patients enrolled in the FLEX study. Of note, responses to cetuximab monotherapy and to cetuximab plus cisplatin in this study were seen in high EGFR-expressing patient-derived xenografts classified as adenocarcinoma, squamous cell carcinoma, and pleomorphic carcinoma (see Supplementary Table S1). This is also in line with the observation made in the FLEX study indicating that the predictive effect of EGFR expression levels for cetuximab activity seemed to be independent of the histologic subtype (11).

The close agreement in this case between the in vivo models and the FLEX study EGFR expression analysis further supports the notion that patient-derived xenograft models might be an effective way to increase the predictive utility of animal tumor models and suggests that they might be used to prioritize candidate biomarkers for testing in the clinical setting.

A final confirmation of the predictive value of high EGFR expression for the efficacy of cetuximab in advanced NSCLC has to come from further clinical studies. Just recently the SWOG trial S0819 (NCT00946712) added measurement of an EGFR IHC score and its association with PFS and overall survival as an additional secondary objective. With an estimated enrollment of 1546 patients, the trial compares overall survival in patients with advanced NSCLC treated with carboplatin, paclitaxel, and bevacizumab, with versus without cetuximab (31). Although the setting is different due to the presence of bevacizumab, the analysis might help to collect further clinical evidence for EGFR expression as a predictive biomarker for cetuximab in NSCLC.

In conclusion, analysis of a panel of patient-derived xenograft models of NSCLC suggested that cetuximab activity was limited to tumors that expressed high levels of EGFR, as defined by an IHC score of ≥200 on a scale of 0–300. This was the case when cetuximab was administered as monotherapy, or when it was administered in combination with cisplatin chemotherapy. The experimental data are therefore consistent with the previously reported biomarker analysis of the phase III FLEX study, which indicated that high EGFR expression (as defined by an IHC score ≥200) may be an effective predictive biomarker for distinguishing patients with NSCLC who are likely to benefit from the addition of cetuximab to platinum-based chemotherapy from those unlikely to benefit (IHC score <200).

No potential conflicts of interest were disclosed.

Conception and design: C. Amendt, C. Stroh

Development of methodology: C. Amendt, C. Stroh

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Friese-Hamim, S. Störkel

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C. Amendt, E. Staub, C. Stroh

Writing, review, and/or revision of the manuscript: C. Amendt, E. Staub, M. Friese-Hamim, S. Störkel, C. Stroh

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C. Amendt, M. Friese-Hamim, C. Stroh

Study supervision: C. Amendt, C. Stroh

This work was sponsored by Merck KGaA. Medical writing services were provided by Dr Jim Heighway of Cancer Communications & Consultancy Ltd and funded by Merck KGaA.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Amado
RG
,
Wolf
M
,
Peeters
M
,
Van Cutsem
E
,
Siena
S
,
Freeman
DJ
, et al
Wild-type KRAS is required for panitumumab efficacy in patients with metastatic colorectal cancer
.
J Clin Oncol
2008
;
26
:
1626
34
.
2.
Hochhaus
A
,
La Rosee
P
,
Muller
MC
,
Ernst
T
,
Cross
NC
. 
Impact of BCR-ABL mutations on patients with chronic myeloid leukemia
.
Cell Cycle
2011
;
10
:
250
60
.
3.
Karapetis
CS
,
Khambata-Ford
S
,
Jonker
DJ
,
O'Callaghan
CJ
,
Tu
D
,
Tebbutt
NC
, et al
K-ras mutations and benefit from cetuximab in advanced colorectal cancer
.
N Engl J Med
2008
;
359
:
1757
65
.
4.
Kudchadkar
R
,
Paraiso
KH
,
Smalley
KS
. 
Targeting mutant BRAF in melanoma: current status and future development of combination therapy strategies
.
Cancer J
2012
;
18
:
124
31
.
5.
Ou
SH
,
Bartlett
CH
,
Mino-Kenudson
M
,
Cui
J
,
Iafrate
AJ
. 
Crizotinib for the treatment of ALK-rearranged non-small cell lung cancer: a success story to usher in the second decade of molecular targeted therapy in oncology
.
Oncologist
2012
;
17
:
1351
75
.
6.
Penault-Llorca
F
,
Bilous
M
,
Dowsett
M
,
Hanna
W
,
Osamura
RY
,
Ruschoff
J
, et al
Emerging technologies for assessing HER2 amplification
.
Am J Clin Pathol
2009
;
132
:
539
48
.
7.
Albarello
L
,
Pecciarini
L
,
Doglioni
C
. 
HER2 testing in gastric cancer
.
Adv Anat Pathol
2011
;
18
:
53
9
.
8.
Carlson
RW
,
Moench
SJ
,
Hammond
ME
,
Perez
EA
,
Burstein
HJ
,
Allred
DC
, et al
HER2 testing in breast cancer: NCCN Task Force report and recommendations
.
J Natl Compr Canc Netw
2006
;
4
Suppl 3
:
S1
22
.
9.
Pujol
JL
,
Pirker
R
,
Lynch
TJ
,
Butts
CA
,
Rosell
R
,
Shepherd
FA
, et al
Meta-analysis of individual patient data from randomized trials of chemotherapy plus cetuximab as first-line treatment of advanced non-small cell lung cancer
.
Lung Cancer
2014
;
83
:
211
4
.
10.
Pirker
R
,
Pereira
JR
,
Szczesna
A
,
von Pawel
J
,
Krzakowski
M
,
Ramlau
R
, et al
Cetuximab plus chemotherapy in patients with advanced non-small-cell lung cancer (FLEX): an open-label randomised phase III trial
.
Lancet
2009
;
373
:
1525
31
.
11.
Pirker
R
,
Pereira
JR
,
von Pawel
J
,
Krzakowski
M
,
Ramlau
R
,
Park
K
, et al
EGFR expression as a predictor of survival for first-line chemotherapy plus cetuximab in patients with advanced non-small-cell lung cancer: analysis of data from the phase 3 FLEX study
.
Lancet Oncol
2012
;
13
:
33
42
.
12.
PL03.05 An intergroup randomized phase III comparison of standard-dose (60 Gy) vs high-dose (74 Gy) chemoradiotherapy (CRT) +/- cetuximab (cetux) for stage III non-small cell lung cancer (NSCLC): results on cetux from RTOG 0617
.
Clin Adv Hematol Oncol
2014
;
12
(
1 Suppl 1
):
2
4
.
13.
Kopetz
S
,
Lemos
R
,
Powis
G
. 
The promise of patient-derived xenografts: the best laid plans of mice and men
.
Clin Cancer Res
2012
;
18
:
5160
2
.
14.
Julien
S
,
Merino-Trigo
A
,
Lacroix
L
,
Pocard
M
,
Goere
D
,
Mariani
P
, et al
Characterization of a large panel of patient-derived tumor xenografts representing the clinical heterogeneity of human colorectal cancer
.
Clin Cancer Res
2012
;
18
:
5314
28
.
15.
DeRose
YS
,
Wang
G
,
Lin
YC
,
Bernard
PS
,
Buys
SS
,
Ebbert
MT
, et al
Tumor grafts derived from women with breast cancer authentically reflect tumor pathology, growth, metastasis and disease outcomes
.
Nat Med
2011
;
17
:
1514
20
.
16.
Bertotti
A
,
Migliardi
G
,
Galimi
F
,
Sassi
F
,
Torti
D
,
Isella
C
, et al
A molecularly annotated platform of patient-derived xenografts (“xenopatients”) identifies HER2 as an effective therapeutic target in cetuximab-resistant colorectal cancer
.
Cancer Discov
2011
;
1
:
508
23
.
17.
Fichtner
I
,
Rolff
J
,
Soong
R
,
Hoffmann
J
,
Hammer
S
,
Sommer
A
, et al
Establishment of patient-derived non-small cell lung cancer xenografts as models for the identification of predictive biomarkers
.
Clin Cancer Res
2008
;
14
:
6456
68
.
18.
Gnirke
A
,
Melnikov
A
,
Maguire
J
,
Rogov
P
,
LeProust
EM
,
Brockman
W
, et al
Solution hybrid selection with ultra-long oligonucleotides for massively parallel targeted sequencing
.
Nat Biotechnol
2009
;
27
:
182
9
.
19.
Therasse
P
,
Arbuck
SG
,
Eisenhauer
EA
,
Wanders
J
,
Kaplan
RS
,
Rubinstein
L
, et al
New guidelines to evaluate the response to treatment in solid tumors
.
J Natl Cancer Inst
2000
;
92
:
205
16
.
20.
Bonetti
M
,
Gelber
RD
. 
A graphical method to assess treatment-covariate interactions using the Cox model on subsets of the data
.
Stat Med
2000
;
19
:
2595
609
.
21.
Khambata-Ford
S
,
Harbison
CT
,
Hart
LL
,
Awad
M
,
Xu
LA
,
Horak
CE
, et al
Analysis of potential predictive markers of cetuximab benefit in BMS099, a phase III study of cetuximab and first-line taxane/carboplatin in advanced non-small-cell lung cancer
.
J Clin Oncol
2010
;
28
:
918
27
.
22.
O'Byrne
KJ
,
Gatzemeier
U
,
Bondarenko
I
,
Barrios
C
,
Eschbach
C
,
Martens
UM
, et al
Molecular biomarkers in non-small-cell lung cancer: a retrospective analysis of data from the phase 3 FLEX study
.
Lancet Oncol
2011
;
12
:
795
805
.
23.
Li
S
,
Schmitz
KR
,
Jeffrey
PD
,
Wiltzius
JJ
,
Kussie
P
,
Ferguson
KM
. 
Structural basis for inhibition of the epidermal growth factor receptor by cetuximab
.
Cancer Cell
2005
;
7
:
301
11
.
24.
Licitra
L
,
Storkel
S
,
Kerr
KM
,
Van Cutsem
E
,
Pirker
R
,
Hirsch
FR
, et al
Predictive value of epidermal growth factor receptor expression for first-line chemotherapy plus cetuximab in patients with head and neck and colorectal cancer: analysis of data from the EXTREME and CRYSTAL studies
.
Eur J Cancer
2013
;
49
:
1161
8
.
25.
Bokemeyer
C
,
Bondarenko
I
,
Hartmann
JT
,
de Braud
F
,
Schuch
G
,
Zubel
A
, et al
Efficacy according to biomarker status of cetuximab plus FOLFOX-4 as first-line treatment for metastatic colorectal cancer: the OPUS study
.
Ann Oncol
2011
;
22
:
1535
46
.
26.
Van Cutsem
E
,
Kohne
CH
,
Lang
I
,
Folprecht
G
,
Nowacki
MP
,
Cascinu
S
, et al
Cetuximab plus irinotecan, fluorouracil, and leucovorin as first-line treatment for metastatic colorectal cancer: updated analysis of overall survival according to tumor KRAS and BRAF mutation status
.
J Clin Oncol
2011
;
29
:
2011
9
.
27.
Rubio-Viqueira
B
,
Jimeno
A
,
Cusatis
G
,
Zhang
X
,
Iacobuzio-Donahue
C
,
Karikari
C
, et al
An in vivo platform for translational drug development in pancreatic cancer
.
Clin Cancer Res
2006
;
12
:
4652
61
.
28.
Laurent
C
,
Gentien
D
,
Piperno-Neumann
S
,
Némanti
F
,
Nicolas
A
,
Tesson
B
, et al
Patient-derived xenografts recapitulate molecular features of human uveal melanomas
.
Mol Oncol
2013
;
7
:
625
6
.
29.
Fiebig
HH
,
Burger
A
. 
Human Tumor Xenografts and Explants
. In:
Teicher
B
,
editor
. 
Tumor Models in Cancer Research
:
Humana Press
; 
2002
.
p.
113
37
.
30.
Hidalgo
M
,
Bruckheimer
E
,
Rajeshkumar
NV
,
Garrido-Laguna
I
,
De Oliveira
E
,
Rubio-Viqueira
B
, et al
A pilot clinical study of treatment guided by personalized tumorgrafts in patients with advanced cancer
.
Mol Cancer Ther
2011
;
10
:
1311
6
.
31.
Redman
MW
,
Crowley
JJ
,
Herbst
RS
,
Hirsch
FR
,
Gandara
DR
. 
Design of a phase III clinical trial with prospective biomarker validation: SWOG S0819
.
Clin Cancer Res
2012
;
18
:
4004
12
.