1. Patient Case
In December, 2019, a 68-year-old female Chinese patient who never smoked but with a history of cavernous sinus hemangioma surgery presented at the Shanghai Chest Hospital. A CT scan showed a 5.2 x 4.5 x 6.2 cm mass in the left upper lobe (Figure 1A), combined with mediastinal lymph nodes, bone and brain metastasis. The upper left lung nodule was biopsied and subjected to pathological tests, next-generation sequencing (NGS) and the establishment of a xenograft at the initial diagnosis. The patient was diagnosed as NSCLC-NOS (not otherwise specified) (Figure 1B). As the patient harbored METex14 mutations, she received crizotinib treatment from January 2020. The best outcome was a partial response, the tumor size was reduced by 35.5% , which resulted in a durable response for 7 months until evidence of disease progression occurred, with a larger primary tumor and hepatic metastasis (Figure 1A). At time of disease progression, the patient again underwent a CT-guided biopsy of the left upper lung lesion for pathology tests, NGS and the establishment of a xenograft. Histopathological examination of the biopsy specimens revealed squamous cell carcinoma characteristics, which were negative for TTF-1, NapsinA and CD56, and positive for CK and P40 (Figure 1C). Due to her poor physical condition and ECOG score, the patient could not tolerate the toxicity associated with chemotherapy and received the best supportive care for 4 more months until she died in December 2020.
2. Establishment of patient-derived xenograft models for human METex14 NSCLC with and without crizotinib resistance
The subcutaneous patient-derived xenograft models were established from biopsy specimens obtained before (PDX pretreatment) or after (PDX resistant) crizotinib resistance had developed. PDXs were serially passaged in animals 3-5 times for tissue expansion. Representative tumor-bearing mice are shown in Figure 2A. No significant body mass loss was observed in mice bearing both PDX tumors.
The histology and degree of differentiation associated with PDX pretreatment was slightly different from the parent tumor histology of which led to a diagnosis of NSCLC-NOS, whereas PDX was diagnosed as squamous cell carcinoma (Figure 2B). Histology of PDX-resistant tissue matched well with its parent tumor (Figure 2C). In addition, both PDX tumors maintained intra-tumor heterogeneity, resembling the original tumors.
To evaluate the responses of PDX models to the standard-of-care agent, we conducted in vivo efficacy studies of crizotinib and tepotinib in 2 PDX models. Crizotinib and tepotinib administration all proved to be highly effective in suppressing PDX pretreatment tumor growth. The quantified TGI (PG-D21) were 115.25% and 112.31%, respectively (Figure 3A). While drug resistant PDXs showed decreased drug sensitivity to crizotinib (TGI = 65.32%, PG-D35) and tepotinib (TGI = 50.40%, PG-D35), they resembled the patient’s drug resistance to crizotinib (Figure 3B).
3. Sequencing data validated the PDX model and reported on aquired target MET mutation and EGFR gene amplification in a patient crizotinib-resistant specimen
WES and RNA-sequencing (RNA-seq) were performed on patient biopsy specimens before treatment and after resistance to crizotinib developed and on their corresponding PDX tissue. RNA-seq results confirmed that both PDX models retained METex14 (Figure 4B). DNA results revealed that both PDX tumors retained the overall pattern of the somatic mutations and copy number variation of their parent tumor tissue. Figure 4A gives a list of a total of 27 mutations, either oncogenic or with higher relevance to the disease. Most of the oncogene alterations in the parent primary tumor were preserved in the PDX specimens including METex14, MTORI2500F, RAD54LG235R, BRCA1Q1240* and TP53H193R (Figure 4A). We noticed the frequency of several mutations, including METex14 (splice site p.E1009fs), TP53H193R, ARID3AP373L and PRSS1T137M , in PDX tissue increased to 100% which may be a result of a homozygous mutation in the parent tumor.
We then compared resistant specimens to non-resistant specimens in patient and PDX models. MTORI2500F, RAD54LG235R, BRCA1Q1240*, TP53H193R and METex14 (splice site p.E1009fs) were fairly consistent between the 2 patient specimens, while mutation of KMT2CG4660E, FGFR1N546K, PBX1E60K and CSF1RL368V were lost in patient-resistant specimens. METD1228N and APCD396N newly occurred in patient-resistant specimens at mutational frequencies of 6.44% and 5.08%, respectively. However, these 2 mutations were lost in PDX-resistant tissue at both the DNA and RNA level, possibly implying clonal selection during establishment. EGFR amplification, a classic drug resistance gene alteration, was acquired in patient-resistant specimens. The copy number of EGFR significantly increased from 2.24 to 7.44 (Figure 4C). Although gene amplification was lost in PDX-resistant tissue, we observed a significant increase in the RNA expression level of EGFR in PDX specimens compared to non-resistant tissue. TPM increased from 37.6 to 102.6, which was consistent with patient specimens. Since EGFR, FGFR3, IGF1R and MET all belong to the RTK family, we further investigated the RNA expression levels in these RTK-related molecules. RNA-seq data showed increased expression levels of EGFR, FGFR3 and IGF1R in resistant specimens compared to pretreatment specimens indicating activation of the bypass signaling pathway (Figure 4C).