Immune Organoids: From Tumor Modeling to Precision Oncology

Abstract

Cancer immunotherapies, particularly immune checkpoint inhibitors, are rapidly becoming standard of care for many cancers. The ascendance of immune checkpoint inhibitor treatment and limitations in the accurate prediction of clinical response thereof have provided significant impetus to develop pre-clinical models that can guide therapeutic intervention. Traditional organoid culture methods that exclusively grow tumor epithelium as patient-derived organoids are under investigation as a personalized platform for drug discovery and for predicting clinical efficacy of chemotherapies and targeted agents. Recently, the patient-derived tumor organoid platform has evolved to contain more complex stromal and immune compartments needed to assess immunotherapeutic efficacy. We review the different methodologies for developing a more holistic patient-derived tumor organoid platform and for modeling the native immune tumor microenvironment.

The expansion of precision oncology and shortcomings of current biomarkers

Precision oncology broadly involves utilizing the characteristics of a patient’s tumor to help guide therapy, with the goal of improving clinical efficacy while limiting adverse effects. This personalized approach has led to a paradigm shift in the clinical practice of oncology. Specifically, the increasing availability of affordable and reliable next-generation sequencing technologies has led to the increased use of targeted therapies against the molecular defects of tumors, such as osimertinib for EGFR-mutated non-small cell lung cancer, or BRAF/MEK inhibitors for BRAF^V600E-positive melanoma [1]. Since the approval of the immune checkpoint inhibitor (ICI; see Glossary) ipilimumab in 2011, ICIs have become an integral part of front-line treatment for many advanced solid cancers [2]. Current ICIs in clinical practice, including antibodies against cytotoxic T lymphocyte-associated protein 4 (CTLA-4) and programmed cell death 1 (PD-1), promote anti-tumor immunity by blocking the engagement of inhibitory co-stimulatory molecules, thereby releasing the brakes on the immune system. Specifically, ICIs drive the expansion, migration, and differentiation of long-lived and poorly cytotoxic progenitor exhausted T cells into terminally exhausted T cells that are short-lived but highly cytotoxic (Box 1) [3].

Text Box 1. Mechanism of response to immune checkpoint inhibitors.

Upon exposure to chronic antigenic stimulation in the tumor microenvironment, CD8+ T cells become hypofunctional or exhausted, thereby limiting anti-tumor immunity. Recent studies have identified exhaustion as a dynamic process, whereby progenitor exhausted CD8+ T cells give rise to terminally exhausted T cells, both with distinctive molecular, transcriptional, epigenetic, and functional characteristics. Stem-like progenitor exhausted T cells express the transcription factor T cell factor 1 (TCF-1) and intermediate levels of PD-1, exhibit high proliferative capacity, are long-lived, poorly cytotoxic, and are responsive to ICIs. Terminally exhausted T cells lack TCF-1 but express high levels of PD-1, exhibit lower proliferative capacity, are short-lived, have higher cytotoxic potential, and are unresponsive to ICIs. ICIs stimulate growth of progenitor exhausted T cells and differentiation into terminally exhausted counterparts as relevant to both the clinical efficacy and potential resistance mechanism of ICIs.

As methods of tumor characterization, including proteomics, metabolomics, bioinformatics, histologic and cellular assays, are developed and implemented clinically, new biomarkers have emerged to guide clinical decision-making in immunooncology. For example, metastatic/unresectable cancers that are mismatch repair protein deficient (dMMR) or harbor high microsatellite instability (MSI-H) are more susceptible to ICI treatment via pembrolizumab blockade of anti-PD-1 [4, 5], leading the Food and Drug Administration (FDA) to grant the first-ever tissue/site-agnostic approval for a cancer therapeutic in 2017. Subsequently in 2020, the FDA approved the use of pembrolizumab for treatment of metastatic/unresectable cancers harboring a high tumor mutation burden (TMB ≥10 mutations/megabase), regardless of the cancer type [6]. These examples also underscore the emerging use of immunotherapeutics for treating various cancers, in which histological evaluation for mismatch-repair proteins or programmed death-ligand 1 (PD-L1) expression levels are more relevant than genomic data to help guide therapy with ICIs [4, 7]. As a result, the definition of precision oncology continues to evolve away from a “one gene, one drug, one disease” paradigm, towards using a wide array of genomic and non-genomic predictive markers to aid therapeutic selection independent of cancer type. Indeed, this is the basis for the National Cancer Institute (NCI)-MATCH trial, which aims to evaluate the response rate of refractory malignancies that are treated based on molecular and immunologic biomarkers [8].

Although precision oncology has expanded the landscape of treatment options for cancer patients, chemotherapy continues to remain the cornerstone of many cancer therapies. Only about 15% of patients with advanced cancers have potentially actionable mutations with about 7% of patients actually responding to genome-informed treatments [9]. Further, most who do respond to targeted therapy eventually progress due to resistance [10]. Additionally, clinical efficacy can vary greatly among different cancers that harbor even the same molecular defect [11], underscoring the heterogeneity of cancer cell dependence on a given mutation. Similar issues also plague the use of predictive biomarkers for immunotherapies. For example, PD-L1, which is currently the most widely used and accepted biomarker to guide anti-PD-1/PD-L1 antibody treatments, has disparate predictive clinical value among different cancers and even within the same cancer type; indeed, therapeutic benefit can be observed in PD-L1-negative tumors [7]. Additional challenges include using different PD-L1 cut-off thresholds in various clinical trials and the inter-assay variability among companion diagnostic assays. Other FDA-approved biomarkers for immunotherapies, such as dMMR/MSI-H or high TMB tumors, have certainly expanded the clinical utility of anti-PD-1 antibodies, but have not been able to improve objective response rates beyond 29–40% for advanced cancers [4, 6]. Given these shortcomings, other predictive biomarkers are under investigation to guide anti-PD-1/PD-L1 therapy, such as quantity of tumor infiltrating lymphocytes (TIL) [12], CD8+ T cell organization within the tumor microenvironment (TME) [13], exosomal or soluble PD-L1 levels [14, 15], tumor inflammation gene signatures [16], and ratio of exhausted CD8+ T cell reinvigoration to tumor burden [17].

Pre-clinical platforms in precision oncology

Due to the complexity of human cancers and the varied predictive value of current clinical biomarkers, pre-clinical platforms (e.g., patient-derived tumor xenografts, patient-derived tumor organoids) have been developed to aid in predicting efficacy of cancer therapeutics based on the characteristics of an individual’s tumor. Unlike the use of 2D cultures of immortalized human cancer cells, patient-derived tumor xenograft (PDX) and patient-derived tumor organoid (PDO) platforms can largely retain the phenotypic and genetic heterogeneity of the original tumors [18, 19].

Patient-derived tumor xenografts

The PDX platform involves the engraftment of primary clinical tumors either subcutaneously or orthotopically into immunodeficient organisms (primarily mice), thereby maintaining native interactions between the epithelial, stromal, and vascular compartments, and is thus an attractive in vivo pre-clinical model for cancer therapeutics [20]. Drawbacks of the PDX platform include limited and/or slow engraftment rates, expenses of immunodeficient mice, replacement of human stroma by mouse stroma with increased passaging, and differences in drug metabolism and pharmacokinetics between mice and humans [21]. Despite such issues, the PDX platform has been employed to predict cancer therapeutic efficacy. For example, a study comparing the response rate of various chemotherapeutic and targeted agents in 92 solid cancer patients versus the corresponding PDXs, displayed an 87% concordance rate using RECIST criteria [22]. Although encouraging, further large-scale prospective trials are needed to evaluate the clinical utility of using PDX in therapeutic selection. Lastly, due to lack of an intact human peripheral immune system, the PDX platform may not optimally evaluate immunotherapy efficacy. To overcome this limitation, humanized mice reconstituted with human-derived blood stem cells [23] or autologous in vitro expanded TIL [24] have successfully demonstrated in vivo anti-tumor immune responses with either ICIs or adoptive cell transfer of TIL, respectively.

Patient-derived tumor organoids

Organoids are self-organizing three-dimensional structures that are derived from primary tissue, and which recapitulate histologic architecture and multilineage differentiation. Since initial descriptions of intestinal organoids in 2009 [25, 26], organoid technology has been adapted to study cancer biology and pathogenesis from both “bottom-up” and “top-down” approaches [19]. On one hand, “bottom-up” approaches engineer oncogenic mutations into wild-type tissue organoids [27–30]. In contrast, the PDO platform is an example of a “top-down” cancer modeling approach, whereby dissociated tumor fragments (obtained from a biopsy or surgical specimen) are embedded into an extracellular matrix scaffold, most commonly Matrigel or collagen. These PDOs can then be used for numerous downstream translational applications, such as drug screening for novel therapeutics, study of TME heterogeneity, and used in precision oncology to predict therapeutic responses. Just as with the PDX platform, PDOs can consistently recapitulate in vivo tumor genotypes, phenotypes, and histopathological heterogeneity [31–40], thereby providing clinically relevant tissue for ex vivo modeling. One major advantage of PDOs over the PDX platform is the shorter time needed for PDO generation (days to weeks rather than months), which can be critical when predicting treatment responses in patients with advanced cancers who have limited lifespans. Another benefit is that normal tissue organoids can be generated in parallel to evaluate any potential off-target effects. One major drawback of PDOs has been the variable success rates in PDO establishment with different histologic subtypes [19], and apart from melanoma, the establishment of other non-epithelial PDOs has been problematic. A review of 17 recent publications attempted to assess the validity of PDOs as a predictive biomarker for clinical response in various cancers using different treatment modalities, including chemotherapies, combined chemoradiation, targeted agents, and ICIs [41]. Five of the 17 studies reported a statistically significant correlation between ex vivo PDOs and patient clinical responses, with a trend towards correlation in 11 of the studies. Although results from these retrospective/observational studies are promising, further large-scale prospective trials are needed to compare the efficacy of PDO-guided therapy versus standard-of-care. Below, we discuss recent publications assessing the clinical utility of using PDOs as a platform to predict therapeutic responses, with a particular focus on ICIs.

Using PDOs to assess chemotherapy/targeted therapy efficacy

PDOs generated with laminin-rich Matrigel or Basement Membrane Extract (BME) submerged below tissue culture medium typically contain exclusively epithelial tumor cells that preserve the genotypic and phenotypic heterogeneity of the original tumors. This suggests their use as personalized platforms to measure responses against chemotherapies or targeted agents. Indeed, human organoid biobanks have been established from multiple types of cancers and subsequently utilized in high-throughput drug screens to discover novel cancer therapeutics [31, 32, 35, 37, 42–44]. Additionally, a multitude of studies have attempted to correlate ex vivo PDO responses with clinical efficacy of chemotherapies or targeted agents in a variety of cancer types, including advanced colorectal [45–47], rectal [43, 48, 49], stomach [32, 37, 50], pancreatic [35, 51, 52], ovarian [53–56], esophageal [57], breast [31, 52], bladder [39], liver [58], prostate [59], and mesothelioma [60]. Several recent reviews have thoroughly summarized these clinical correlation studies [41, 61, 62] and attempted to analyze the level of evidence supporting the use of PDOs in clinical decision making. Results have encouragingly suggested positive correlations between PDOs and patient outcomes for particular cancers and therapies with the notable caveat that many of these studies had limited sample sizes and publication bias may favor successful results.

For example, a multicenter, prospective, observational clinical study (TUMOROID), showed a more than 80% correlation between ex vivo PDO drug responses and clinical responses in 22 metastatic colorectal cancer patients receiving irinotecan-based treatment [45]. The same research group further initiated the first formal, prospective intervention trial (SENSOR) to evaluate the feasibility of treating metastatic colorectal cancer patients, who had progressed on standard-of-care treatments, with investigational mTOR or AKT inhibitors based on their organoid drug response profile [63]. To limit the time between PDO response evaluation and initiation of next-line therapy, PDO drug screening was carried out in parallel to standard-of-care treatment. Of 61 enrolled patients, 19 had PDOs that responded to one of the 8 investigational drugs, 6 eventually received PDO-directed treatments, of which none achieved an objective clinical response. Although the authors concluded that organoid technology had limited value for precision oncology in this cancer population, several potential areas were identified for future improvements, such as organoid culture success rate and testing a larger panel of drugs. Future studies focused on optimizing culture conditions, standardizing assay approaches, and increasing sample size will be critical to achieving feasible organoid-based precision medicine. Additionally, meaningful survival advantages remain to be demonstrated.

Using PDOs to assess ICI efficacy

Unlike PDOs used to predict therapeutic responses to chemotherapies or targeted agents, PDOs used to assess efficacy of ICIs require both tumor and immune cell compartments. Recent progress has increased the complexity of the PDO platform to more reliably recapitulate the in vivo immune TME (Figure 1, Key Figure and Table 1), including in bladder cancer [64], pancreatic cancer [65, 66], CNS tumors [67], melanoma [68, 69], colorectal cancer [66, 70, 71], renal cell carcinoma [66, 69], and non-small cell lung cancer [66, 69] as described below.

Figure 1. Patient-derived tumor organoid (PDO) platforms that recapitulate the immune tumor microenvironment.

The immune tumor microenvironment can be constructed in PDOs via several approaches. (A) In the assembloid PDO platform, organoids containing exclusively tumor cells, often from physically and enzymatically dissociated tissues, are cultured in an extracellular matrix dome (Matrigel or BME) and submerged beneath tissue culture medium. In parallel, immune cells are isolated from peripheral sources (i.e., peripheral blood, lymph node) or from the tumor (i.e., tumor infiltrating lymphocytes, TIL) and subsequently co-cultured with the tumor organoid. (B) Alternatively, these isolated immune cells can be combined with physically and enzymatically dissociated tumors before submerging in an extracellular matrix dome. (C) Several PDO platforms have incorporated native tumor-infiltrating immune cells. In the patient- and murine-derived organotypic tumor spheroid (PDOTS/MDOTS) platform, tumors are physically and enzymatically dissociated, filtered to obtain appropriately sized spheroids, resuspended in a collagen matrix, then the spheroid-collagen mixture is injected into a 3D microfluidic culture device. (D) In the air-liquid interface (ALI) platform, physically dissociated tumor fragments are embedded in a collagen matrix on top of a transwell insert that is exposed to air with culture medium below it. (E) In the patient-derived tumor fragment (PDTF) platform, 1–2 mm³ tumor fragments are cut from different regions within a tumor and individually embedded into a collagen-based extracellular matrix with culture medium placed on top. Figure created with BioRender.com.

Table 1.

Detailed comparison of patient-derived tumor organoid (PDO) platforms that recapitulate the immune tumor microenvironment.

	Assembloid	PDOTS/MDOTS	ALI	PDTF
Source of tumor tissue	Surgical specimen, biopsy	Surgical specimen, biopsy	Surgical specimen, biopsy	Surgical specimen
Tissue dissociation	Tumors are dissociated physically and enzymatically (e.g., collagenase); lymph nodes and peripheral blood undergo Ficoll to isolate lymphocyte cell fractions; TIL are expanded from tumors.	Tumors are dissociated physically and enzymatically (e.g., collagenase), filtered to collect 40–100 μm-sized spheroid fractions.	Tumors are physically minced into fine tissue fragments.	1–2 mm3 fragments are cut from different regions within a tumor and frozen in a cryovial until later usage.
Extracellular matrix	Matrigel, basement membrane extract (BME), or collagen-based hydrogel	Collagen	Collagen	Collagen plus Matrigel
Culture instrument	Any size cell culture plate or dish	3D microfluidic culture device	Any size cell culture plate and insert	96-well plate
Culture method	• Tumor-only organoid cultures in Matrigel or BME are first generated then co-cultured with isolated immune cells [70–72]. • Isolated immune cells are combined with tumor cells before suspension in collagen-based hydrogel and dispensed as round droplet, then medium is added on top [68].	Spheroid-collagen mixture is injected into 3D microfluidic device with medium added into media channels.	Finely minced tumor tissue is mixed with collagen and dispensed on bottom of a transwell insert to solidify, then medium is added into the surrounding compartment with the top exposed to air.	Individual tumor fragments are thawed and placed in the collagen plus Matrigel matrix to solidify, then medium is added on top.
Cell types retained in culture	Tumor cells, primarily T cells (from peripheral blood, lymph nodes, or TIL expansion).	Tumor cells, native tumor-infiltrating immune cells (T and B cells, granulocytes, dendritic cells, myeloid-derived suppressor cells, tumor associated macrophages); determined by FACS and immunofluorescence.	Tumor cells, native tumor-infiltrating immune cells (T and B cells, myeloid cells, macrophages, NK cells), stromal fibroblasts; determined by FACS, single cell RNA-seq, and immunofluorescence.	Tumor cells, native tumor-infiltrating immune cells (T cells and non-T cells), non-immune cells (CD45-CD3-); determined by FACS.
Culture duration	Long-term culture required to maintain and expand tumor-only organoids; once reconstituted with immune cells, assembloids are cultured short term (i.e., days).	Short-term culture; long-term culture and ability to be serially passaged not reported.	Tumor cells can survive long-term and be serially passaged; immune cells and fibroblasts decline over a month but can persist longer with IL- 2 supplementation.	24–48h; long-term culture and ability to be serially passaged not reported.
Storage	Tumor-only organoid cultures can be cryopreserved and re-propagated; cryopreservation of assembloids not tested.	Cryopreservation of cultures not reported.	Cultures can be cryopreserved and re-propagated.	Individual tissue fragments can be cryopreserved and thawed at time of culture.
Possibility for immune reconstitution	Tumor-only organoids are reconstituted with immune cells isolated from several different sources (e.g., peripheral blood, lymph nodes, TIL expansion).	Can be adapted to include peripheral immune components (e.g., PBL, lymph nodes).	Can be adapted to include peripheral immune components (e.g., PBL, lymph nodes).	Can be adapted to include peripheral immune components (e.g., PBL, lymph nodes).
Modeling immune checkpoint blockade response	• 2 week coculture of tumor- only organoids and autologous immune cells can enrich for tumor- reactive T cells. • Addition of ICIs in assembloids increases T cell activation (via secretion of IFN-γ or upregulation of CD137) and increased tumor cell cytotoxicity.	Within 3 days of ICI addition in PDOTS, T cell activation can be measured via secreted cytokines and tumor cell cytotoxicity; MDOTS recapitulate differential sensitivity to ICI in different tumors.	5–7 day treatment with anti-PD-1/PD-L1 increases CD8+ T cell numbers and activation, resulting in increased tumor cell cytotoxicity.	Within 24–48 h of PD-1 blockade in PDTF cultures, T cell activation can be measured via activation markers and secreted cytokines.
Advantages	• Tumor-only organoids can be expanded longterm; preserves genetic and morphological heterogeneity of original tumor. • Specific Immune populations can be added to assess effect of different peripheral immune components (e.g., PBL, lymph nodes).	• Preserves diverse and functional immune cell populations. • Smaller number of cells and reagents are needed for microfluidic device (i.e., 250–300 μl of media). • Can be established in days. •Combinatorial therapies with small molecule inhibitors and ICIs have been validated.	• Preserves original tumor genetic and histopathological heterogeneity. • Preserves diverse and functional immune cell populations and T cell receptor repertoire. • Preserves stromal fibroblasts. • Can be cryopreserved, repropagated, and used in subsequent in vivo studies. • Can be established in days.	• Small tumor fragments allow for preservation of cellular architecture of tumor and sufficient nutrient and reagent access. • Mixing of fragments from different areas preserve tumor heterogeneity. • Collagen plus Matrigel matrix prevents immune cell efflux. • Can be established in days.
Limitations	•Establishment of tumor-only organoids can be technically challenging and time-consuming. • Lack of native tumor-infiltrating immune and stromal components.	•Requires Specialized equipment. • Lack of peripheral immune components (e.g., PBL, lymph nodes) may not fully recapitulate in vivo responses to ICIs.	•Lack of peripheral immune components (e.g., PBL, lymph nodes) may not fully recapitulate in vivo responses to ICIs. • Long-term (i.e., months) culture of immune TME not yet feasible.	• Requires pooling of multiple tumor fragments to reduce intratumorally heterogeneity, which may not be feasible with biopsies. • Lack of peripheral immune components (e.g., PBL, lymph nodes) may not fully recapitulate in vivo responses to ICIs.
Refs	[68, 70–72]	[73–75]	[66, 76]	[69]

PDOs to assess ICI responses: assembloids

In one strategy, exclusively tumor PDOs have been reconstituted with various stromal and immune components to form an “assembloid” structure. For example, tumor-only PDOs from biopsies of locally advanced rectal cancer patients undergoing neoadjuvant chemoradiation were combined with autologous TIL (Figure 1A and Table 1). Tumor cell death in the assembloids identified all 6 patients who would eventually obtain a pathologic complete response from the neoadjuvant treatment [71], underscoring the underappreciated immunomodulatory role of chemotherapy and radiation. The addition of anti-PD-1 to a subset of these assembloids increased TIL cytokine production and cytotoxicity, confirming responsiveness of assembloids to immunotherapy. Another study created assembloids by combining dissociated melanoma surgical specimens and matched lymph nodes or autologous peripheral blood lymphocytes (PBL), into a hydrogel extracellular matrix and screened against various treatment modalities (nivolumab, pembrolizumab, ipilimumab, and dabrafenib/trametinib) for ex vivo tumor cell death (Figure 1B and Table 1) [68]. Although assembloid responses to ICIs were correlated to clinical outcomes in 85% (6/7) of cases, the study population was entirely clinical non-responders, and assembloids created from the same patient’s blood and lymph nodes were counted as separate biologic replicates. In this study, the lymph node-enhanced assembloids were also used to enrich for tumor-reactive T cells from co-cultures with autologous PBL, a method previously described to improve TIL therapy [72]. As part of the NICHE study examining the efficacy of neoadjuvant ICIs (ipilimumab plus nivolumab) in early-stage mismatch-repair protein proficient (pMMR) and dMMR colon cancers, tumor-only PDOs were first generated and subsequently co-cultured with autologous PBL (Figure 1A and Table 1) [70]. Using CD8+ T cell IFN-γ as a readout for immunological reactivity, ex vivo responses correlated with clinical outcomes in 75% (9/12) of patients. Autologous T cell IFN-γ production was seen in 0 of 6 clinical non-responder patients but was observed in only 3 of 6 clinical responders suggesting greater specificity than sensitivity.

PDOs to assess ICI responses: spheroids, air-liquid interface, tumor fragments

A possible drawback of studies that utilize peripheral immune components (lymph nodes or PBL) in conjunction with tumor-only PDOs to predict clinical ICI efficacy is the lack of native tumor-infiltrating immune cells, particularly ICI-responsive progenitor exhausted T cells (Box 1). Indeed, tumor-intrinsic CD8+ T cell infiltration has been investigated as a potential biomarker for ICI efficacy [12]. A more holistic approach that incorporates native tumor-infiltrating immune cells into PDO construction could improve modeling of the in vivo immune TME. Two studies reported on the development of patient- and murine-derived organotypic tumor spheroids (PDOTS/MDOTS), which are ex vivo 3D microfluidic culture systems that retained native tumor-infiltrating immune cells without reconstitution (Figure 1C and Table 1) [73, 74]. To demonstrate that MDOTS could recapitulate in vivo sensitivity to anti-PD-1, MDOTS derived from PD-1-resistant B16F10 melanoma exhibited significantly less cell death in response to PD-1 blockade compared to MDOTS derived from PD-1-sensitive MC38 colon tumors. Notably, MDOTS were used to evaluate the pre-clinical efficacy of combining a novel TBK1/IKKε inhibitor with anti-PD-1 therapy, which correlated with prolonged survival in mice implanted with CT26 colon tumors and treated with the same combinatorial inhibitors. PDOTS were further used to identify CCL19 and CXCL13 as potential novel predictive/prognostic biomarkers by treating PDOTS from PD-1-responsive cancers (melanoma, Merkel cell carcinoma) with anti-PD-1 and observing an increase in CCL19/CXCL13 production [73]. Interestingly, this mirrored upregulation of CCL19/CXCL13 in a separate cohort of paired patient biopsy specimens (pre- versus post-PD-1 blockade). The limited clinical samples hindered determination of whether PDOT CCL19/CXC13 production was sufficient to reliably predict clinical outcomes in anti-PD-1-treated patients. In a separate report, PDOTS/MDOTS modeled the synergistic anti-tumor effect of combining various cyclin-dependent kinase inhibitors with anti-PD-1 treatment [75].

Another study described the generation of >100 human and murine PDOs via an air-liquid interface (ALI) method, whereby mechanically processed tumor fragments are grown in a collagen matrix on top of a transwell insert that is exposed to air with culture medium below it (Figure 1D and Table 1). The tumors were derived from both primary and metastatic sites, and encompassed numerous cancers including colon, small intestine, pancreas, lung, bile duct, kidney, esophagus, stomach, salivary gland, uterus, testes, and skin. This platform retained native epithelial, stromal, and infiltrating immune populations, including various T cell subsets, B cells, natural killer cells, and macrophages [66, 76]. Single cell gene expression analysis confirmed that ALI PDOs accurately preserved the diversity of immune cell infiltrates and T cell receptor (TCR) repertoire from the original tumor. Importantly, the ALI TIL retained their functional capacity to respond to anti-PD-1/PD-L1 treatment ex vivo via expansion and activation of antigen-specific CD8+ T cells with concomitant decline in tumor viability. Although 6 of 20 human ALI PDOs exhibited response to nivolumab via upregulation of cytolytic markers (i.e., IFN-γ, granzyme B, perforin-1), the study was not designed to investigate the correlation with ICI clinical responses.

Lastly, the patient-derived tumor fragment (PDTF) platform involves individually embedding 1–2 mm³ tumor fragments in a collagen plus Matrigel extracellular matrix to characterize early anti-PD-1-induced immunological changes (Figure 1E and Table 1) [69]. In a recent report, PDTFs were created using 37 tumors from five different cancer types (melanoma, non-small cell lung cancer, breast, ovarian, renal cell carcinoma), then treated ex vivo with anti-PD-1 for 48 hours and assayed for induction of 13 cytokines, 13 chemokines, and 4 T cell activation markers. This allowed for creation of an unbiased immunological score that could distinguish PDTF responders versus non-responders, which was then retrospectively correlated with clinical outcomes in patients who would eventually go on to receive anti-PD-1 treatment. Surprisingly, there was 100% concordance (12 of 12) between the ex vivo PDTF immunological responses and patient clinical outcomes (i.e., best radiographic response). In a separate validation cohort of 26 patients, full concordance between PDTF immunological responses and clinical outcomes was also observed. Caveats include the absence of radiographic confirmation in one of the 5 clinical responders, and two complete responses correlated with PDTFs that were derived from different tumors on the same patient. Interestingly, heterogenous PDTF immunological responses could be observed with different tumor lesions from the same patient. Lastly, three patients in the validation cohort had mixed clinical responses: two with associated PDTF responses and one with PDTF non-response.

Concluding Remarks

The increasing use of ICIs for solid malignancies is transforming the landscape of clinical cancer care. To maximize ICI therapeutic efficacy while limiting adverse effects, advances in predictive biomarkers and/or individualized ex vivo/in vivo platforms are needed. In conjunction with increased ICI usage, PDO platforms have evolved complexity to better model the native immune TME, each with their own advantages and disadvantages (Table 1; see Outstanding Questions).

Outstanding Questions Box.

• Which PDO platform(s) can best recapitulate the native immune TME and are most relevant for assessing ICI response?

• How well do PDOs predict clinical efficacy towards combinational immune and non-immune treatment modalities?

• Does addition of peripheral immune components (lymph nodes, PBLs) to PDOs improve prediction of ICI clinical response?

• Can PDOs reliably assess ICI efficacy in solid malignancies other than melanoma or non-small cell lung cancer?

• Can PDOs be used to evaluate responses or improve upon other immunotherapies besides ICIs?

• What steps in the PDO-guided process need further optimization to realize the routine use of PDOs in precision oncology?

Early attempts have included the reconstitution of tumor-only PDOs with stromal and immune components. In small cohorts, these assembloids have shown some promise in assessing clinical ICI responses, however, native tumor-infiltrating immune cells, particularly ICI-responsive progenitor exhausted T cells (Box 1), are absent. Additionally, the time-consuming process to generate tumor-only organoids is a limiting factor in clinical settings. Overcoming these deficits, more holistic PDO platforms (PDOTS/MDOTS, ALI, PDTF) can be constructed quickly (days instead of weeks) and incorporate the native immune TME and other stromal components. Of these different PDOs, the immune TME of the ALI platform has been characterized extensively, showing preservation of diverse functional immune cell populations beyond T cells, such as B, myeloid and NK lineages. This makes the ALI platform an attractive option for potentially modeling different classes of immunotherapies and network immune responses. Notably, the prevalence of progenitor exhausted T cells has not been examined in any PDO method, which could greatly facilitate ICI testing. The smaller sample volumes inherent to 3D microfluidic culture systems (PDOTS/MDOTS) could be beneficial in high throughput drug screens and when tumor tissue is sparse. As current standard of care for many metastatic solid cancers involves combinatorial approaches between ICIs and other treatment modalities (e.g., chemotherapies, targeted agents, radiation) [2], PDOTS/MDOTS have the added benefit of being validated against multimodal therapies (see Outstanding Questions). To date, only the PDTF platform has been validated to predict clinical ICI efficacy, although one caveat may include difficulty in recapitulating tumor heterogeneity in small biopsy specimens.

While holistic PDO platforms that contain native tumor-resident immune cells could predict clinical responses to ICIs, recent publications clearly implicate the importance of the peripheral immune system in supplying tumor-specific T cells [77], particularly progenitor exhausted T cells [78], which could be modeled with recently described lymphoid organoids [79]. Conceivably, PDO platforms with native tumor-infiltrating immune cells might better predict initial responses to ICIs, but PDO platforms that also incorporate peripheral immune components (lymph nodes, PBL) might better predict durable clinical responses. The natural evolution of the PDO platform may be to combine peripheral immune components with a PDO containing native tumor-infiltrating immune cells to best recapitulate the in vivo immune TME (see Outstanding Questions).

Understandably, these early PDO studies assessing ICI responses utilized PDOs generated primarily from ICI-responsive cancers like non-small cell lung cancer and melanoma, which tend to have higher tumor mutational burden and immune cell infiltrates. Therefore, future studies will need to address the applicability of PDOs to ICI efficacy in less-responsive malignancies (see Outstanding Questions). Although this review primarily focuses on ICIs, tumor-only PDOs are also being used to investigate the efficacy of cell-based immunotherapies (e.g., TIL, CAR T cells) in solid malignancies [80], which could conceivably also be modeled with more holistic PDOs containing native tumor-filtrating immune cells and/or peripheral immune components (see Outstanding Questions). For instance, co-cultures of tumor-only PDOs and PBL have been used to enrich for tumor-reactive T cells as relevant to TIL therapy [72]. In this study, IFN-γ was added to the co-cultures to induce both MHC-I and PD-L1 expression, an important factor to consider when accessing PDO response to ICIs.

Lastly, the practicality of using PDOs to guide ICI usage needs to be considered (see Outstanding Questions). The first prospective intervention trial (SENSOR) using tumor-PDOs was unable to demonstrate patient benefit from PDO-directed treatment with mTOR or AKT inhibitors [63]. Many steps from the acquisition of sufficient tissue for culture to testing PDOs in a timely manner will require further optimization for clinically meaningful translation towards the ultimate goal of organoid-based precision tumor immunotherapy.

Highlights.

• Recent advances in the patient-derived tumor organoid (PDO) platform have allowed modeling of the in vivo immune tumor microenvironment.

• In assembloids, tumor-only PDOs are reconstituted with autologous immune cell components from various sources.

• Diverse culture methods have been developed to generate PDOs that retain native tumor-infiltrating immune cells.

• Ongoing efforts are correlating PDO immunological responses to immune checkpoint inhibitor treatment with patient clinical outcomes.

Glossary

Air-liquid interface (ALI)

ex vivo 3D culture platform whereby minced tissue fragments are embedded in a collagen extracellular matrix on top of a transwell insert that is exposed to air

B16F10

Murine melanoma cell line

Basement Membrane Extract (BME)

formulations that mimic the extracellular matrix for cell culture purposes

CT26

Murine colon carcinoma cell line

Cytotoxic T lymphocyte-associated protein 4 (CTLA-4)

negative co-stimulatory molecule that downregulates T cell immune responses

High microsatellite instability (MSI-H)

increase in repeated DNA sequences that results from impaired DNA mismatch repair

Immune checkpoint inhibitor (ICI)

type of immunotherapy that blocks negative regulators of the immune response, thereby increasing anti-tumor immunity

MC38

Murine colon adenocarcinoma cell line

Mismatch repair protein deficient (dMMR)

loss of function of one or more enzymes involved in the repair of nucleotide mismatch errors

Mismatch repair protein proficient (pMMR)

enzymes involved in the repair of nucleotide mismatch errors are intact and functional

Murine-derived organotypic tumor spheroids (MDOTS)

ex vivo 3D culture platform whereby dissociated murine tumor tissues are embedded in a collagen extracellular matrix and injected into a 3D microfluidic device

Patient-derived organotypic tumor spheroids (PDOTS)

ex vivo 3D culture platform whereby dissociated human tumor tissues are embedded in a collagen extracellular matrix and injected into a 3D microfluidic device

Patient-derived tumor fragment (PDTF)

ex vivo 3D culture platform whereby different areas of a human tumor are minced into 1–2 mm³ tumor fragments and individually embedded into a collagen-based extracellular matrix

Patient-derived tumor organoid (PDO)

ex vivo 3D culture platform whereby dissociated human tumor tissues are embedded into an extracellular matrix scaffold, allowing for retention of the phenotypic and genetic heterogeneity of the original tumor

Patient-derived tumor xenograft (PDX)

in vivo pre-clinical model platform that involves the engraftment of primary clinical tumors into immunodeficient organisms, allowing for retention of the phenotypic and genetic heterogeneity of the original tumor

Peripheral blood lymphocytes (PBL)

lymphocytes isolated from the peripheral blood

Programmed cell death 1 (PD-1)

negative co-stimulatory molecule that downregulates T cell immune responses

Programmed death-ligand 1 (PD-L1)

expressed by antigen-presenting cells and tumor cells, it is the principal ligand for PD-1

Response Evaluation Criteria in Solid Tumors (RECIST)

methodology to measure tumor burden in evaluating the efficacy of cancer therapeutics

T cell receptor (TCR)

protein complex found on the surface of T cells that recognize peptide antigens bound to major histocompatibility complex molecules

Tumor infiltrating lymphocytes (TIL)

lymphocytic cell populations that have invaded the tumor tissue, which can be isolated, expanded, and used as a type of immunotherapy

Tumor microenvironment (TME)

complex environment surrounding a tumor, including fibroblasts, blood vessels, infiltrating immune cells, and extracellular matrix

Tumor mutation burden (TMB)

quantification of gene mutations that occur in the genome of a cancer cell