Cracking the Resistance Code: The Molecular Reclassification of Cancer and Precision Therapy Strategies

It’s about time: CML as a success story with lessons for solid cancers about the critical role of treating early—at diagnosis

Metastatic cancers are characterized by multiple genomic alterations [1–3] and ongoing growth leads to acquisition of additional mutations over time [14]. Treating cancers early with targeted therapy likely provides the best chance at improving outcomes. Chronic myeloid leukemia (CML) was historically a hematologic malignancy with poor prognosis and a median survival of 4 years after diagnosis [15]. A better understanding of the molecular underpinnings of this cancer and early treatment has transformed this disease into a chronic condition with a near normal life expectancy. Indeed, the three key factors that led to CML outcome transformation were as follows: (i) identifying the molecular driver (the aberrant BCR-ABL fusion and its altered enzymatic activity); (ii) developing drugs such as imatinib to target that driver; and, importantly, (iii) treating at the time of diagnosis, rather than in late-stage disease after failing chemotherapy (as mostly occurs in solid cancers).

Impacting CML occurred through several steps. The first was the discovery of the Philadelphia (Ph) translocation between chromosome 9 and 22 [16] and the resultant p210^BCR-ABL protein with an abnormally activated kinase enzymatic function [17, 18]. BCR-ABL aberrant enzymatic activity is critical for CML growth and has been dubbed oncogene addiction [19]. This addiction of CML to the kinase activity led to the critical realization that the aberrant BCR-ABL alteration was a potential target for therapy.

The development of tyrosine kinase inhibitors (TKI) to effectively block BCR-ABL was the second step. The first generation TKI imatinib led to a dramatic improvement in outcomes with 69% complete cytogenetic response at 1 year and 83% 10-year overall survival [20, 21]. Additional chromosomal abnormalities can evolve in response to imatinib therapy leading to resistance. The most common alterations include an additional Ph+ chromosome, trisomy 8, and isochromosome 17q. Resistance to imatinib was more commonly felt to be due to clonal evolution (58%) than the presence of BCR-ABL1 mutations (45%). Mutations in the kinase domain of BCR-ABL however, can lead to imatinib resistance [22]. Further clonal evolution occurs during progression to blast crisis [23, 24].

BCR-ABL1 dependent and independent pathways can drive resistance to TKIs such as imatinib. BCR-ABL dependent resistance typically occurs through acquisition of a second Ph chromosome or upregulation of transcriptional level of BCR-ABL1 [25, 26]. Multiple point mutations encoding for single amino acid substitutions in the kinase domain of BCR-ABL1 have been reported which can lead to imatinib resistance [27]. BCR-ABL1 independent pathways can occur via changes in pharmacokinetics due to altered efflux proteins (P-glycoprotein) or CYP3A4 metabolism [28]. Secondary chromosomal alterations and mutations may also develop. Non-BCR-ABL1 mediated leukemia stem cells (LSCs) can persist and enter a resting phase regardless of TKI treatment. Surrounding stromal cells can act on LSCs by stopping their cell cycle through growth factors and cytokines [29]. Second-generation TKIs are more potent and have a broader spectrum of activity against BCR-ABL1 mutations than imatinib and have shown benefit in patients who have developed resistance to imatinib [30]. Thus, second-generation (dasatinib, nilotinib) and third-generation (bosutinib, ponatinib) TKIs offer additional therapeutic options [21].

A critical step in the transformation of CML from a lethal leukemia to one of with a near-normal life expectancy was to treat CML at diagnosis with BCR-ABL-targeted drugs in order to achieve maximal suppression of BCR-ABL. As CML progresses from chronic phase to blast crisis, it acquires additional alterations which can make TKI-based therapy even less effective including TP53 alterations, ASXL1 alterations, complex copy number alterations, i(17q), and +21 leading to extremely poor prognosis [31]. CML evolving to blast crisis can acquire additional translocations and inversions. Myeloid blast crisis is more commonly associated with i(17q) and TP53 alterations while lymphoid blast crisis more commonly has monosomy 7, hypodiploidy, and CDKN2A deletions [32]. The accumulation of somatic alterations over time can be suppressed by TKI-based therapy which can maintain CML in chronic phase and potentially eliminate any rapidly cycling BCR-ABL containing cells which are at higher risk of acquiring novel alterations [31]. For patients whose CML is treated at diagnosis while in chronic phase, response rates are close to 100% with a median survival of over 20 years which approaches a normal life expectancy [15, 33]. If the disease is allowed to progress or is not discovered until late stage blast phase, response rates are negligible at less than 15% with a median survival of around 1 year [15].

The success seen in CML provides a blueprint for precision oncology therapy in solid tumors. The discovery of key molecular drivers of disease followed by the development of relevant inhibitors of these molecular drivers is critical for therapeutic success. However, even with a strong inhibitor of a powerful oncologic driver, treating later in disease can lead to poor outcomes as seen for CML. Treating prior to the development of advanced, metastatic disease is often not possible as patients may not come to medical attention early in the disease course. However, precision oncology clinical trials and approaches are frequently implemented after patients have failed one or more untargeted cytotoxic chemotherapy regimen. Additional clonal evolution occurs over the course of progression and when the drivers are finally targeted, outcomes are limited, akin to treating CML in blast crisis.

Evolution of precision oncology approaches

Solid tumors may go undetected prior to the onset of metastatic disease, which makes effective treatment challenging given clonal evolution and the development of multiple genomic alterations. Metastatic solid tumors have multiple drivers. Traditional approaches to metastatic solid tumor treatment involved regimens of one or more cytotoxic chemotherapy agents to target enhanced cell proliferation. These regimens are assessed and implemented based on histologic tumor type, rather than the molecular underpinnings of each patient’s individual disease. More recent approaches explore genomically driven, tissue-agnostic, and targeted therapies. Most often tumor-agnostic strategies involve biomarker-based monotherapies, but most recently, biomarker-based combination therapeutic approaches have been developed.

Cytotoxic chemotherapy regimens have traditionally been tested and approved based on the tissue of origin of cancer. Clinical trials (often randomized) will compare a novel regimen to the current standard of care to determine if there is benefit to the new regimen. In general, therapies for metastatic solid tumors that improve overall survival compared to the standard of care are approved for treatment, although at times approvals may be based on progression-free survival [34]. Response rates vary widely between tumor types and within tumor types. While some tumors such as testicular cancer have remarkable response rates to cytotoxic chemotherapy regimens with high rates of cure [35], the vast majority of metastatic cancers are not curable, and traditional cytotoxic strategies yield low response rates and/or limited duration of response. Five-year survival rates for patients with metastatic solid cancers ranges from 5–30% across malignancies [36]. It is difficult to predict which patients will benefit the most from a given treatment in the absence of biomarkers prior to administration of treatment. When a patient’s tumor fails to respond to first-line therapy, a second chemotherapy regimen is selected and administered, which usually has a lower response rate than the first-line treatment due to development of additional resistance pathways.

More recently, therapies have been tested for specific alterations using genomically driven approaches. Biomarker testing has been implemented in trials and multiple therapies have been approved based on genomic biomarkers in a specific type of cancer, e.g., EGFR inhibitors for EGFR-mutant lung cancer. The next evolution was clinical trials that were genomically based but tissue-agnostic [37]. Tissue-agnostic approvals have now been realized, with nine therapies (10 approvals, since pembrolizumab is approved for both microsatellite unstable and for tumor mutational burden high tumors) authorized by the FDA (pembrolizumab, larotrectinib, entrectinib, pemigatinib, dostarlimab, darbrafenib/trametinib, selpercatinib, trastuzumab deruxtecan-nxki, repotrectinib) [34]. While remarkable responses to single-agent targeted therapies have been seen, these responses are often not durable, which speaks to the importance of combination therapy as well as treating earlier in the disease course, before molecular evolution has occurred.

Given that metastatic tumors are molecularly complex, akin to malignant snowflakes, combination approaches are critical to limit resistance and prevent further clonal evolution. The ComboMATCH (Combination Therapy Platform Trial with Molecular Analysis for Therapy Choice) platform study is a trial conducted by National Cancer Institute (NCI). The goal of this study is to test molecularly targeted combination with at least one drug, but not usually all drugs, in the combination matched to a biomarker. Combinations may include either two targeted therapies or a targeted drug with a cytotoxic chemotherapeutic agent [38].

Each patient with metastatic cancer has a unique molecular profile and likely has multiple drivers for tumor growth; thus tailoring therapeutic combinations to each patient’s tumor profile may provide opportunities for further next-generation optimization of therapeutic approaches [9–12]. The I-PREDICT and WINTHER trials are the first examples of N-of-1 patient-centered customized combination, biomarker-based clinical trials. Each patient received a unique combination of therapeutic agents primarily based on genomic and immune profiling of the individual tumor landscape. These trials found higher degrees of matching drugs with the tumor’s profile led to higher response rates with longer progression-free and longer overall survival. Patients with the same histologic type of cancer were on dramatically different combination therapies since molecular alterations do not segregate well by tumor type [39]. Thus, there is evidence for improved outcomes with biomarker-matched targeting of multiple drivers simultaneously.

Synergy/additivity vs. targeting different clonal populations within the same patient versus targeting multiple alterations in the same clone

While combination therapy has historically often provided improved outcomes over single-agent therapy, even without biomarker-matching, it is often unclear if the combination of agents are delivering an additive or synergistic effect, or simply targeting different populations of patients within the clinical trial. Indeed, Palmer et al [40] showed that most of the effects of non-biomarker-based traditional combinations can be mathematically attributable to targeting different groups of patients within the population treated. Each patient benefits solely from the drug to which his or her tumor is most sensitive, with no added benefit from other drugs. However, other trials show evidence of additive or synergistic effects in certain settings. For instance, the BRAF and MEK inhibitor dabrafenib and trametinib both target the same pathway and, in patients with mutations such as BRAF V600E in that pathway, the combination achieves better outcomes than single agent dabrafenib [41].

Even with the above caveats, with biomarker-matching, the best impact of combinations may be achieved wherein each patient gets the best combination match for their tumor [9–11]. For instance, colon cancer was excluded from the tissue-agnostic approval of dabrafenib/trametinib given poor response rates seen in BRAF-altered colon cancer with BRAF/MEK inhibition [34, 42]. However, additional studies found that BRAF upregulates EGFR [42] and combining cetuximab with BRAF/MEK inhibition led to improved therapeutic outcomes [43]. Thus, blocking a resistance pathway can lead to an improved outcome.

Biomarker matching is however complicated by the heterogeneity of metastatic tumors, even within individual patients. Performing next generation sequencing (NGS) on blood circulating tumor DNA, which is comprised of shed DNA from multiple tumor sites, can help reveal the multiple alterations present in tumors regardless of whether these pathogenic molecular alterations occur in the same cells or in different clones of cells in the same patient [44, 45].

Figure 1 demonstrates tumor growth occurring through more than one pathway; blocking multiple pathways can be of benefit regardless of whether the driving pathways for growth are the same throughout the tumor (Figure 1A–C) or in separate cells/clones within a single tumor or within a single patient (Figure 2).

Convergent/Divergent Pathways Leading to Resistance

A concept that has sometimes emerged is the possibility of finding convergence points when there is more than one pathogenic alteration in a tumor and targeting that convergence point. As an example, pathways such as PDGFR and FGFR (Figure 3A) appear to converge on MEK; however, MEK inhibition alone may be ineffective because the pathways branch out into multiple new signals before the convergence point. NGS can identify alterations seen within a tumor sample; however, heterogeneity can exist between cells within a single tumor. If the activating PDGFR and FGFR alterations primarily activate JAK and PI3K, respectively, rather than MEK, then the pathways will not converge. If the pathways have minimal or lack of MEK activation, there will be limited responses to MEK inhibition. Oncologic signaling pathways can thus be convergent or divergent and classification can be challenging in the absence of additional technology such as single cell sequencing. Regardless, improved responses will likely occur by targeting each pathway separately upstream of MEK (Figure 3B). This example oversimplifies the complex oncologic signaling pathway network, but illustrates the importance targeting upstream effectors whenever possible.

Using serial liquid biopsies to identify new resistance drivers, to monitor residual disease, and for early detection of response or progression

Metastatic cancers develop additional alterations through growth and following therapy. As sensitive clones are destroyed with therapy, additional cancer cells will grow; these cells may have novel alterations allowing them to circumvent targeted therapies and requiring a change in therapy. It is helpful to have updated NGS information from a repeat tissue biopsy at progression (especially if the enlarging lesions are biopsied) in order to identify resistance alterations and optimize next line of therapy. However, the time required to obtain a biopsy and complete NGS testing, especially for rapidly progressive disease, may limit the ability to act on new alterations prior to starting the next line of therapy. Tumor location may also make repeat biopsy challenging. Liquid biopsy, usually by blood testing, may offer a viable and faster alternative. Liquid biopsy is thus of value at progression to determine if additional alterations have developed, which would help guide a change in therapy. Even for patients with disease that is amenable to tissue biopsy, the liquid NGS testing may provide faster results, enabling delivery of therapy modifications while awaiting NGS testing from repeat tissue biopsy [45, 46].

As an example, patients with BRCA1/2 inherited germline alterations have tumors that are sensitive to PARP inhibitors and platinum therapy. One mechanism of resistance is BRCA reversion whereby, after treatment with PARP inhibitors or platinum-based therapy, the tumor develops additional alterations to restore BRCA1/2 function [47]. This event can be identified through NGS of the tumor or liquid biopsy and leads to resistance to therapy.

Liquid biopsy with NGS testing of blood-based circulating tumor (ctDNA) has also been suggested for use in surveillance to determine recurrence prior to measurable disease detection by imaging [45]. It may also help monitor response to therapy. Panels such as Signatera are covered by certain United States insurance payors for monitoring response to immune checkpoint therapy in solid tumors and for evaluation of recurrence for colorectal cancer, breast cancer, bladder, ovarian, and non-small cell lung cancer [48]. Finally, some data suggest that serial liquid biopsies, including from fluids such as urine, may detect patterns of ctDNA fluctuation (spike followed by dramatic drop) indicative of response in the first week of therapy [49]. Blood biopsies after immunotherapy suggest that early response detection can occur but may not be evident for about three weeks [50]. Circulating tumor cells (CTCs) are another form of liquid biopsy which may also provide earlier detection of response or resistance [51, 52]. While standard ctDNA/cfDNA (cell-free DNA) testing generally require prior identification of tumor-specific alterations, low-coverage whole genome sequencing to determine cfDNA fragmentomes may allow for a tumor- and mutation-independent approach to estimating tumor burden and monitoring treatment response thus providing earlier detection of resistance [53]. In summary, serial liquid biopsies of blood and other fluids may provide crucial information regarding early detection of response or resistance.

While certain United States payors such as Medicare will reimburse for ctDNA testing for specific cancer types, other United States payors will not cover liquid biopsies [54]. Within the European Union, the adoption of liquid biopsy applications to enhance cancer care has not routinely been implemented. The European Liquid Biopsy Society was assembled to provide guidelines and help implement multiple liquid biopsy technologies including CTCs and ctDNA [55]. Liquid biopsy is generally reimbursed in Japan and South Korea, but is not routinely reimbursed throughout Asia [56]. Liquid biopsy availability has also been noted to be extremely limited in low- or middle-income countries [54].

Mixed response/intra-tumoral heterogeneity

A mixed response in metastatic cancers signifies one or more locations having tumor growth while others show response to therapy. Oncology practice has historically considered progression in any part of the tumor a sign that therapy is no longer effective and should be changed for a different treatment. When considering that metastatic cancers are heterogeneous, this practice may be suboptimal. If one portion of a tumor is responding, then the current therapy has benefit and should be continued. To address the non-responding areas, it is likely necessary to add another drug to the regimen [57].

As an example, Figure 4 depicts a cancer metastatic to the liver and the lungs. The lung disease is primarily driven by the PDGFR pathway while the liver disease is primarily driven by the FGFR pathway. Administration of a single agent targeted therapy will allow for tumor growth in one of the sites despite tumor shrinkage at the other site whereas combination therapy targeting both alterations will lead to responses in both metastatic sites. Adding one or more therapies to a regimen in the setting of a mixed response, rather than discontinuing a partially effective therapy, appears to be a more rational treatment strategy.

In addition to distinct drivers occurring in separate metastatic sites, different clones within the same site and tumor mass may have separate drivers [58]. Treating with a single matched therapy may lead to an initial response or stable disease; however, as clones with different alterations continue to grow, progression will be seen on scans. Figure 2 demonstrates tumor growth in part of the tumor is driven by the PDGFR pathway while other cells in the tumor are dependent on FGFR pathway activation. Administration of single agent targeted therapy can lead to an initial radiographic response followed by progression, whereas targeting both pathways can lead to more durable responses. It can be challenging to ascertain if progression is due to resistance to the initial therapy in the entirety of the tumor, which would necessitate switching therapy, vs. adding a second agent to target the clones lacking the driver targeted by the initial agent, when both drivers occur within cells occupying the same tumor mass.

Fusions as critical drivers of tumor growth

A better understanding of drivers of cancer growth and which drivers need to be prioritized for precision medicine approaches is critical for developing effective therapies. A gene fusion is a gene made by joining parts of two different genes. Fusions, such as BCR-ABL for CML, are considered to be critical drivers of tumor growth. There are multiple therapies approved for genomic fusions for both solid tumors and hematologic malignancies in oncology. The fusions with therapeutics currently available in the United States include BCR-ABL (CML, acute lymphoblastic leukemia), ROS1 (non-small cell lung cancer (NSCLC)), ALK (NSCLC), PML-RARA (acute promyelocytic leukemia), FGFR (myeloid/lymphoid neoplasm; cholangiocarcinoma), NRG1 (pancreatic; NSCLC), RET (solid tumors), and NTRK (solid tumors) (United States Food and Drug Administration Solid Tumor and Hematologic Malignancy Approvals [59]). A prior study explored differences in response rates to fusions as compared to non-fusions for FDA-approved therapies and found significantly higher response rates for therapies approved for fusions as compared to non-fusion alterations for both solid and hematologic malignancies. A further exploration into a single institution cohort found that, for patients with genomic fusions, patients whose therapies were matched to the fusions (versus to other types of alterations, such as mutations, also present in the tumors) had a significantly longer progression-free survival. Indeed, patients whose tumors harbored fusions who were matched to other alterations present in the tumor had a median progression-free survival that was indistinguishable from patients with fusions who were treated with unmatched therapy [60].

Fusions are likely strong drivers of cancer growth and developing therapies to target fusions is critical for effective cancer therapy. Treatment of a tumor with untargeted therapy or with targeted therapy without addressing the fusion will likely lead to poor outcomes with limited responses and rapid development of resistance. Examples of fusions commonly seen solid tumors without current targets include: EWSR1-FLI1 in Ewing’s sarcoma, SS18-SSX in synovial sarcoma, and TMPRSS2-ERG in prostate cancer [61]. Development of therapeutics to target these “undruggable” fusions will be important for improving therapeutic outcomes.

Strategies for Addressing Undruggable Genomic Drivers

Despite best efforts to follow genomically driven approaches, at times there are undruggable targets. While additional therapies targeting novel genomic alterations are under development, there remain strong cancer drivers that currently lack targeted therapeutics. These include undruggable fusions (i.e. EWSR1-FLI1 in Ewing’s sarcoma, SS18-SSX in synovial sarcoma, and TMPRSS2-ERG in prostate cancer [61]) and other undruggable genomic alterations. MYC and TP53 are amongst the most common driver alterations in oncology and considered to be undruggable genomic alterations [62]. KRAS, another common driver alteration, was previously considered an undruggable target as it lacked classic inhibitor-binding regions [63] until novel therapies were developed to target this alteration [64]. There are a number of approaches that may provide a pathway forward to circumvent undruggable alterations. These include targeting the immunome, RNA transcripts, and cell-surface proteins along with functional studies and synthetic lethality. These methodologies may identify alternative effective therapeutic approaches to allow for effective cancer cell destruction without targeting the undruggable alteration.

Strategies for Addressing Multiple Targets

Frequently, NGS identifies multiple targets for therapy in individual cancers. Previously treated patients on the I-PREDICT study had a median of 5 genomic alterations in their cancers, with a range of 1–20; physicians were able to administer a median of two matched therapies with a range of 1–6 [9]. Previously untreated patients with advanced cancers on the I-PREDICT study still had a median of 5 genomic alterations with a range of 0 to 15 [10]. Thus, heterogenous, advanced cancers may have multiple targets. Deciding on safe, efficacious starting doses for novel therapeutic combinations without guidance from phase I clinical trials is critical for a precision medicine approach. Administering multiple therapies simultaneously was accomplished through reducing initial doses of therapies [101–105]; prior studies suggested that patients on lower doses did not fare worse in phase I clinical trials of targeted therapies [106]. However, as the number of therapies increases beyond three to four, it can be challenging to administer tolerable, efficacious doses, given overlapping toxicities of many oncology therapeutics. A minimal dose of a targeted therapy is likely required to achieve target engagement leading to therapeutic efficacy.

Thus, for patients with a large number of alterations, alternative strategies are needed beyond targeting all possible alterations. It may be possible that not all alterations are strong drivers of tumor growth and may not need to be targeted. Literature knowledge of stronger oncogenic drivers and clinical trials data on approved therapies can be helpful in evaluating which alteration(s) to target. However, given the heterogenous nature of metastatic tumors, they may change dependence on pathways over the course of metastatic development and in response to treatment. Different areas of the tumor may be driven by multiple separate drivers. If the number of critical drivers exceeds the number of therapies that can be safely administered, synthetic lethality, immunotherapy, antibody-drug conjugates, and computationally driven approaches may help circumvent the need to target each individual alteration.

Computational/artificial intelligence driven approaches

Computational modeling of tumors has great potential to develop virtual tumor models especially in conjunction with artificial intelligence (AI) [111]. Patient-informed cellular and tissue-level models for cancer can be developed computationally that can serve as a digital twin to the patient and will predict tissue-level responses to different therapeutic combinations in a tumor microenvironment customized to patient information [112]. AI and machine learning can be used to test multiple combinations of potential therapeutics in the digital twin and determine which are most likely to be effective. In the setting of multiple potential targets, AI and machine learning can also be used to analyze multimodal-omics data (genomic, proteomic, immunomic, and transcriptomic data) and provide optimal personalized therapeutic recommendations.

Precision immunotherapy

Immunomics involves understanding the tumor immune microenvironment and host-tumor interaction. Immune checkpoint inhibition with CTLA-4 and PD-1 inhibitors has led to dramatic improvements in the outcomes for metastatic melanomas and many other solid tumors [113]. Biomarkers such as PD-L1 expression, microsatellite status (MSI), and tumor mutational burden (TMB) have provided insight into which patients are more likely to respond to PD-1/PD-L1 immune checkpoint inhibition [72, 114, 115]. Elevated tumor mutational burden and microsatellite status high have led to tissue-agnostic approvals for PD-1 directed monoclonal antibodies (dostarlimab for dMMR; pembrolizumab for MSI-high and tumor mutational burden >=10 Mutations/Megabase) [34, 116]. However, solid tumors may develop resistance to therapy after initial responses while other tumors may be immunologically “cold” with a low probability of response.

A number of immune checkpoints exist beyond the PD-1/PD-L1 axis, which may contribute to resistance to therapy; thus, combination therapy or biomarker-matched immunotherapies may help improve responses and overcome resistance. Combination immune checkpoint inhibition with anti-PD-1/PD-L1 and CTLA-4 or Lag3 inhibitors are FDA-approved and can provide improved responses over single agent therapy alone [59, 117–119]. The CTLA-4 inhibitor ipilimumab in combination with the PD-1 inhibitor nivolumab has been approved for hepatocellular cancer, MSI-high colorectal cancer, esophageal squamous cell cancer, renal cell carcinoma, mesothelioma, melanoma, and non-small cell lung cancer [59, 119]. Nivolumab and relatlimab-rmbw is a co-formulation of a PD-1 inhibitor with a LAG3 inhibitor which is approved metastatic melanoma [120]. IDO inhibitors were explored in combination with PD-1 directed therapies, however, had disappointing results in phase III studies possibly due to variability between different patients and lack of biomarker selected trials [121]. Other immune checkpoints that may contribute to immune resistance are being explored in clinical trials. ICOS agonists to affect the ICOS/ICOSL axis have been studied in combination with CTLA-4 and/or PD-1 inhibitors [122, 123]. TIM-3 inhibitors in combination with PD-1 inhibitors or as a TIM-3/PD-1 bispecific antibody inhibitor have been explored in clinical trials [124, 125]. VISTA [126, 127] and TIGIT [125] inhibitors are also under evaluation in combination immunotherapy approaches.

A comprehensive evaluation of immunomic expression patterns through RNA sequencing as well as other technologies can help better select patients for immune checkpoint therapies and novel clinical trials. While single agent PD-1/PD-L1 checkpoint inhibition may have improved responses over traditional cytotoxic chemotherapy in some cancers, a better understanding of the immunome and optimization of combination therapy will be critical to further optimizing outcomes. Determining expression patterns of other immune checkpoints at the start of therapy and following immune checkpoint inhibitor therapy may provide benefit in predicting primary or later resistance after an initial response [82, 83]. Novel bioinformatic tools can also be used with genomic and transcriptomic data to evaluate neo-antigens and T-cell and B-cell clonotype expansion [81]. Multiple factors may influence the efficacy of immune checkpoint inhibitor therapy and may include the number of mutanome-derived neo-antigens, immunogenicity of neo-antigens, ability of major histocompatibility complex (MHC) to present neo-antigens, the ability of patient’s T-cell receptor repertoire to recognize neo-antigens, and on which checkpoint(s) the tumor is dependent to evade the immune system [128–130].