Ex vivo drug testing of patient-derived lung organoids to predict treatment responses for personalized medicine

Abstract

Lung cancer is the leading cause of global cancer-related mortality resulting in ~1.8 million deaths annually. Systemic, molecular targeted, and immune therapies have provided significant improvements of survival outcomes for patients. However, drug resistance usually arises and there is an urgent need for novel therapy screening and personalized medicine. 3D patient-derived organoid (PDO) models have emerged as a more effective and efficient alternative for ex vivo drug screening than 2D cell culture and patient-derived xenograft (PDX) models. In this review, we performed an extensive search of lung cancer PDO-based ex vivo drug screening studies. Lung cancer PDOs were successfully established from fresh or bio-banked sections and/or biopsies, pleural effusions and PDX mouse models. PDOs were subject to ex vivo drug screening with chemotherapy, targeted therapy and/or immunotherapy. PDOs consistently recapitulated the genomic alterations and drug sensitivity of primary tumors. Although sample sizes of the previous studies were limited and some technical challenges remain, PDOs showed great promise in the screening of novel therapy drugs. With the technical advances of high throughput, tumor-on-chip, and combined microenvironment, the drug screening process using PDOs will enhance precision care of lung cancer patients.

Introduction

Lung cancer remains the leading cause of cancer-related mortality worldwide accounting for approximately 1.8 million deaths in 2020[1]. Non-small cell lung cancer (NSCLC) is the predominant histologic subtype, accounting for 85% of all lung cancer cases in the United States[2]. In comparison, small cell lung cancer (SCLC) represents 15% of lung cancer cases, occurs almost exclusively in smokers and has the most aggressive clinical course with survival outcomes of 2 to 4 months in untreated patients[3, 4]. Immune checkpoint inhibitors (ICI) have revolutionized treatment paradigms and have been broadly integrated into the first line setting of non-oncogene driven NSCLC and small cell lung cancer therapy in combination with chemotherapy. ICI monotherapy has been shown to be effective in selected patients with NSCLC with high PD-L1 expression (>50%) or high tumor mutational burden (TMB >10). With the advent of molecular targeted therapies, there have also been significant improvements in survival outcomes for patients with oncogene driven NSCLC[5, 6]. Although these systemic therapies have durable responses, drug resistance often occurs with long-term use of these therapies due to tumor intrinsic factors and complex tumor immune microenvironments that sustain tumor growth and metastasis[7-9]. For these reasons, there is an urgent need to identify novel therapies in patients with treatment-refractory lung cancer of different molecular subtypes who fail standard systemic treatments.

An advanced understanding of lung tumor biology has led to novel therapeutic strategies that target other signal transduction or angiogenesis pathways, as well as leverage the immune system in favor of an anti-tumor response[8, 10-13]. Historically, cancer cell lines have been used to test novel anti-cancer therapies, but they do not accurately predict treatment responses due, in part, to interpatient variability and molecular tumor heterogeneity[14-17]. Patient-derived xenograft (PDX) mouse models can predict better treatment responses but they are costly and difficulties in high-throughput expansion have confined their application to preclinical drug screening[16, 18]. Patient derived organoids (PDOs) are human tissue derived tumors that can overcome some of these challenges[19-21]. PDOs (3-D cultures of cancer cells) can be established from different lung tumor specimens collected from lobectomies, wedge resections, , core needle biopsies, body fluids (ascites, pleural fluid, cerebral spinal fluid), peripheral blood (circulating tumor cells) [22-29]. Lung organoids from tumor tissue from tumor resections can best capture the spatial architecture and molecular signatures [30-35]. Lung organoids grow exponentially from single stem cells embedded in matrigels/culture mediums and display unique morphological patterns[30, 36, 37]. PDO models can demonstrate test drug efficacy of anti-neoplastic drugs based on morphological changes, tumor volume reduction or reduction in cell viability[38]. Thus, PDO models can serve as a surrogate tumor for genomic and transcriptomic profiling, biomarker identification, genetic manipulation and high-throughput drug sensitivity screening. Moreover, they are suitable for long-term expansion or cryopreservation as living organoid biobanks [39-41]. An important clinical aspect of PDOs is the potential for genomic and transcriptomic profiling and therapeutic screening for precision medicine[42-46] (Figure 1). In this way, PDOs of lung cancer can further our understanding of lung cancer pathophysiology and predict drug efficacy for individual patients in the clinic. Here, we conduct a comprehensive literature review to explore the research to date surrounding ex vivo drug testing using organoids derived from lung cancer patients. We discuss the translational work, challenges and future applications of this most promising technology.

Figure 1.

The illustration of applications of patient-derived lung cancer organoids for precision medicine. Created with BioRender.com

Methods

This is a comprehensive review of the published literature focusing on ex vivo drug testing using organoids from lung cancer patients. PubMed, Cochrane Library and Scopus databases were searched. All full-text, peer-reviewed publications published through March 2023, were included for consideration. The search was restricted to full-text, peer-reviewed articles and reports published in English. Abstract-only publications were excluded, as were any papers that failed to fully elucidate outcomes (i.e., editorials). The search strategy for this review was the result of prior research in the fields. The search terms included (but were not limited to): “patient derived organoids”, “human derived organoids”, and “drug screening”. Further, appropriate MeSH terms and subject headings were also employed. A full search strategy was developed for Ovid Medline and is shown in Table 1.

Table 1:

Literature search strategy in Database(s): Ovid MEDLINE(R) ALL from 1946 to March 23, 2023

#	Searches	Results
1	(((patient-derived adj3 organoid*) or patient derived) adj3 organoid*).ti,ab.	1187
2	exp Organoids/de [Drug Effects]	1031
3	(((human-derived adj3 organoid*) or human derived) adj3 organoid*).ti,ab.	35
4	1 or 3	1218
5	2 and 4	96
6	drug screening.mp. or exp Drug Evaluation, Preclinical/	297911
7	4 and 6	269
8	5 or 7	312

Results

Organoid technology for high-throughput drug screening

Historically, cancer cell lines have been widely used as preclinical models to evaluate the efficacy of antineoplastic drugs. Despite its wide application, the sensitivity and specificity of drug candidates in 2D-cell culture have been widely variable. Lung cancer organoid technology was developed for high-throughput drug screening to better predict clinical responses to drug treatment. In one study, Li et al. created a living biobank of frozen NSCLC PDOs (n=10) and used high-throughput drug screening assay to demonstrate the anti-tumor effects of natural compounds (chelerythrine chloride, cantharidin and harmine)[47]. In a second study, Chen et al. evaluated drug sensitivity of 26 anti-cancer therapeutic agents in organoids derived from 12 patients with locally advanced NSCLC, including tumors that harbored EGFR and KRAS G12C mutations[45]. They found that PDOs retained the histological and genetic characteristics of the primary tumors with more than 80% concordance[45]. Furthermore, they showed that PDOs correlated with molecular profiles of primary lung tumors with mutations involving the top 20 NSCLC-related genes[45]. In a separate study, Li et al. tested 24 antineoplastic drugs in PDOs derived from 12 patients with NSCLC using a high throughput system and found drug efficacy correlated with mutational profiles[48]. Tamura et al. demonstrated the feasibility of using a high throughput system for PDO drug screening from a large compound library comprised of 61 anti-cancer agents that included targeted therapies and chemotherapy[49]. Takahashi et al. used their high-throughput assay to evaluate 86 anti-cancer agents (targeted agents, immune checkpoint inhibitors and cytotoxic chemotherapies) in PDOs developed from lung tumors with squamous and adenosquamous histologies[50]. In all these studies, the pharmacogenomic profiles of PDOs were highly comparable to that of tumor tissue indicating that organoid models can predict therapeutic responses on a larger scale to inform therapeutic strategies[17, 51, 52].

PDOs in Preclinical Studies

In addition to large-scale screening, PDOs are being employed for personalized drug screening for patients with treatment resistant lung cancer. Jung et al. observed that PDOs derived from SCLC display an inherent drug resistant phenotype to conventional chemotherapy with cisplatin and etoposide[53]. Gmeiner et al. found that dysregulated pyrimidine biosynthesis contributes to drug resistance in SCLC and they demonstrated that the novel fluoropyrimidine polymer (CF10) could overcome drug resistance to conventional chemotherapy using a PDO model system[54]. In another study, PDOs derived from a NSCLC patient demonstrated the efficacy of an oral potent tubulin destabilizing agent, S-40, in overcoming paclitaxel resistance[55]. These study models highlight the utilization of organoids for assessing drug resistant phenotypes which can be directly translated to the clinic for individual lung cancer patients who develop tumor progression on standard chemotherapy.

The identification of oncogenic activation of tyrosine kinases in NSCLC tumors, most notably epidermal growth factor receptor (EGFR) mutations or anaplastic lymphoma kinase (ALK) gene or c-ROS oncogene 1 (ROS1) gene rearrangements, has led to the development of targeted agents. Matching a specific targeted drug to the identified driver mutation for an individual patient has resulted in significantly improved survival outcomes, compared with chemotherapy and/or immunotherapy [51, 56]. Despite dramatic tumor shrinkage initially seen in NSCLC patients with driver mutations, drug resistance to targeted therapy often develops. PDOs are currently being used to test small molecular inhibitors with the potential to overcome this secondary resistance. One study conducted by Saraon et al. evaluated the efficacy of EMI66 and other small molecules that inhibit mutant EGFR signaling and alters ER stress response pathway[57]. They found that EMI66 and other derivatives reduced viability of organoids derived from three PDOs with EGFR mutations, suggesting a bypass mechanism to target EGFR driven tumor growth[57]. Hu et al. found that PDOs with the EML4-ALK rearrangement mutation and EGFR activation mutations showed reduced viability and sensitivity to TKI inhibition, compared to wild-type[58]. They demonstrated that lung organoids with TKI sensitive L858R mutation and absent T790 mutation could develop drug resistance early, highlighting the potential role of PDOs to predict drug resistance better than the T790M molecular marker[58]. These collective studies suggests that lung organoids can be used as a treatment prediction model to allow clinicians to prioritize targeted by comparing the responses of the individual PDOs to different drug combinations in real-time. These previous studies of ex vivo drug testing based on patient-derived lung cancer organoids are summarized in Table 2 and organoids culture medium are summarized in Table 3.

Table 2.

Summary of organoids as an enhanced model for personalized drug screening

Cancer type	Organoid Model	PDX Model (n)	Compounds Tested	Technology/ Unique culture supplement	Limitations	Ref
Adenocarcinoma	Primary lung tumor spheroids (n=14) were cultured in 3D-Matrigel culture methods	N/A	Erlotinib	Stem Pro hESC Supplement	The number of different drug and dose combinations that can be investigated at one time is limited.	[74]
Adenocarcinoma (NSCLC patients with tumors stage I–III, EGFR L858R, EGFR Ex20 ins, KRAS G12C)	1) Fresh tumor samples harvested for organoid culture. 2) Primary tumor samples and PDOs were analyzed via whole-exome sequencing and IHC. (n=7)	N/A	26 antineoplastic drugs tested (gefitinib, osimertinib, afatinib)		Small cohort sized used and will require further large-scale analyses to validate the findings.	[45]
Small cell lung cancer – refractory (tumors stage I–III, EGFR L858R, EGFR Ex20 ins, KRAS G12C)	PDOs (n=4) were developed from human SCLC PDX samples to test if TS inhibition could be a viable strategy for SCLC treatment.	PDXs were generated from SCLC tumor biopsy samples.	Cisplatin Thymidylate synthase inhibitors: CF10, 5-FU	HyStem-HP hydrogel kits		[54]
1)Adenocarcinoma 2)Squamous cell carcinoma 3)Adenosquamous carcinoma 4)Large cell carcinoma 5)Small cell lung cancer	Surgically resected lung cancer tissues from 36 patients were embedded in Matrigel and submerged in MBM to create PDOs (n=80).	PDXs were generated from tissue samples. PDXs from 10 samples (43%)	Docetaxel Olaparib Erlotinib Crizotinib		Limitation of cancer organoid models is the lack of a cancer microenvironment.	[17]
NSCLC	PDOs derived from NSCLC were cultured in vitro (n=10)	N/A	Chelerythrine chloride Cantharidin Harmine Berberine Betaine			[47]
Adenocarcinoma	Developed PDOs from human lung adenocarcinoma biopsy samples (n=12)	N/A	24 antineoplastic drugs tested		Small sample size limits the power to detect molecular markers of drug response.	[48]
Small cell lung cancer (n=1)	Tumor organoids were generated from primary lung cancer cells from patients with SCLC.	N/A	Cisplatin Etoposide	Microfluidic-based lung cancer organoid culture platform for testing drug sensitivity. Patient-derived lung cancer organoids (PDOs) were cultured and expanded in Matrigel droplets in 24-well plates.	Small sample sized used and will require further large-scale analyses to validate the findings.	[53]
Adenocarcinoma Metastatic lung cancer	Developed PDOs from pleural effusion of patients with lung adenocarcinoma (n=2)	N/A	Cisplatin + pemetrexed; carboplatin + pemetrexed; crizotinib	3D hydrogel-based model	Small sample size. Will require further large-scale analyses to validate the findings.	[82]
Non-small cell lung cancer	PDO (n=1)	N/A	S-40 (oral potent tubulin destabilizing agent)	No detail description of organoid culture	Small sample size	[55]
1)Squamous cell carcinoma 2)Adenosquamous carcinoma	PDOs developed from human tumor lung cancer surgical specimens, (n=3)	N/A	86 antineoplastic drugs tested, including molecular targeted drugs, immune checkpoint inhibitors, and cytotoxic chemotherapy	High throughput 96- and 384-well screening Organoids were cultured in FBIM001 medium. Organoids culture condition as previous describe [49]	Few and limited PDO examples were selected for each type drug screen.	[50]
1)Lung adenocarcinoma 2)Squamous cell	PDOs developed from 19 surgically resected lung adenocarcinomas (LUAD) and 15 lung squamous cell carcinomas (LUSC),	LPTO and PDXO were used to denote organoid models derived from lung primary patient tumor and PDX, respectively. 16 LUAD PDXs, and 26 LUSC PDXs were processed for organoid establishment.	Study evaluated the efficacy of clinically approved EGFR-targeted therapy in NSCLC in four short- term organoid models. BGJ398 (FGFR inhibitor) Trametinib (MEK inhibitor) Selumetinib (MEK inhibitor) Afatinib (EGFR inhibitor)			[69]
Lung adenocarcinoma	Three primary lung cancer organoid models: XDO-137 (EGFR ex19del) PDXO-4000 (EGFR ex19del) XDO-344 (wild-type EGFR)	N/A	EMI66 and other derivatives.	Organoids culture condition previously described [69]		[57]
1) Squamous cell carcinoma 2) Adenocarcinoma 3) small cell lung cancer	PDOs (n=21) generated from 16 adenocarcinoma and 4 squamous cell carcinoma, two small cell lung cancer organoids	On-chip drug responses of the PDX derived organoids (n=3) were consistent with the in vivo PDX results	Pemetrexed-cisplatin Gemcitabine+cisplatin Docetaxel+cisplatin Paclitaxel-cisplatin Afatinib Erlotinib Gefitinib Osimertinib Crizotinib	Organoids were cultured with Matrigel in integrated superhydrophobic microwell array chip (InSMAR-chip) Perform drug testing within a week.	Well-designed pilot study would improve sensitivity and specificity of the assay.	[58]
Lung adenocarcinoma.	Organoids (n=84) were established from patients with advanced lung adenocarcinoma.		Six EGFR inhibitors Two EGFR/Her2 inhibitors Vandetanib Lenvatinib Cabozantinib Pralsetinib Poziotinib	Organoids culture medium previously described [91]	This was a retrospective study.	[51]
Lung adenocarcinoma	PDOs (n=5) were developed from fresh lung tumors obtained from treatment naïve patients.	N/A	Dubermatinib (TP-0903) Ruxolitinib TP0903 + ruxolitinib	Single cell CyTOF analysis of tumor tissues	Not for high throughput screening with a higher cost	[76]
Adenocarcinoma HER2 mutant	PDO (n=1) was developed from human lung tumor specimens,	Human lung tumor fragments were subcutaneously implanted in mice for generating PDX	Pyrotinib Afatinib T-DM1		One PDO and small size of clinical trial patient cohort	[44]
NSCLC	PDOs (n=2) were developed from malignant pleural effusions of patients with NSCLC with Exon20ins mutations.	Human lung tissue specimens were implanted subcutaneously in mice	Amivantamab	Organoids culture medium previously described [91]		[72]
EGFR-Mutant NSCLC	Patient surgical resection or tumor biopsy specimens (n=3).	N/A	Osimertinib	Organoids culture medium previously described [91]	The presence of the driver oncogene and secondary mutations are needed. no tumor microenvironment included	[92]

Table 3.

Summary of organoids culture medium

Ref	Base medium	Matrigel	N2	B27	EGF (ng/mL)	FGF-10 (ng/mL)	Y-27632 (μM)	GlutaMax (L-glutamine alternative)	R-spondin 1 (ng/mL)	Noggin (ng/mL)	Nicotinaminde (mM)	SB202190 (μM)	N- acetylcysteine (mM)	A83-01	Specific components
[76]	Advanced DMEM/F12	Yes	Yes	Yes	50	10	10		500	100	4	5		500 nM	20 ng/ml HGF, 1 μM Prostaglandin E2, 20 ng/mL bFGF
[47]	Advanced DMEM/F12	Yes	Yes	Yes	50	20			250	100	10	10	1	500 nM	2mM L-glutamine, 100 ng/ml Wnt3a, 10 nM gastrin 1, 0.01% BSA, 1 μM Prostaglandin E2, 1ng/mL bFGF
[48]	Advanced DMEM/F12	Yes	Yes	Yes		20	10 mM	Yes	500	100	10	10 mM	1.25	500 nM	25 ng/mL FGF-7
[44]	Advanced DMEM/F12	Yes		Yes	50	10		Yes	Yes*	Yes*	10	10	1.25	500 nM	R-spondin and Noggin form condition medium, 1 ng/ml FGF2, Dihydrotestosterone (DHT), 10mM HEPES
[51],[72],[92]	Advanced DMEM/F12	Yes		Yes		100	5	Yes	500	100	5	0.5	1.25	500 nM	25 ng/mL FGF-7, 10mM HEPES
[45]	DMEM/F12	Yes	Yes	Yes	50		10
[17], [53]	DMEM/F12	Yes	Yes	Yes	50		10								20 ng/mL bFGF
[58]	DMEM/F12	Yes	Yes	Yes	50		10	Yes			5	3	1	5 μM	10 μM Forskolin, 3 nM Dexamethasone, 10mM HEPES
[74]	DMEM/F12	Yes					10	Yes							Stem Pro hESC Supplement, 0.1mM 2-mercaptoethanol, 25% BSA, bFGF
[69], [57]	RPMI 1640	Yes	Yes	Yes	50	100	10	Yes		100				500 nM	250 nM CHIR 99021, 100 nM SAG, 100 ng/mL FGF4, 10mM HEPES
[82]	RPMI 1640														Hydrogel cultures, 5% fetal bovine serum
[54]		N/A													HyStem-HP hydrogel kits
[50]															FBIM001 medium [49]
[55]	No detailed description

PDOs in clinical studies - an enhanced model for personalized drug screening

In a large clinical study, 84 organoids were established from patients with oncogene-driven lung adenocarcinoma and found that PDOs could predict treatment in patients receiving clinically approved targeted agents (EGFR inhibitors, ERBB2 targeted therapies, multikinase inhibitors)[51]. They found that combination treatment with dabrafenib and trametinib demonstrated favorable ex vivo and clinical responses in a NSCLC patient harboring both an EGFR exon 19 deletion and a BRAF G464A mutation[51]. They also demonstrated pre-clinical and clinical efficacy of afatinib against PDOs that harbored the rare EGFR L747P mutation. Among ERBB2 targeted therapies (erlotinib, lapatinib, neratinib, and afatinib), poziotinib demonstrated the greatest potency in lung organoids with ERBB2 exon 20 insertions[51]. The preclinical studies mirrored the phase 2 study (ZENITH20-2 Trial) of poziotinib in patients with advanced or metastatic NSCLC (n=90 patients)[59]. With a median follow-up of 9.0 months, this study yielded an overall response rate (ORR) of 28%, median progression free survival (PFS) of 5.5 months, median duration of response (DOR) of 5.1 months[59]. Most importantly, clinical benefit was seen regardless of prior lines of therapy, presence of brain metastasis and type of HER2 mutation[59]. Both the preclinical and clinical studies indicated robust and durable anti-tumor activity of poziotinib in NSCLC patients with ERBB2 exon 20 insertions[51]. Among the multi-kinase inhibitors (vandetanib, cabozantinib, and lenvatinib), pralsetinib demonstrated clinical efficacy in RET-rearranged lung tumors ex vivo and in vivo[51]. In a recent phase I/II clinical trial (ARROW trial), robust treatment responses were observed in patients with RET fusion-positive NSCLC (n=233) and similar drug efficacy were recapitulated in respective PDOs[60]. In 53 of 87 NSCLC patients previously treated with platinum-based chemotherapy, there was a 61% ORR to therapy including five patients with a complete response[60]. In 19 of 27 treatment naïve NSCLC, a 70% overall response was observed with a complete response in three patients[60]. Overall, these studies highlight the potential of organoid-based drug sensitivity testing to predict clinical outcomes better than molecular markers. This accuracy in treatment prediction offers clinicians the ability to prioritize targeted and chemotherapies prospectively by comparing the individual responses of the generated PDOs to different drugs.

About 2%–3% human epidermal growth factor receptor 2 (HER2, ERBB2) mutation have been identified as one of tumorigenic drivers and observed in of NSCLC[61]. Several HER2-targeted agents (afatinib, dacomitinib, neratinib and trastuzumab) have not shown long term responses in patients with HER2-mutant NSCLC[62-65]. Wang et al. recently studied the anti-tumor effects of pyrotinib, a pan-HER inhibitor, in NSCLC with HER2 exon 20 mutations in both preclinical and clinical models [44]. Pyrotinib showed significant growth inhibition of organoids relative to afatinib[44]. In PDX model, the pan-HER inhibitor reduced tumor volumes showed a superior antitumor effect compared with afatinib and trastuzumab emtansine (humanized monoclonal antibody trastuzumab covalently linked to cytotoxic agent DM1)[44]. Specifically, mice treated with pyrotinib had >50% reduction in tumor volumes[44]. This anti-tumor drug effect was attributed to the inhibition of HER2 and its downstream signaling pathways, including ERK and Akt. Preclinical findings were validated in a phase 2 study of pyrotinib (400 mg oral daily dose) in patients HER2-amplified NSCLC (n=27)[66]. In this single arm study, pyrotinib demonstrated a median Progression free survival (PFS) and overall survival (OS) reported as 6.3 months and 12.5 months, respectively[66]. Furthermore, patients administered pyrotinib as first-line treatment achieved a median PFS of 12.4 months[66]. Approximately 30% of the patients who had progressed on EGFR tyrosine kinase inhibitor responded to pyrotinib. Strikingly, patients with brain metastases had an overall response rate (ORR) of 40%. Collectively, the preclinical and clinical study demonstrates the applicability of lung organoid models to study patient populations with rare mutations.

The fibroblast growth factor/fibroblast growth factor receptor (FGF/FGFR) is a tyrosine kinase signaling pathway that plays a critical role in oncogenesis (tumor proliferation, angiogenesis, migration, and survival) via gene amplification, activating mutations, or gene translocation in lung tumors[67, 68]. In this way, the FGFR signaling pathway represents an important target for lung cancer. In vitro studies suggest that combination of MEK and PI3K inhibitors with FGFR inhibitors may be effective in FGFR-aberrant cancers[69]. To test this hypothesis in preclinical models, Shi et al. developed organoids from surgically resected lung tumors derived from 19 patients with lung adenocarcinoma and 15 patients with squamous cell lung carcinoma[69]. Additionally, they processed 16 lung adenocarcinoma PDX and 26 squamous cell lung cancer PDX tumors for organoid establishment[69]. They tested trametinib (MEK inhibitor) and the PI3K inhibitor BKM120 with BGJ398 (inhibits pFGFR and pAkt) in their FGFR1-amplified organoid models[69]. Strong synergy was observed in the BGJ398 + trametinib combination (combination index < 0.5), whereas weaker synergy (combination index >0.5) was observed in the BGJ398 + BKM120 (PI3K inhibitor) combination[69]. Furthermore, PDOs retained the histologic and molecular features of their parental tumors[69]. The spectrum of mutations was highly concordant between the organoid and matched patient tumor and PDX tissue[69]. Mutational burden in the five long-term established PDOs were similar in parental tumor and PDX tumors[69]. This indicates that organoid culture conditions did not destabilize the cancer genome. Copy number variation (CNV) profiles of the parental tumors were largely preserved during PDO culture[69]. There have been early phase clinical studies in patients with lung cancer (and other solid tumors) evaluating FGFR inhibitors as monotherapy or in combination with existing therapies. As more clinical trials of both selective and non-selective FGFR inhibitors emerge[70], patient selection as it pertains to predicting response to therapy should be undertaken with patient derived organoids given the accuracy of this model system.

Insertion mutations in exon 20 of the epidermal growth factor receptor (EGFR) gene are the largest class of EGFR mutations in NSCLC for which there are no FDA approved targeted therapies[71]. Yun et al. tested amivantamab (an EGFR-MET bispecific antibody) in tumors derived from organoids and xenograft models harboring diverse Exon 20 insertion mutations and found that the drug showed anti-tumor activity and suppressed EGFR and MET signaling pathways[72]. Furthermore, they established their PDOs from malignant pleural effusions collected from two patients with NSCLC with Exon 20 insertion mutations[72]. In the Phase I (CHRYSALIS) study, amivantamab produced robust and durable tumor responses in patients with EGFR Exon 20 insertion NSCLC (n = 81) who progressed on platinum chemotherapy. The ORR was 40% (including three complete responses) with a median DOR of 11.1 months and median PFS was 8.3 months[73]. This study highlights the importance of ex vivo drug testing to inform clinical trial design[72]. Clinical trials have been summarized in Table 4.

Table 4.

Clinical trials related to organoid studies

NCT	Study description	Drug mechanism of action	Results	Adverse events	Conclusion	Ref
NCT03318939	ZENITH20, a multicenter, multicohort, open-label phase II study, evaluated poziotinib in patients with advanced or metastatic NSCLC. In cohort 2, patients received poziotinib (16 mg) po once daily.	Poziotinib is an irreversible pan Her2 inhibitor with activity agains HER1, Her2, and Her4. Pozitinib has been shown to inhibit kinase activity of EGFR with exon 20 insertion mutations.	ORR was 27.8% (95% CI, 18.9 to 38.2) DCR was 70.0% (95% CI, 59.4 to 79.2) Median PFS was 5.5 months (95% CI, 3.9 to 5.8) Median DOR was 5.1 months (95% CI, 4.2 to 5.5)	TRAE (grade 3) Rash (48.9%) Diarrhea (25.6%) Stomatitis (24.4%)	Poziotinib shows antitumor activity in patients with HER2 exon 20 insertion NSCLC.	[51, 59]
NCT03037385	ARROW is a multi-cohort, open-label, phase 1/2 study done at 71 sites in 13 countries. Patients aged 18 years or older with locally advanced or metastatic solid tumours, including RET fusion-positive NSCLC. In phase 2, patients received pralsetinib 400 mg po once daily.	Pralsetinib is highly-selective RET Inhibitor	ORR was 61% (95% CI, 50–71)	TRAE (grade 3, 48%) Neutropenia (18%) Hypertension (11%) Anaemia (10%)	Pralsetinib is a well-tolerated, promising, once-daily oral treatment option for patients with RET fusion-positive NSCLC.	[51, 60]
ChiCTR1800020262	A prospective, multicenter, single-arm trial, patients with advanced NSCLC with HER2 amplification. Patients administered pyrotinib 400 mg po daily.	Oral irreversible pan-HER receptor tyrosine kinase inhibitor that target HER1, HER2, and HER4	6-month PFS rate was 51.9% (95% CI, 34.0-69.3) Median PFS was 6.3 months (95% CI, 3.0-9.6 months) Median OS was 12.5 months (95% CI, 8.2-16.8 months) ORR of 22.2% (95% CI, 10.6%-40.8%)	TRAE (grade 3, 22.2%) Diarrhea (7.4%) Vomiting (7.4%) Alanine aminotransferase (ALT) increase (3.7%) Rash (3.7%)	Pyrotinib demanstrates antitumor efficacy in HER2-amplified NSCLC patients with a manageable safety profile.	[44, 66]
NCT02609776	CHRYSALIS is a phase I, open-label, dose-escalation, and dose-expansion study, which included a population with EGFR Exon20ins NSCLC. Recommended phase II dose of 1,050 mg amivantamab (1,400 mg, ≥ 80 kg) given once weekly for the first 4 weeks and then once every 2 weeks starting at week 5.	EGFR-MET bispecific antibody	ORR was 40% (95% CI, 29 to 51) Median DOR of 11.1 months (95% CI, 6.9 to not reached) Median PFS was 8.3 months (95% CI, 6.5 to 10.9).	TRAE (grade 3) Hypokalemia (5%) Rash (4%) pulmonary embolism (4%) Diarrhea (4%) Neutropenia (4%)	Amivantamab yielded robust and durable responses with tolerable safety in patients with EGFR Exon20ins mutations after progression on platinum-based chemotherapy.	[72, 73]

^*Non-small cell lung cancer (NSCLC), overall response rate (ORR), disease control rate (DCR), progression-free survival (PFS), overall survival (OS), duration of response (DOR), Rearranged during transfection (RET) gene, treatment-related adverse events (TRAE)

Mechanistic studies involving PDOs

PDOs also serve as a valuable model to study tumor evolution and the acquisition of secondary mutations associated with treatment resistance. Banda et al., treated lung tumor organoids with erlotinib, a first-generation EGFR-tyrosine kinase inhibitor[74]. Upon subsequent passaging, the organoid cultures developed additional mutations in BRAF, EGFR, KRAS and PIK3CA genes which have been commonly reported in patients who develop resistance to EGFR inhibitors[74]. These observations were supported by a comprehensive mutation profiling analysis (multiplex testing) of tumor specimens collected from 153 NSCLC patients identified the co-existence of EGFR with KRAS (n=29), or BRAF (n=2) or PIK3CA somatic mutations (n=58)[75].

Therapeutic targeting of tumor subpopulations based on their single cell expression of oncogenic markers can lead to robust treatment responses in PDO and PDX mouse models. Taverna et al. used mass cytometry by time-of-flight (CyTOF) to identify continuous AXL and JAK/STAT3 signal activation in lung adenocarcinoma tumors derived from 11 treatment naïve patients[76]. Using single cell profiling, they could stratify tumor subpopulations based on AXL and STAT3 signaling[76]. They found that tumor subpopulations with high AXL/STAT3 expression displayed hybrid epithelial-to-mesenchymal phenotype and cancer stemness, suggesting high tumorigenic and metastatic potential[76]. They subsequently demonstrated that PDO with high AXL/STAT3 expression responded robustly to combination treatment with dubermatinib (AXL inhibitor) and ruxolitnib (JAK inhibitor) as compared with single agents[76]. Conversely, PDOs with low AXL/STAT3 expression did not respond to combination treatment and/or single agents[76]. In a separate study, they created PDX mouse models derived from PDOs expressing high AXL and JAK/STAT3. The PDX mouse models with high AXL/STAT3 expression demonstrated reduced tumor volumes following combination treatment with AXL and JAK inhibitors, when compared to single agent[76]. These studies highlight that PDOs can capture the oncogenic signaling of primary lung tumor cell populations and could help determine treatment sensitivity to targeted therapies (bench to bedside).

Another major strength for PDOs is that they can be subjected to genetic manipulation, which can be a powerful tool for understanding oncogenic signaling in lung cancer. For example, Dost et al. used organoids model to study transcriptional hallmarks of oncogenic KRAS activation in lung epithelial progenitor cells[77]. They developed organoid systems from primary mouse and human induced pluripotent stem cell-derived lung epithelial cells to model early-stage lung adenocarcinoma, showing organoid approaches can be utilized for uncovering the early consequences of oncogenic KRAS expression.

Conclusions and future directions

Lung cancer PDOs have emerged as an effective alternative model for ex vivo drug screening. PDOs have become widely used for testing anti-neoplastic agents for therapeutic efficacy in clinical trials. The ability to generate organoid models from individual patients enrolled in clinical trials allows investigation of patient-specific responses to therapeutic drugs in real-time[51]. One research team established a PDO biobank from patients with metastatic gastrointestinal cancer who were actively undergoing treatment in clinical trials and demonstrated that the ex vivo responses in organoid cultures closely mirrored clinical responses with 100% sensitivity, 93% specificity, 100% negative prediction accuracy and 88% positive prediction accuracy[78]. This study provides a strong rationale for designing co-clinical trials with organoid models to better predict drug responses. Organoids have also been adopted to predict treatment response to radiotherapy and immunotherapy[79-81]. By simulating cancer behavior ex vivo, organoid technology can integrate molecular biology with treatment decisions to improve patient selection for particular therapeutic agents.

Based on our review of the literature, we found that organoid models can be established from pleural effusions which is scarce source of lung cancer cells[72, 82, 83]. Strikingly, organoids derived from pleural fluid aspirate can developed tumor-specific cellular and molecular characteristics and preserved tumor heterogeneity[82, 83]. Since pleural effusion aspiration is fairly a non-invasive and routine procedure among lung cancer patients, generation of organoid models from these samples can allow effective disease modeling, monitor disease progression and treatment responses in real-time. Another advantage of using pleural effusion-derived cells is that they constitute both tumor cells and stromal cells such that tumor microenvironment and disease pathophysiology can be preserved and cellular interactions in response to disease progression or anti-cancer agents can be studied simultaneously[83, 84]. Co-culturing tumor organoids with other immune cells can mimic complex cellular microenvironment and tumor-intrinsic interactions [50]. For example, Takahashi et al. recently demonstrated that lung PDOs exposed to antibody-drug conjugates (trastuzumab, pertuzumab and trastuzumab emtansine) can effectively model complex interactions with THP-1 effector cells and reflect the antibody-dependent cellular cytotoxicity (ADCC) response of the drug[50]. The study also expanded on the use of anti-PD1 monoclonal antibodies to evaluate efficacy of immune checkpoint inhibitors.

Although organoid culture systems have several advantages over traditional 2D cell-based models, there are certain challenges that need to be overcome. First, establishing organoids from lung cancer is a demanding endeavor. The overall organoid establishment rate from lung cancer patients is limited and variable in different lung cancer types[85, 86]. Moreover, the outgrowth of normal airway organoids could replace tumor organoids after long-term culture[41, 85]. The variable successful rates and the overgrowth of normal airway organoids would impede the implementation of organoids for precision medicine and personalized drug screening in clinical trials. The organoid media, culture procedures, and tumor sources could be the potential factors leading to these challenges. Therefore, characterization of organoids via immunohistochemistry or immunofluorescence to evaluate the malignancy of organoids would be necessary before ex vivo drug testing. Moreover, obtaining an adequate amount of tumor tissues and fine-tuning the organoid media can potentially promote tumor growth and hinder the growth of normal airway organoids[85].

Second, the most common organoid models lack stromal components and paracrine signals unique to the tumor microenvironment[87]. To overcome this limitation, microfluidic organ-on-chips have been developed to model cell-cell and cell-extracellular matrix (ECM) interactions. These novel systems are designed to recapitulate organ-level functionality including physical forces that mimic in vivo cyclic strain and fluid shear stress. Moreover, the microfluidic nature of these systems and microsensors within the microchip allows for the collection of real-time data such as barrier function and effluent collection to monitor byproducts as an indirect measure of tissue functionality[88]. Tumor-on-chip technology can mimic the mechanical and biochemical properties of the tumor microenvironment (oxygen and nutrient gradients, extracellular matrix stiffness, and cell-cell interactions). By creating a miniature model of lung tumors on a microfluidic chip, this technology allows researchers to study how cancer cells directly engage with their environment and allows for the testing of novel therapeutics mechanistically.

Tumor-on-chip platform also provides insights into the underlying mechanisms of metastasis. Metastasis is a complex process that involves multiple steps, including the invasion of tumor cells into surrounding tissue, intravasation into the blood or lymphatic vessels and colonization at the secondary site. One example is a study by Zhang et al. where they use tumor-on-chip system to study the interaction of tumor cells with endothelial cells that line blood and lymphatic vessels and demonstrated that an anti-angiogenic drug can inhibit the intravasation and extravasation of breast cancer cells[89]. In another study, researchers developed microfluidic chips to study microenvironmental conditions of metastatic tumor niches and demonstrated that hypoxic conditions at the secondary site promoted tumor growth and migration of prostate cancer cells.

In conclusion, the future development of PDO on chip technology will revolutionize drug discovery efforts and help inform clinical trials by allowing for more personalized treatment, reflecting individual patients’ drug responses in their unique tumor microenvironment. These model systems will be able to mimic multiple aspects of tumor microenvironments and cell-cell interactions between tumor cells, stromal fibroblasts, endothelial cells, immune subtypes and their individual responses (cytokine release, mechanotransduction) to immunotherapy and other systemic treatments[90].