Anti-cancer effect of afatinib, dual inhibitor of HER2 and EGFR, on novel mutation HER2 E401G in models of patient-derived cancer

Human lung cancer cell line NCI-H2170 was purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA); cells were cultured in RPMI-1640 medium supplemented with 10% FBS. Afatinib and lapatinib were purchased from LC laboratories (Woburn, MA, USA), and trastuzumab and pertuzumab were purchased from Chugai Pharmaceutical Co. (Tokyo, Japan). Immunodeficient Balb/c Rag-2^−/− Jak3^−/− (BRJ) mice, which lack mature T and B lymphocytes and natural killer cells, were provided by Prof. Seiji Okada (Kumamoto University, Kumamoto, Japan).

To examine drug efficacy of the agents mentioned above, we established PDX using cancer tissue obtained from the patient of unknown primary. The cancer tissue used to prepare the cancer models was obtained from a portion of a skin biopsy specimen collected for clinical use with Institutional Review Board approval (2020–07-R-10) and written informed consent from the patient for use for this research. The cancer tissue was subcutaneously transplanted into the dorsal flanks of 7-week-old female BRJ mice (n = 2) under anesthesia with midazolam, medetomidine, and butorphanol tartrate, and three or four in vivo passages were performed to establish PDX for drug efficacy evaluation experiments (n = 42). Tumors excised from mice engrafted patient tumors were defined as the first generation (G1), and the generations of tumors excised at passaging were defined as G2, G3, and G4, in that order. These characteristics were evaluated using morphology, immunohistochemistry (IHC), and droplet digital PCR. CTOS was established according to the method previously described [13, 14], using a portion of the G3 tumor of PDX in this study. Briefly, tumors were minced and digested using Liberase™ DH (Roche, Basel, Switzerland), after which CTOS size fractions of 100- or 40- μm were collected using a cell strainer. Fractions were cultured in StemPro™ hESC SFM (Life Technologies, Carlsbad, CA) supplemented with 8 ng/mL bFGF (Life Technologies, Carlsbad, CA) and 0.1 mM 2-mercaptoethanol (Wako Chemical Co., Tokyo, Japan). CTOS were established as CTOS lines after at least two in vivo passages of subcutaneous transplantation of CTOS into BRJ mice under anesthesia, and CTOS in the 40–70 μm size fraction were used for evaluation of drug efficacies.

Patient tumor samples and tumors excised from patient-derived models were stained with IHC and Hematoxylin Eosin (HE). Primary antibodies used for IHC were as follows: HER2 (HER2/neu, Agilent Technologies, Glostrup, Denmark), CK7 (OV-TL12/30, Agilent Technologies, Glostrup, Denmark), CK20 (Ks20.8, Agilent Technologies, Glostrup, Denmark), and GATA3 (L50–823, Cell Marque, The Hague, Netherlands). Glass slide samples were digitized with a NanoZoomer S60 digital slide scanner (Hamamatsu Photonics, Hamamatsu, Japan).

Droplet digital PCR (ddPCR™) for analyzing the ratio of E401G and wild type HER2 was performed as described previously [10]. Copy number assay using ddPCR was performed with a HER2 probe set (dHsaCP1000116, Bio-Rad Laboratories, Inc., USA) and the reference gene, RPP30, probe set (dHsaCP2500350, Bio-Rad Laboratories, Inc., USA), according to the manufacturer’s protocol.

To explore appropriate treatment for the patient, we examined drug efficacy using PDX described above (n = 6 per group). We also used xenograft using H2170 cells with wild-type HER2 amplification to compare the effects with PDX with HER2 E401G amplification (n = 3 or 4 per group). H2170 cells (1 × 10⁷ each) were injected into the dorsal flanks of 7-week-old female BRJ mice under anesthesia with midazolam, medetomidine, and butorphanol tartrate. Tumor sizes were measured with calipers twice per week and tumor volumes were calculated as V = 1/2 × [(the shortest diameter)² × (the longest diameter)]. For drug efficacy evaluation, drug or vehicle administration was started on the day that the tumor size reached 150–200 mm³ for PDX and 100–200 mm³ for H2170 xenograft (in both cases defined as “day 1”). On day 22, mice were euthanized by cervical dislocation and excised tumors were photographed. Regarding the administration of the drugs, sub-maximal tolerated doses of afatinib and lapatinib were reported to be 20–25 mg/kg [11, 12, 15] and 100–150 mg/kg [11, 16, 17], respectively, so the initial experiments were performed with 20 mg/kg of afatinib and 100 mg/kg of lapatinib. However, due to a lack of efficacy in the lapatinib 100 mg/kg group, a lapatinib 150 mg/kg group was added. Afatinib and lapatinib were prepared in 0.5 w/v (%) methyl cellulose and administered orally once daily for 5 days per week. Trastuzumab and pertuzumab were administered intraperitoneally once a week at 30 mg/kg diluted in saline according to the literature published during the time of their development [18, 19]. PDX excised tumors were cryopreserved and analyzed for HER2 copy number by droplet digital PCR. In each experiment, the mice were randomly assigned into two or three groups. Mice were maintained in a specific pathogen-free facility within the Analytical Research Center for Experimental Sciences, Saga University. Mice were euthanized by cervical dislocation when a loss of more than 20% baseline body weight occurred or other human endpoints such as lethargy. All animal studies were performed in accordance with animal research protocols approved by the Institutional Review Board of Saga University (A2021–022-0) and the university’s institutional guidelines. The study was completed and reported in compliance with the recommendations of Animals in Research: Reporting In Vivo Experiments (ARRIVE) guidelines.

CTOSs prepared by the method described above were counted and seeded into 96-well plates at approximately 10 CTOSs/well. Twenty-four hours later, medium containing each drug was added to each well to achieve the targeted final concentration, then the culture was continued. Seven days after the addition of the drugs, ATP content was measured with CellTiter-Glo Luminescent Cell Viability Assays (Promega, Madison, WI, USA), and the content was adjusted with vehicle-treated control.

H2170 cells were seeded into 96-well plates at 7 × 10³ cells/well; 24 hours later each drug was added, then the culture was continued. Cell growth was measured with a Cell Counting Kit-8 (Dojindo Molecular Technology, Kumamoto, Japan) on the third day after the addition of the drugs, and the content was adjusted using vehicle-treated control.

Data are expressed as mean and standard deviation (SD). Differences between two groups were tested with the Wilcoxon rank-sum test. Differences among three groups were tested with the Kruskal-Wallis test and Dunn’s test for pairwise comparisons. For all comparisons, P < 0.05 was considered statistically significant. Most calculations were performed using JMP Pro 15.2.0 (SAS Institute Inc., USA). Dose response curves were plotted and half maximal inhibitory concentration (IC₅₀) values were calculated with the DRC package (version 3.0.1) for R (version 4.1.3) in RStudio (version 2022.2.1.461; RStudio PBC, Boston, Massachusetts, USA).

The clinical course of a 67-year-old Japanese woman with carcinoma of unknown primary up to enrollment in the JUPITER trial (jRCT2031180150) [20], a basket trial of trastuzumab and pertuzumab combination therapy, was described in detail in our previous report [10]. After initial treatment with carboplatin and gemcitabine, the patient’s cancer gradually became refractory to the therapy. The FoundationOne CDx® test, a multi-gene panel test, showed HER2 amplification with coexisting extracellular domain HER2 E401G variant (Fig. 1A, B). Because the patient met the JUPITER trial eligibility criteria of HER2 amplification detected by an NGS-based assay, she was enrolled into the trial and received the combination of trastuzumab and pertuzumab. The best overall response to treatment according to RESIST version 1.1 [21] was “Stable Disease” with a certain degree of tumor control but not sufficient tumor volume reduction (Fig. 1C). Eight months later, skin metastasis was observed and diagnosed by skin biopsy, which pathological findings similar to those at initial diagnosis (Fig. 1D, E); treatment in the trial was therefore discontinued.

As an ancillary part of the trial, liquid biopsy using Gardant360®, a multi-gene panel test, was performed at the time of enrollment and at the end of treatment in clinical trials. In this test, a copy number (CN) of 2.2 or higher is considered to indicate a clinically significant increase in the CN of the gene [22]. The results in this patient showed a HER2 CN of 2.25 and HER2 E401G variant allele fraction (VAF) of 6.78% at the time of enrollment. At the end of treatment in the clinical trial, CN was 2.43 and E401G VAF was 10.9%, with a slightly increasing trend reflecting disease progression (Fig. 1A). Based on the Ethics Committee review and the patient’s written informed consent, we initiated investigation to determine the next treatment regimen, by using PDX and other approaches, utilizing a portion of the skin biopsy tissue.

We used histological and genetic approaches to evaluate whether the PDX models retained the characteristics of the original tumor. Evaluation of drug efficacy in PDX was performed after at least three in vivo passages by transplantation into BRJ mice, and the morphology and IHC profiles of the tumors excised during passages (PDX G1, G2, G3 and G4) were similar to those of the original tumor (Fig. 2A). In addition, the HER2 CN did not change during in vivo passages, and E401G remained more frequent than wild type in the amplified HER2 gene (Fig. 2B). In addition, CTOS model preparation was performed concurrently with PDX establishment for evaluation in a distinct experimental system. After the initial CTOS preparation process, we used CTOS obtained after two in vivo passages in BRJ mice for drug efficacy evaluation. As was the case with PDX, tumors (CTOS G1 and G2) during passages in the CTOS preparation process retained the histological and genetic characteristics of the tumors (Fig. 3A, B).

Initially, we evaluated the efficacy of the drugs in the CTOS model for cancers with the HER2 E401G and its amplification. Afatinib, a TKI with an inhibitory effect on EGFR in addition to well as on HER2, was the most effective in suppressing tumor growth, with an IC₅₀ of 0.35 μM (Fig. 3C). On the other hand, lapatinib, an anti-HER2 TKI with a weaker inhibitory effect on EGFR than afatinib, was less effective in inhibiting tumor growth (IC₅₀: 1.8 μM) (Fig. 3C). The effect of trastuzumab plus pertuzumab (T + P), for which there is clinical evidence of efficacy against cancers with wild-type HER2 amplification, was very weak (Fig. 3C). Because both trastuzumab and pertuzumab are anti-HER2 monoclonal antibody drugs, antibody-dependent cellular cytotoxicity (ADCC) and complement-dependent cytotoxicity in vivo are also involved in their efficacy. The effect of T + P was considered necessary to validate these drugs in the in vivo model.

Next, we evaluated the efficacy of the same drugs in the in vivo PDX models. Afatinib was the most effective among all treatments, with a statistically significant inhibition of tumor growth on day 22 after initiation of drug treatment (Fig. 4A, B). T + P tended to suppress tumor growth, but no statistically significant difference was observed. Lapatinib showed less inhibition of tumor growth in both the 100 mg/kg and 150 mg/kg dose settings. In addition, HER2 copy number in tumors in each mouse on day 22 after anti-HER2 therapy decreased was significantly lower in the afatinib-treated group than in the vehicle-treated group, but was not significantly lower with the other treatments (Fig. 4C). To compare the difference in efficacy with and without E401G in coexistence with HER2 amplification, we evaluated the drug efficacy on H2170 cells, a lung cancer cell line with wild-type HER2 amplification. The results showed that T + P was profoundly effective against H2170 cells in vivo and provided the greatest effect of all drugs (Fig. 5A, B). This finding was contrary to the results in the cancer model with coexisting E401G mutation. Afatinib also produced statistically significant tumor shrinkage in H2170, and its effect was superior to that of lapatinib (Fig. 5A, B). The trend was similar in vitro, although efficacy of T + P were relatively weak (Fig. 5C).