Presented at the 2011 European Multidisciplinary Cancer Conference; September 23–27, 2011; Stockholm, Sweden.
ELK, VKC and AJI contributed to concept and design. ELK, H-JL, W-FC, W-CS, MR, FLM, MTL, VKC and AJI were involved in the collection and assembly of data. ELK, H-JL, W-FC, MR, FLM, MTL, VKC and AJI contributed to data analysis and interpretation. ELK, GIS, H-JL, MR, FLM, MTL, VKC and AJI were involved in manuscript writing. ELK, GIS, SMC, H-JL, CRB, W-FC, W-CS, MR, FLM, MTL, VKC and AJI gave final approval of manuscript. ELK, GIS, SMC, CRB, H-JL, and W-FC contributed to the provision of study materials or patients.
The efficacy of afatinib, an irreversible ErbB Family Blocker, was evaluated in patients who had 1 of 4 categories of solid tumors with epidermal growth factor receptor/human epidermal growth factor receptor 2 (EGFR/HER2) gene amplification or EGFR-activating mutations.
Patients with previously treated but ErbB inhibitor-naive esophagogastric, biliary tract, urothelial tract, or gynecologic cancers (lung cancers were excluded) harboring EGFR/HER2 gene amplification or high polysomy were identified by fluorescence in situ hybridization (FISH). Tumors were also screened for EGFR mutations. The primary endpoint was the objective response rate; secondary endpoints included the clinical benefit rate, pharmacokinetics, and safety.
Of 385 prescreened patients, 38 had FISH-positive tumors (10 with EGFR amplification and 29 with HER2 amplification or high polysomy [1 tumor had EGFR/HER2 high polysomy]; none had EGFR-activating mutations), and 20 patients received treatment with afatinib 50 mg daily. The objective response rate was 5% (1 of 20 patients), and the best objective response included 1 complete response. Eight patients experienced stable disease. The most frequently reported adverse events were diarrhea, rash, and decreased appetite. The trial closed early because of slow recruitment.
Dysregulation of the ErbB Family has been observed in numerous malignancies, making this signaling pathway an attractive therapeutic target. The ErbB (erythroblastic leukemia viral [v-erb-b] oncogene homolog) receptor Family comprises 4 structurally related, membrane-bound receptors: epidermal growth factor receptor (EGFR/ErbB1), human epidermal growth factor receptor (HER2/ErbB2), ErbB3 (HER3), and ErbB4 (HER4). In cancers that depend on a single signaling pathway for survival and growth, commonly termed “oncogene addiction,” targeted inhibition of the signaling pathway can lead to tumor death. In a growing number of tumors, oncogene addiction has been observed in the setting of a dysregulated ErbB signaling pathway. For instance, activating mutations in the tyrosine kinase domain of EGFR occur in a subset of nonsmall cell lung cancers (NSCLCs), and inhibition of these oncogenic drivers with EGFR inhibitors, such as gefitinib and erlotinib, has produced dramatic responses.[4-10] EGFR or HER2 gene amplification also has been identified as an oncogenic driver in a variety of tumor histologies, which are sensitive to EGFR or HER2 inhibition.[11-14]
Afatinib (BIBW 2992) is a potent, orally bioavailable ErbB Family Blocker that irreversibly blocks the signaling from EGFR/HER1 (half-maximal inhibitory concentration [IC50], 0.5 nM), HER2 (IC50, 14 nM), and ErbB4 (IC50, 1 nM) tyrosine kinases and transphosphorylation of ErbB3.[15, 16] In vitro, afatinib suppresses the cellular kinase activity of wild-type and activated EGFR and HER2 mutants, including erlotinib-resistant isoforms. Afatinib also has demonstrated preclinical activity across a range of ErbB-driven, in vivo cancer models.[15, 17] Preclinical data suggested that afatinib has the potential to benefit patients with tumors driven by the ErbB pathway regardless of histology, which prompted a study design based on tumor molecular characteristics rather than the primary site of origin. Such biomarker-driven patient recruitment has been used successfully in trials of other targeted cancer therapies.[18-20] For the current phase 2, open-label, exploratory trial, we evaluated the efficacy of afatinib in patients most likely to benefit from ErbB Family blockade who were prospectively screened for EGFR/HER2 gene amplification or EGFR-activating mutations.
MATERIALS AND METHODS
This was a multicenter, phase 2, open-label trial in patients whose tumors had been genetically prescreened for EGFR/HER2 gene amplification or EGFR-activating mutations. It was conducted in 17 centers (14 in the United States and 3 in Taiwan) between November 2008 and November 2010. The study design, which involved a 2-step consent process, is illustrated in Figure 1. Patients' tumors were required to have documented gene amplification before patients received afatinib treatment; therefore, genetic prescreening of archived tumor samples comprised fluorescence in situ hybridization (FISH) testing for EGFR and HER2 amplification, which was performed by a central laboratory (Massachusetts General Hospital, Boston, Mass). Patients who had FISH-positive tumors provided written consent for further screening for suitability and, if they were eligible, afatinib treatment. Given the known sensitivity to EGFR inhibitors, patients who harbored externally documented, known EGFR-activating mutations also were allowed to enroll in the trial. The study was conducted in accordance with the Declaration of Helsinki, local laws, and the International Conference on Harmonization Good Clinical Practice Guidelines, and it was approved by the relevant regulatory and independent ethics committees or institutional review boards.
Eligible patients aged ≥18 years required a histologically confirmed diagnosis of advanced cancer of 1 of either: category 1 (gastric, gastroesophageal junction, or esophageal cancer), category 2 (biliary or gallbladder cancer), category 3 (transitional cell carcinoma [TCC] of the urothelial tract), or category 4 (gynecologic cancers). Other key eligibility criteria included measurable disease according to Response Evaluation Criteria in Solid Tumors version 1.0 (RECIST 1.0), life expectancy ≥3 months, and an Eastern Cooperative Oncology Group performance score from 0 to 2. All patients had to have progressed after ≥1 line of prior chemotherapy and were excluded if they had received previous treatment with gefitinib, erlotinib, lapatinib, or other EGFR tyrosine kinase inhibitors (TKIs). Patients were required to have tumors with EGFR or HER2 gene amplification or tumors that harbored known activating EGFR mutations. However, the protocol was amended later to also allow patients who had tumors with high EGFR/HER2 polysomy (≥4 gene copies in ≥40% of cells), as determined by FISH, to enter the study. The rationale for including patients who had high EGFR/HER2 polysomy was that multiple gene copies result in higher protein expression, which may be amenable to EGFR inhibition.
Tumor Genetic Testing
Tumor samples were screened for molecular genetic abnormalities by FISH; and individual cell EGFR, HER2, and centromeric probe 7 (CEP7) (the reference probe) signals were recorded. FISH-positive tumors had either high-level amplification (numerous loose or tight clusters of EGFR/HER2 signals, or atypically large signals, or an EGFR or HER2/CEP7 ratio >5.0) or low-level amplification (EGFR or HER2/CEP7 ratio >2.2 and <5.0). After a protocol amendment, tumors that had ≥4 gene copies of EGFR or HER2 in ≥40% of cells (high polysomy) were considered FISH-positive. EGFR and v-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog (KRAS) mutation analyses (KRAS exon 2 mutations at codons 12 and 13 and EGFR mutations, including exon 19 deletions, exon 20 insertions, and the mutations G719X, T790M, L858R, and L861Q) were performed retrospectively on eligible tumor samples using a high-sensitivity, allele-specific mutation-detection assay (SNaPshot Multiplex Kit; Applied Biosystems/Life Technologies, Carlsbad, Calif).
Patients received afatinib 50 mg daily administered orally or through a gastrostomy tube in 28-day cycles until they developed either disease progression or undue adverse events (AEs). Two dose reductions to 40 mg daily and, if deemed necessary, subsequently to 30 mg daily were permitted in patients who experienced certain AEs; further dose reductions were not allowed.
The primary endpoint was objective response rate (ORR; complete response [CR] or partial response [PR] according to RECIST 1.0 criteria). Imaging assessments for tumor response were performed at 6 weeks and at 12 weeks and then every 8 weeks until treatment completion. Secondary efficacy endpoints included the clinical benefit rate (CR, PR, or stable disease according to RECIST 1.0 criteria) and progression-free survival (PFS). Safety assessments encompassed the occurrence and intensity of diarrhea, skin rash, AEs resulting in dose reduction or treatment discontinuation, and the incidence of other AEs according to the Common Terminology Criteria for Adverse Events version 3.0.
Plasma samples were collected during Cycle 1 on Days 1 and 15 (before drug administration), during Cycle 2 on Days 1 (1, 3, and 6 [2 patients] hours after drug administration) and 15 (before drug administration), during Cycle 3 on Day 1 (before and 2 hours after drug administration), and during Cycle 4 on Day 1 (before drug administration). Afatinib plasma concentrations were analyzed by validated high-performance liquid chromatography/tandem mass spectrometry at Boehringer Ingelheim (Biberach, Germany).[21, 22]
Sample Size Determination and Statistical Methods
The sample size was based on the assumption that the underlying response rate for the selected patient population would be 20%. Sample size calculations also met the requirement for the exact 90% confidence interval (CI) to rule out a 10% response rate if the true response rate was ≥20%. The recommended total sample size was 48 patients, with 12 patients in each of the 4 tumor categories; 10 responders were required overall to meet study assumptions. The planned primary analysis was to estimate the proportion of patients who achieved an objective response. An exact 90% Clopper–Pearson CI was to be calculated for this estimate for the overall population and for each tumor category. Because of early study termination, only data on the primary efficacy endpoint were summarized; no CIs were produced, and efficacy and safety results were presented for the overall population rather than by tumor category.
Overall, 385 patients were prescreened for study inclusion; demographics and characteristics are listed in Table 1. Of the 38 patients who had FISH-positive tumors identified by central laboratory testing, 10 tumors tested positive for EGFR amplification/high polysomy (6 esophagogastric tumors, 4 urothelial TCCs), and 29 had HER2 amplification/high polysomy (17 esophagogastric tumors, 5 biliary tract tumors, 3 urothelial TCCs, and 4 gynecologic tumors). One of the urothelial TCCs mentioned above tested positive for EGFR and HER2 high polysomy. Twenty eligible patients received ≥1 afatinib dose (Table 2); none had tumors with EGFR-activating mutations. Among those 20 patients, 4 had EGFR-amplified tumors (3 with high-level amplification and 1 with low-level amplification), and 13 tumors were positive for HER2 amplification (10 with high-level amplification and 3 with low-level amplification). Two tumors had high polysomy (EGFR, n = 1; EGFR and HER2, n = 1). In the same group of 20 patients, 1 had a tumor that was positive for EGFR amplification and high polysomy. Locally confirmed FISH-positive results could not be confirmed subsequently by central laboratory testing in 2 patients. This study was terminated early because of recruitment challenges; no safety or efficacy findings influenced that decision.
Table 1. Patient Demographics and Characteristics: Prescreened Patients
Gastric, Gastroesophageal Junction, or Esophageal Cancer: Category 1, n = 128
Biliary or Gallbladder Cancer: Category 2, n = 61
Transitional Cell Carcinoma of Urothelial Tract: Category 3, n = 46
Gynecologic Cancer: Category 4, n = 150
Abbreviations: EGFR, epidermal growth factor receptor; FISH, fluorescence in situ hybridization; HER2, human epidermal growth factor receptor 2.
FISH status was determined by the central laboratory.
FISH positive indicates either EGFR or HER2 amplification, or EGFR or HER2 high polysomy.
Table 2. Patient Demographics and Characteristics: Treated Patients, n = 20
Afatinib 50 mg Daily: No. of Patients (%)
Abbreviations: ECOG, Eastern Cooperative Oncology Group; EGFR, epidermal growth factor receptor; FISH, fluorescence in situ hybridization; HER2, human epidermal growth factor receptor 2; KRAS, v-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog; SD, standard deviation.
In total, 18 patients had FISH-positive results based on central laboratory findings, and another 2 patients were enrolled based on FISH-positive results from the local laboratory.
High-level amplification was defined as the presence of loose or tight clusters of EGFR or HER2 signals too numerous to count, or atypically large signals, or an EGFR or HER2/centromeric probe 7 (CEP7) ratio of >5.0. Low-level amplification was defined as tumors with an EGFR or HER2/CEP7 ratio of >2.2 and <5.0.
EGFR and HER2 high polysomy, based on central laboratory findings, were only tested for patients who enrolled after a protocol amendment that allowed their inclusion.
The 20 patients who received ≥1 dose of afatinib were included in the efficacy assessment. Responses are summarized in Table 3, and EGFR/HER2 status is also presented. The ORR for afatinib was 5% (1 of 20 patients) in this previously treated but ErbB TKI-naive population. There were no PRs; however, 1 patient achieved a CR. This patient was an Asian woman aged 64 years with stage IV endometrial carcinoma whose tumor was FISH-positive for high-level HER2 and FISH-negative for EGFR amplification (no EGFR or KRAS mutations were detected); prior therapies included palliative chemotherapy with paclitaxel and carboplatin. The PFS in this patient was censored at 252 days, which was the last available tumor assessment. Eight patients (40%) had stable disease as their best response, and the clinical benefit rate was 45% (9 of 20 patients). Figure 2 depicts a waterfall plot of data from imaging analyses in the 17 patients who had evaluable RECIST data. Approximately 50% of patients had a decrease in tumor size (in terms of the maximum decrease in any 1 dimension, based on sums of the greatest dimensions, collected as part of the RECIST assessment), although the change was small in most patients.
Table 3. Best Overall Response, According to Response Evaluation Criteria in Solid Tumors, Version 1.0
No. of Patients (%)
EGFR or HER2 Gene Amplification or High Polysomy in the Central Laboratory-Confirmed, FISH-Positive Subgroup, n = 18
Two patients on the trial who received afatinib tested FISH-negative according to the central laboratory (a KRAS [v-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog] mutation [exon 2, G12-34G] was detected in 1 of these patients).
Overall, the median treatment duration was 83.5 days (range, 9–237 days). The most common reason for treatment discontinuation was disease progression (n = 15; 75%), followed by drug-related AE (n = 2; 10%), other AE, patient refusal to continue with drug, and other reason (n = 1 each; 5% each). All patients reported ≥1 AE, most commonly diarrhea, rash, and decreased appetite (Table 4). Drug-related AEs were reported in 19 of 20 patients. Twelve grade 3 treatment-related events were reported in 9 of 20 patients (45%), including diarrhea (n = 4), dermatitis acneiform (n = 1), fatigue (n = 3), decreased weight (n = 1), paronychia (n = 2), and decreased appetite (n = 1). Eight patients required dose reductions because of AEs. All drug-related grade 3 AEs resolved after dose reduction (n = 6) or drug discontinuation (n = 2), except for paronychia (n = 1), decreased appetite (n = 1), and fatigue (n = 2). Serious AEs (SAEs) were reported in 6 patients, possibly related to treatment in 2 patients. One patient with metastatic bladder cancer experienced grade 5 dyspnea possibly related to afatinib. That patient had multiple comorbidities, including a history of deep vein thrombosis; workup revealed a low-probability ventilation-perfusion scan for pulmonary embolism with no significant pulmonary infiltrate. This patient subsequently died. Autopsy revealed extensive disease progression, including microscopic pulmonary tumor emboli. The second SAEs possibly related to afatinib were grade 3 diarrhea and grade 2 dehydration, which required hospitalization in a patient with gallbladder carcinoma. Five patients discontinued trial medication because of AEs; 3 discontinuations were attributed to the treatment (hematuria [n = 1], fatigue, increased creatinine, hyperbilirubinemia, and increased aspartate aminotransferase and alkaline phosphatase levels [n = 1], and weight loss [n = 1]).
Table 4. Adverse Events Observed in ≥10% of Patients at Any Common Terminology Criteria for Adverse Events Grade or in Any Patient at Common Terminology Criteria for Adverse Events Grade ≥3, n = 20
Afatinib 50 mg Daily: No. of Patients (%)
These included 1 grade 5 adverse event that led to death.
In 12 evaluable patients, steady-state afatinib levels were reached at the latest by Day 15, and predose plasma concentrations appeared stable over the observed treatment periods (Table 5). Overall variability of predose plasma concentrations was moderate to high, with a geometric coefficient of variation (gCV) from 65.1% to 82.3%. On Day 1 of Cycle 2, afatinib plasma concentrations increased in the first 3 hours after initial administration of afatinib 50 mg, with geometric mean (gMean) values of 42.4 ng/mL (range, 18.5–117 ng/mL) at 1 hour and 53.4 ng/mL (range, 24.3–95.2 ng/mL) at 3 hours.
Table 5. Trough Plasma Concentrations (Cpre,ss Values) of Afatinib After Multiple Oral Administrations of Afatinib 50 mg Daily
Afatinib 50 mg Daily
No. of Patients
Abbreviations: Cpre,ss,15 (29,57,85), trough plasma concentration after the 15th (29th, 57th, and 85th) oral dose; gMean, geometric mean; gCV, geometric coefficient of variation.
This was a phase 2, open-label, exploratory trial of afatinib in patients with previously treated solid tumors that were screened prospectively for EGFR or HER2 gene amplification. The underlying rationale was that determinants of sensitivity to targeted agents may be genetic rather than histologic; however, the trial was designed to include 4 specific tumor categories (gastric, gastroesophageal junction, or esophageal cancer; biliary or gallbladder cancer; TCC of the urothelial tract; and gynecologic cancers) to facilitate data interpretation. The trial was terminated early because of recruitment challenges, which prevented full accrual within the planned 2-year study period. No safety or efficacy findings influenced this decision.
Review of the literature suggests that, in gastric cancer, the estimated frequency of HER2 gene amplification is 22%, and the estimated frequency of HER2 overexpression ranges from 7% to 43%. HER2 and EGFR are overexpressed in 10% and 46%, respectively, of gallbladder cancers. Adenocarcinoma of the gastroesophageal junction rarely presents with EGFR mutation, and no mutations have been detected within the tyrosine kinase domain or exons 19/21 of EGFR in patients with bladder cancer.[27, 28] In the current study, the observed HER2/EGFR amplification rates were relatively lower than those in published reports,[23-26] whereas the observed absence of EGFR-activating mutations was consistent with earlier studies.[27, 28]
There was evidence of antitumor activity in this heavily treated, but ErbB TKI-naive, population. One patient with endometrial cancer and high-level HER2 amplification achieved a CR. It is noteworthy that no responses were observed in a phase 2 trial of trastuzumab in patients with HER2-overexpressed or HER2-amplified endometrial cancers, suggesting possible differences between antibodies and TKIs in this setting. With afatinib treatment, 8 of 20 patients experienced a best response of stable disease, suggesting modest single-agent activity in patients with other ErbB-driven tumors. Although the response rates appear to be low in this broad patient population, the results need to be interpreted with caution in light of the limited sample size. In select patients, afatinib therapy may improve outcomes, as previously reported.[9, 10, 30] The observed AE and pharmacokinetic profiles for afatinib were consistent with previous studies.[21, 30, 31] The most frequently reported AEs included diarrhea and rash.
Single-agent therapy with afatinib in this small, pretreated patient population that was selected for EGFR/HER2 gene amplification demonstrated a low ORR (5%). However, the small sample size because of the early trial termination and the low patient numbers by tumor category meant that the study objectives could not be fully assessed, and interpretation of the findings was difficult. Factors potentially influencing the low response rate include the possibility that gene amplification does not act as the sole oncogenic driver in the included tumor types. EGFR mutations are drivers in subsets of NSCLCs, and, it was reported recently that HER2 mutations confer sensitivity to afatinib in NSCLC. However, EGFR mutations were not identified in our population, and the prospective identification of HER2 mutations was not an eligibility criterion. The HER2 mutation status of the patient with an endometrial tumor who experienced a CR is unknown. Another explanation for low response rates in the setting of gene amplification is that resistance mechanisms are recruited so readily that the roles of ErbB signaling and targeted inhibition are masked. The role of polysomy in oncogenic signaling remains unclear, because only 3 patients with polysomy were recruited.
The development and evaluation of a biomarker within a single study of multiple tumor types represents a further level of complexity over studies evaluating a single biomarker within 1 tumor type, because each biomarker may have a different impact or weight on clinical outcomes. For instance, although BRAF V600E mutations predict a dramatic benefit with vemurafenib in patients with melanoma, the drug has demonstrated only modest activity in colorectal cancers with BRAF V600E mutations. Nonetheless, although exploratory studies are important for testing clinical scientific hypotheses, small sample sizes and the absence of control treatment arms limit the predictive value of the biomarker in question unless drug activity with respect to the biomarker is dramatic. In such cases, the use of biomarker-related eligibility criteria may allow for a “signal” that could be lost or delayed in an unselected study design.
Prescreening as part of our trial design was a major challenge to patient recruitment, and only 20 patients received afatinib rather than the planned 12 patients per category (48 patients in all). Prescreening the desired number of patient tumors for uncommon genetic lesions relies on oncologists to refer patients for screening, the coordination of which is particularly challenging when multiple cancer types are involved within 1 trial. This obstacle may be obviated in the future as genetic screening across tumor types for multiple molecular lesions becomes standard clinical practice.
In conclusion, the single-agent activity of afatinib in this small trial was limited, yet encouraging, in selected patients with manageable tolerability. The trial was terminated early because of the difficulty in recruiting sufficient patients within the prespecified timeframe. The implementation of a biomarker-driven approach for patient selection in this setting was challenging, and alternative models for such a setting are needed for the future.
This study was supported by Boehringer Ingelheim. Medical writing assistance, supported financially by Boehringer Ingelheim, was provided by Aurora O'Brate of Ogilvy Healthworld.
CONFLICT OF INTEREST DISCLOSURES
Ms. Robohn, Ms. Le Maulf, Dr. Lobmeyer, and Dr. Chand are employees of Boehringer Ingelheim.