Heterogeneity in precision oncology

Abstract

Precision oncology is a rapidly evolving concept that holds great promise in cancer treatment. However, a cancer complexity attributed to genomic and acquired tumour heterogeneity limits treatment effectiveness and increases toxicity. These limitations refer to both systemic therapies and radiotherapy, which are two mainstays of non-invasive cancer treatment. By understanding cancer heterogeneity and utilising advanced tools to personalise treatment strategies, precision oncology has the potential to revolutionise cancer care. In this article, we review the current status of precision oncology in solid tumours, specifically focusing on the impact of tumour heterogeneity and genomic patient features on systemic therapies and radiation. We also discuss the implementation of novel tools, such as next-generation sequencing and liquid biopsies, to overcome this problem.

Impact statement

Precision oncology, one of the most promising applications of precision medicine, uses molecular and genetic information to customise cancer treatments, considering the individual characteristics of each patient’s tumour. To further advance the field, precision oncology increasingly incorporates knowledge of cancer heterogeneity, on both spatial and temporal levels. Addressing these complexities with modern precision radiotherapy and systemic therapies is the key to targeting all cancer cell subpopulations. The future vision of precision oncology involves continuous advancements in technological and analytical methods, leading to further treatment personalisation. This progress will ultimately contribute to a paradigm shift in cancer care to improve patient outcomes significantly. Access to advanced tools should be improved in terms of availability and affordability while addressing the need for routine genomic profiling across various regions of primary and metastatic tumours to understand cancer heterogeneity comprehensively.

Introduction

Precision medicine is a novel approach to treatment and prevention that tailors strategies to the unique characteristics of individual patients, including their genetics, environment and lifestyle. It differs from conventional evidence-based medicine, which generally relies on average clinical benefits in the studied populations (Tonelli and Shirts, 2017; Blackstone, 2019). Precision medicine is supported by advances in technology and medical research, such as using genomic sequencing and big data analysis to identify individualised treatment options.

The decision-making process in precision medicine is based on predictive biomarkers, which offer insights into the underlying molecular mechanisms of tumorigenesis and allow the identification of potential therapeutic targets. In clinical settings, biomarkers can predict which patients are most likely to benefit from specific therapies, optimise treatment efficacy and reduce toxicity. Further, biomarkers enable early cancer detection and treatment monitoring, thereby increasing its efficacy. In essence, biomarkers are transformative tools of personalised medicine, driving more accurate, effective and safer cancer treatments (Slikker, 2018).

The two most commonly used markers are prognostic and predictive biomarkers. A prognostic biomarker is a clinical or biological indicator that offers insights into the probable health outcome of an individual patient, such as disease recurrence or death, regardless of the treatment pursued. In turn, a predictive biomarker signifies the potential advantage to the patient, resulting from a specific treatment (Sechidis et al., 2018). Other biomarkers include predisposing biomarkers, indicating the potential for developing a disease (Califf, 2018) and pharmacogenomic biomarkers, informing about the drug efficacy and toxicity based on the underlying genetic composition (Lauschke et al., 2017). The United States Food and Drug Administration and the National Institutes of Health published the Biomarkers, EndpointS and other Tools (BEST) resource, describing the extensive list of biomarkers used in translational science (Cagney et al., 2018).

Precision oncology is a concept that customises oncological care based on unique patient genomics and clinical, genetic, proteomic, transcriptomic or phenotypic tumour features (de and Ashworth, 2010; Collins and Varmus, 2015). Precision oncology has achieved unprecedented advancements through rigorous scientific evidence and extensive computational analyses (Mirnezami et al., 2012). However, challenges such as accurate data interpretation, precise patient stratification and the development of successful targeted therapies for specific genomic aberrations require further efforts (Prasad et al., 2016).

To overcome these obstacles, precision oncology requires innovative clinical trial designs that account for patient and tumour heterogeneity and the dynamic nature of cancer evolution (Chen and Snyder, 2013). Integrating precision oncology into clinical practice is a key goal of the Precision Medicine Initiative, which was launched by the US government in 2015.

In the present article, we discuss the impact of tumour and patient heterogeneity on treatment outcomes in solid tumours oncology and explore how precision systemic therapies and radiotherapy can mitigate these obstacles. The analysis will focus on scrutinising pivotal studies, such as the Molecularly Aided Stratification for Tumour Eradication Research (MASTER) (Horak et al., 2017), the National Cancer Institute Molecular Analysis for Therapy Choice (NCI-MATCH) trial (Flaherty et al., 2020 and other pertinent research, to better understand customised cancer therapy. We also present current investigative endeavours and interdisciplinary collaborations to optimise cancer therapy in all patients, regardless of their genetic makeup (Topol, 2014; Jameson and Longo, 2015; Figure 1). The examples provided here should be considered illustrative, as no comprehensive literature analysis on this topic was attempted.

Figure 1. Timeline showing the highlights of clinical precision medicine. ALK, anaplastic lymphoma kinase; ALL, acute lymphoblastic leukaemia; BCa, breast carcinoma; BRAF, v-Raf murine sarcoma viral oncogene homologue B’ BRCA, BReast CAncer gene; CAR-T, chimeric antigen receptor T-cell therapy; CML, chronic myeloid leukaemia; CR, complete response; ER+, oestrogen receptor-positive; HER2, human epidermal growth factor receptor-2; KRAS, Kirsten rat sarcoma virus; MET, hepatocyte growth factor receptor; MMRd, mismatch repair deficiency; MSI-H, high microsatellite instability; NGS, next generation sequencing; NSCLC, non-small cell lung cancer; PD-1, programmed death receptor-1; RET, Ret Proto-Oncogene; VEGF, vascular endothelial growth factor.

Systemic therapies

Systemic therapies, which involve drugs circulating throughout the body, are fundamental to cancer treatment (Chabner and Roberts, 2005). Precision oncology has revolutionised systemic therapies by better allocating standard chemotherapy and has paved the way for specific targeted therapies (Schwaederle et al., 2015). However, cancer heterogeneity, both spatial and temporal, highly impacts the effectiveness of these therapies (Greaves and Maley, 2012). As a result, one of the major challenges in oncology is customising systemic therapies for each patient and tumour characteristics (Leichsenring et al., 2019).

Tumour heterogeneity

Whereas cytotoxic chemotherapy is essential for many malignancies, it is generally recognised as a one-size-fits-all approach, which may not be optimal for patients with genetically diverse tumours. Precision oncology considers tumour genetic heterogeneity, thus can improve the efficacy of standard treatments, identify druggable targets for specific tumours and select patients who are more likely to benefit from customised treatments (Massard et al., 2017).

The relationship between specific genomic alterations, genetic inter- and intratumour heterogeneity and the effectiveness of cancer treatment has been well established (Schwaederle et al., 2015; McGranahan and Swanton, 2017). Tumours with certain genetic alterations differ in their susceptibility to classical cytotoxic chemotherapy. For example, mutations in the TP53, KRAS, PTEN or RB1 genes are associated with resistance to chemotherapy (Custodio et al., 2009; Perrone et al., 2010), BRCA1 and BRCA2 mutations denote chemosensitivity to platinum compounds (Pennington et al., 2014), and MGMT methylation in glioblastoma is associated with a better response to temozolomide (Stupp et al., 2005).

Knowledge of genetic tumour heterogeneity has been extensively used in targeted cancer therapies (Figure 2). If druggable, genetic alterations are primarily used as therapeutic targets; however, many also serve as predictive markers for treatment effectiveness. In colorectal cancer, cetuximab, which is a chimeric antibody against the epidermal growth factor receptor (EGFR), is effective only against wild-type rat sarcoma (RAS) family oncogenes (Van Cutsem et al., 2009; Douillard et al., 2013). Conversely, in lung cancer, EGFR tyrosine kinase inhibitors are less effective in patients with coexisting TP53 (Aggarwal et al., 2018; Sun et al., 2023) or KRAS mutations (Massarelli et al., 2007), which can activate alternative signalling pathways bypassing the EGFR pathway. In breast cancer, human epidermal growth factor receptor-2 (HER2) inhibitors are widely used to treat patients with HER2-overexpressing or HER2-amplified tumours, but they are less effective in patients with coexisting mutations in fibroblast growth factor receptor-1 (FGFR1) or receptor-2 (FGFR2) genes (Hanker et al., 2017). FGFR alterations correlate with resistance to several targeted and standard therapies across different malignancies (Babina and Turner, 2017), while the mechanistic and prognostic role of FGFR1–4 protein overexpression remains equivocal (Piasecka et al., 2019).

Figure 2. Examples of precision medicine biomarkers used in oncologic practice, together with respective targeted therapies approved in patients harbouring such lesions. CPI, checkpoint inhibitor; CTL, cytotoxic lymphocyte; DDR, DNA damage response; MMR, mismatch repair; MSI, microsatellite instability; PD-L1, programmed death-ligand 1; TMB, tumour mutational burden.

Some genetic alterations are druggable only in specific tumours, whereas others can be targeted across biologically and clinically different malignancies. The inhibitors of cyclin-dependent kinases 4 and 6 (CDK4/6) are effective and routinely administered to treat advanced hormone receptor-positive, HER2-negative breast cancer, though CDK4/6 alterations are not a hallmark of these cancers and do not predict the effectiveness of this therapy (Suski et al., 2021; Cristofanilli et al., 2022). Furthermore, CDK4/6 inhibitors are inefficient in liposarcomas harbouring the amplification of CDK4/6 and murine double minute 2 (MDM2) genes (Sbaraglia et al., 2021). Other examples are ivosidenib, an isocitrate dehydrogenase-1 (IDH1) inhibitor, and enasidenib, an IDH2 inhibitor, which effectively targets relapsed or refractory acute myeloid leukaemia with IDH1/2 mutations (Cerchione et al., 2021) but that are not effective in gliomas bearing these mutations. In turn, some targeted therapies, for example, entrectinib, which targets neurotrophic tyrosine receptor kinase (NTRK) fusions, and ROS oncogene 1 (ROS1) rearrangements, are effective across different solid tumour types, including lung cancer, colorectal cancer and thyroid cancer (Doebele et al., 2020; Drilon et al., 2020). Similarly, V-raf murine sarcoma viral oncogene homologue B1 (BRAF) and mitogen-activated protein kinase (MEK) inhibitors were recently approved with a tumour-agnostic indication for unresectable or metastatic solid tumours harbouring the BRAF V600E mutation (Gouda and Subbiah, 2023).

Immune-oriented therapies, such as immune checkpoint inhibitors (ICIs) or chimeric antigen receptor T-cells (CAR-T), have revolutionised cancer treatment. However, correctly identifying good responders remains challenging. Genetic heterogeneity in tumours may elicit variable responses to ICIs. Malignancies with a high tumour mutational burden (TMB) and neoantigen load are more responsive to ICIs. Patients with high-TMB non-small cell lung cancer (NSCLC) or melanoma achieve significant improvements in survival with ICIs compared with those with low TMB (Ning et al., 2022; Ricciuti et al., 2022). However, intratumour or intersite (primary vs. metastatic foci) heterogeneity leading to spatial neoantigen expression variability might result in the escape of certain subclones from immune surveillance (McGranahan and Swanton, 2017). Different tumour types (e.g., colorectal or endometrial cancers) with microsatellite instability or mismatch repair deficiency are highly responsive to ICIs (Cercek et al., 2022; O’Malley et al., 2022), whereas tumours with some mutations may be ICI resistant. For instance, EGFR-mutant NSCLCs are less sensitive to ICIs than wild-type EGFR (Mazieres et al., 2019). Melanomas with overactive WNT/β-catenin signalling are less infiltrated by T-cells and, thus, less susceptible to ICIs (Spranger et al., 2015).

Cancers with high levels of intrinsic heterogeneity, which can be defined as the presence of different genetic clones within a single tumour, are usually less likely to benefit from chemotherapy and targeted therapies (McGranahan and Swanton, 2017) because of the presence of drug-resistant subclones within the tumour that can contribute to rapid relapse after the initial response to therapy (Burrell et al., 2013; Almendro et al., 2014). Computational modelling and in situ analyses have shown that genetic and phenotypic heterogeneity can greatly affect tumour evolution during chemotherapy and treatment outcomes (Almendro et al., 2014). Recent advances in genomics and single-cell sequencing have shed light on the molecular mechanisms underlying tumour heterogeneity, paving the way for the development of novel personalised therapeutic strategies (Dagogo-Jack and Shaw, 2018; Ramón et al., 2020; Labrie et al., 2022). Further development of precision medicine for systemic anticancer therapies heavily relies on better understanding and addressing intratumour heterogeneity. However, because of diagnostic limitations, intratumour heterogeneity cannot yet be routinely exploited in guiding treatment options. It is expected that circulating tumour DNA (ctDNA) and single-cell analysis techniques may enable a detailed characterisation of tumour cell populations and better inform personalised treatment strategies (Nath and Bild, 2021; Tivey et al., 2022). ctDNA dynamic profiling allows for real-time monitoring of tumour evolution and adapting treatment strategies as the tumour mutates and evolves. Recently, ctDNA-guided therapy was shown to be beneficial in patients with NSCLC and colorectal cancer (Jee et al., 2022; Tie et al., 2022).

Precision medicine approaches may considerably improve cancer treatment outcomes, provided that the complex interplay between tumour genetics and response to systemic treatment is better understood. Basic and translational studies are essential for identifying next-generation predictive biomarkers. Novel clinical trial designs, such as basket-type trials assessing the druggability of specific targets across different tumour types, and umbrella-type trials evaluating the efficacy of specific or various targeted therapies in specific cancer diagnoses (Figure 3 and Table 1), may prompt the development of new tailored therapies (Subbiah, 2023).

Figure 3. Types of precision medicine clinical trials. BRAF, v-Raf murine sarcoma viral oncogene homologue B; SOC, standard of care.

To compile the data presented in Tables 1 and 2 (please see below), we employed a multi-pronged search strategy. Firstly, a targeted search was conducted on PubMed using the following query: ("precision medicine"[Title/Abstract] OR “targeted therapy”[Title/Abstract] OR “personalised medicine”[Title/Abstract]) AND “clinical trial”[Publication Type] AND ("2013/01/01"[PDAT]: “2023/12/31”[PDAT]). This query was designed to yield articles classified as “clinical trials” focusing on “precision medicine,” “targeted therapy,” or “personalised medicine,” and published between January 1, 2013, and December 31, 2023. In addition to PubMed, we supplemented the articles with data from the clinical trials registry ClinicalTrials.gov and information gathered from sessions at the European Society for Medical Oncology (ESMO) Meetings and American Society of Clinical Oncology (ASCO) Annual Meetings held between 2018 and 2023. These conferences are recognised platforms that regularly feature key developments in precision medicine trials in oncology.

Table 1. Selected clinical trials investigating personalised cancer therapies

ALK, anaplastic lymphoma kinase; CBR, clinical benefit rate; CI, confidence interval; ctDNA, circulating tumour DNA; DCR, disease control rate; DMFS, distant metastasis-free survival; DNA, deoxyribonucleic acid; EGFR, epidermal growth factor receptor; FDA, Food and Drug Administration; FFPE, formalin-fixed, paraffin-embedded; ICD – O, International Classification of Diseases for Oncology; IHC, immunohistochemistry; KRAS, Kirsten Rat Sarcoma Viral oncogene homologue; MSI, microsatellite instability; MTB, molecular tumour board; NGS, next-generation sequencing; NR, not reported; NSCLC, non-small cell lung cancer; OR, odds ratio; ORR, objective response rate; OS, overall survival; PD-L1, programmed death-ligand 1; PFS, progression-free survival; PFSr, PFS interval associated with molecularly informed therapy (PFS2) divided by the PFS interval associated with the last prior systemic therapy (PFS1); RNA, ribonucleic acid; RNA-seq, RNA sequencing; TMB, tumour mutational burden; VEGF, vascular endothelial growth factor; VEGFR, vascular endothelial growth factor receptor; WES, whole exome sequencing; WGS, whole genome sequencing.

Table 2. Summary of selected clinical studies investigating radiosensitivity-predicting genomic signatures

95%CI, 95% confidence interval; HR, hazard ratio; sHR, subdistribution hazard ratio.

Germline heterogeneity

Response and adverse reactions to systemic therapies can vary substantially between individuals. This variability has been attributed mainly to the inherited genomic variants that inactivate protein-coding genes (Karczewski et al., 2020). The therapy may be impacted on several levels, including direct drug–target interactions, drug metabolism (including drug activation and removal) and downstream effects (e.g., DNA damage). Hence, considering patient pharmacogenomics can improve treatment outcomes, decrease toxicity and reduce costs.

The drug-metabolising enzyme known as cytochrome P450 2D6 (CYP2D6) is the most thoroughly examined and variable polymorphic enzyme (Zhou and Lauschke, 2022). Its deficiency is inherited through an autosomal recessive trait, and individuals carrying this alteration are classified as poor metabolisers. However, the remaining subjects (extensive metabolisers) display considerable variability in their enzymatic activity (Bertilsson et al., 2002). The gene encoding the CYP2D6 protein is highly polymorphic, with over 100 allelic variants described to date (Gaedigk et al., 2018). Specific genetic variations in the CYP2D6 gene can cause altered activity of the cytochrome P450 2D6 enzyme, which is involved in the metabolism of tamoxifen (a selective oestrogen modulator used to treat breast cancer). Individuals with reduced CYP2D6 activity (e.g., *4, *5 and *6 alleles) account for up to 10% of patients (Ingelman-Sundberg, 2004; Crews et al., 2014). These individuals have a lower efficacy of tamoxifen than those with normal activity (Lim et al., 2007; Schroth et al., 2007; Goetz et al., 2013). Likewise, the ultrarapid metabolisers (e.g., with *1xN or *2xN alleles, about 1–2% of patients) show lower endoxifen (a tamoxifen metabolite) concentrations and worse outcomes compared with patients with normal CYP2D6 activity (Wegman et al., 2005; Schroth et al., 2007; Crews et al., 2014).

Similarly, variations in the DPYD gene (e.g., *2A or *13), which encodes the dihydropyrimidine dehydrogenase, an enzyme involved in the metabolism of fluoropyrimidines, such as 5-fluorouracil and capecitabine, can lead to reduced enzyme activity and an increased risk of severe toxicity (Amstutz et al., 2009; Offer et al., 2014; Henricks et al., 2018).

Finally, the UGT1A1 gene, which encodes for the uridine diphosphate glucuronosyltransferase 1A1 enzyme, is involved in the metabolism of irinotecan (a topoisomerase I inhibitor); variations reducing its activity (including *6, *27 and *28) may cause severe toxicity, including neutropenia, diarrhoea or infection (Iyer et al., 2002; Innocenti et al., 2004; Marcuello et al., 2004; Xu et al., 2016).

Genomic polymorphisms in drug targets may also affect interactions with drugs. Several HER2 gene variants have been shown to impact the effectiveness of trastuzumab (a monoclonal antibody targeting HER2) in HER2-positive breast cancer patients. The F117L variant, which is located in the extracellular domain of the HER2 protein, impairs trastuzumab binding by approximately threefold compared with wild-type HER2. Decreased binding affinity is attributed to the introduction of a leucine residue, which causes a steric hindrance and disrupts the protein conformation at the binding site (Gaborit et al., 2011). In turn, a variant within the HER2 intracellular kinase domain (T798I) leads to increased kinase activity, conferring resistance to lapatinib (Bose et al., 2013). Another example of EGFR polymorphism is R521K (rs2227983), which may decrease the response to cetuximab in patients with metastatic colorectal cancer (Graziano et al., 2008). As a result, patients who carry this allele experience a lower incidence of skin toxicity during cetuximab treatment (Klinghammer et al., 2010; Fernández-Mateos et al., 2016).

Finally, variants affecting the efficiency of DNA damage repair may affect the response and toxicity of numerous drugs, including platinum agents, alkylating agents, topoisomerase II inhibitors, antimetabolites and poly-ADP ribose polymerase inhibitors. The rs3212986 variant (C8092A) of the ERCC1 gene, which is a component of the nucleotide excision repair pathway, is associated with a poor response to platinum agents in NSCLC and ovarian cancer (Zhou et al., 2004; Krivak et al., 2008). Similarly, the rs13181 (L751G) single nucleotide variant (SNV) in the XPD gene is associated with the poor efficacy of platinum agents in NSCLC (Park et al., 2001).

Radiotherapy

Radiotherapy is a vital component of cancer treatment, applicable in around 60% of patients (Citrin, 2017). Until recently, radiotherapy was prescribed on the empirical basis of a one-fits-all approach, assuming a similar response to the same radiation dose. Recent advances in precision medicine have enabled the use of more targeted and personalised radiotherapy regimens tailored to the specific characteristics of individual patients and their tumours. Cancer heterogeneity poses a significant challenge for radiotherapy, as it can cause variable tumour responses and the emergence of radioresistant cell populations. Precision radiotherapy, by considering the comprehensive molecular and genetic tumour makeup, may overcome these challenges and allow for the development of tailored treatment plans. By integrating genomic data and other biomarkers, precision radiotherapy has the potential to maximise tumour control while minimising toxicity to surrounding healthy tissues.

Tumour response

Technological advancements in radiotherapy have increased the potential of physical radiation tailoring to personalise treatment. However, the optimisation process typically focuses on dose conformality, ignoring biological factors and assuming that all tumours react similarly to radiation (Price et al., 2023). Unlike medical oncology, where genomic signatures have become part of routine practice (e.g., MammaPrint tests, Oncotype DX or PAM50), their use in radiotherapy has been limited (Parker et al., 2023). Meanwhile, radiation impacts several molecular pathways, such as DNA damage, hypoxia or proliferation (Reisz et al., 2014; Wang et al., 2018; Huang and Zhou, 2020).

Several somatic mutations have already been established as conferring radioresistance. Numerous studies, including breast (Jameel et al., 2004), colorectal (Munro et al., 2005) and head and neck cancers (Hutchinson et al., 2020), gliomas (Werbrouck et al., 2019) and sarcomas (Casey et al., 2021) have shown that TP53 mutations might impair radiotherapy response. Other notable examples are KEAP1 and NFE2L2/NRF2 mutations in NSCLC and head and neck cancers (Binkley et al., 2020; Guan et al., 2023). In addition, the coexistence of KRAS and SMAD4 mutations is an indicator of radioresistance in cervical cancer (Oike et al., 2021).

To date, there has been scarce data on molecular predictive signatures in radiotherapy. These examples comprise the PORTOS classifier encompassing 24 genes to predict the efficacy of postoperative radiotherapy in prostate cancer (Zhao et al., 2016) and the Adjuvant Radiotherapy Intensification Classifier (ARTIC) and POLAR classifiers, which incorporate 27 and 16 genes, respectively, to predict outcomes of postoperative radiotherapy in breast cancer (Sjöström et al., 2019, 2023).

Unlike PORTOS or POLAR, which relate to specific cancers and clinical situations, a radiosensitivity index (RSI) has also been proposed as a pan-cancer and specific marker of cellular radiosensitivity. This index is based on the expression of 10 genes (AR, c-JUN, STAT1, PKC, Rel A, cABL, SUMO1, CDK1, HDAC1 and IRF1) related to DNA damage response, cell cycle, apoptosis and proliferation (Eschrich et al., 2009). Based on RSI, a quantitative metric for the biological effect of RT, the genomic-adjusted radiation dose (GARD) has been developed. GARD was initially validated in patients with breast cancer, lung cancer, pancreatic cancer and glioblastoma (Scott et al., 2017). This signature was further tested in a pooled, retrospective, pan-cancer cohort and reported as a continuous variable associated with time to first recurrence and overall survival (Scott et al., 2021). Recently, GARD has been employed in a provocative in silico analysis to explain the unexpected results of the seminal RTOG 0617 trial (unsuccessful radiotherapy dose escalation in locally advanced NSCLC) (Scott et al., 2021). The authors assumed that this model allows for deriving an optimal radiation dose in each patient. Another study employing prospectively collected tissues showed that low RSI values (denoting higher radiosensitivity) are associated with increased immune infiltration and activation (Grass et al., 2022). Recently, based on the reanalysis of the publicly available datasets – Merged Microarray-Acquired Dataset (Bin Lim et al., 2019) and the Cancer Genome Atlas (Weinstein et al., 2013) – RSI was shown to be associated with immune-related features and predicted response to PD-1 blockade (Dai et al., 2021). However, a recent analysis showed that RSI is not associated with survival and should not be used for radiation dose adjustments (Mistry, 2023). It was also suggested that the RSI of tumour clones remaining after RT, instead of the initial tumour population, should be evaluated to better predict the RT outcome (Alfonso and Berk, 2019).

Incorporating genomic signatures in radiotherapy decision-making has shown significant advancement through recent research, such as the GARD-based trial, to optimise radiotherapy for triple-negative breast cancer (NCT05528133). The European Organisation for Research and Treatment of Cancer has appraised the evidence from RSI/GARD studies as a priority for phase 3 clinical trials in radiotherapy (Thomas et al., 2020). However, the clinical utility of these approaches warrants an evaluation that integrates molecular data into prospective clinical trials and routine clinical practice (Table 2).

Radiotherapy tolerance

The impact of genetic heterogeneity on normal tissue toxicity following radiotherapy is a significant concern in cancer treatment. Individual genetic variations can influence the severity of radiation-induced side effects (Barnett et al., 2009). Normal tissue complications can range from mild to severe and may include skin reactions, inflammation, fibrosis and organ dysfunction (Bentzen, 2006). However, except for several radiosensitivity syndromes related to biallelic pathogenic mutations in DNA repair genes and deleterious heterozygous ATM mutations in young patients, no genomics-guided radiotherapy is currently used (Bergom et al., 2019).

Since the beginning of the twenty-first century, more than 100 articles analysing the impact of DNA sequence changes on the frequency and severity of radiation-induced complications have been published (Andreassen et al., 2016). Most of these studies have addressed SNVs, which typically affect the genes responsible for processes such as DNA break or inflammation. However, these studies were usually small (median of approximately 150 patients), hence lowering the statistical power for comparisons (Andreassen et al., 2016).

To reduce the bias associated with the publication of numerous low-quality studies, the Radiogenomics Consortium (https://epi.grants.cancer.gov/radiogenomics/) was created in 2009 (West et al., 2010). This initiative allowed for assembling adequate groups of patients with diverse clinical characteristics and validating presumed associations of SNVs with radiation toxicity. However, the results of the prospective study published in 2012 were a huge disappointment because none of the reported relationships (98 SNVs in 46 genes) were confirmed (Barnett et al., 2012). However, this experience prompted the development of research employing large-scale techniques such as genome-wide association studies (GWAS). As a result, potentially interesting SNV associations with radiation reactions were found, such as variants at the locus of the TANC1 gene that was found to be encoding a protein responsible for muscle cell regeneration (Fachal et al., 2014). The strength of these associations is much higher, with odds ratios of 1.3–1.5, compared with 1.1–1.2 observed in typical GWAS studies (Zhong and Prentice, 2010). The Radiogenomics Consortium remains active, and a significant increase in sample size has led to the discovery of several potentially relevant relationships. An analysis of breast and prostate cancer patients from 17 cohorts indicated that the ATM rs1801516 SNP is associated with an increased risk of radiation toxicity (Andreassen et al., 2016). A recent study has revealed a strong association between radiation-induced mucositis and the rs1131769*C locus in the STING1 gene on chromosome 5 (Schack et al., 2022).

A definitive answer to the SNVs’ role in healthy tissues’ response to radiation may come from the international, multicentre REQUITE project (www.requite.eu) funded by the European Union through its 7th Framework Programme (West et al., 2014). The project, performed in collaboration with the Radiogenomics Consortium, aimed at predicting and reducing the risk of long-term side effects of radiotherapy and completed patient recruitment (Seibold et al., 2019). REQUITE reported that polygenic risk scores (PRS) may be clinically useful and that incorporation of SNP-SNP interactions improves patient classification and prediction of radiotherapy-related toxicity (Franco et al., 2021). This project significantly advanced the collaborations among stakeholders, including healthcare professionals, researchers and industry partners, highlighting the importance of personalised radiotherapy. Other collaborative genetic association studies at both the national and global levels include Gene-PARE (Ho et al., 2006), RadGenomics (Iwakawa et al., 2002) and RAPPER (Burnet et al., 2013). Understanding the impact of radiogenomic heterogeneity on normal tissue radiation toxicity is essential for developing more effective and safe personalised strategies.

In summary, for a long time, radiotherapy optimisation was focused on dosage conformity rather than biological factors. It is critical to factor in tumour heterogeneity when considering radiotherapy outcomes; hence, developing and verifying molecular signatures such as PORTOS, ARTIC and POLAR, together with pan-cancer RSI, constitute the base for resolving this predicament.

Other factors

Circadian heterogeneity

The efficacy of anticancer treatment may also be affected by circadian rhythm, a biological phenomenon displaying endogenous, untrainable 24-h oscillation (Lee, 2021). The circadian clock regulates several key processes in the human body, including metabolism and cell division, with 40% of the transcriptome under circadian control in at least some tissues (Ruben et al., 2018). Recently, a comprehensive analysis of clock genes across different human cancers was performed using primary solid tumour data from The Cancer Genome Atlas (Ye et al., 2018). Based on the available evidence, the International Agency for Research on Cancer classified shift work that involves circadian disruption as potentially carcinogenic to humans (IARC Monographs Vol 124 group, 2019).

Chronotherapy involves administering treatment at specific times of day to optimise its effectiveness and minimise side effects (Zhou et al., 2021). This approach has been tested in several clinical trials with conflicting results: some showed improved efficacy and reduced toxicity (Lévi et al., 1997; Giacchetti et al., 2006), whereas others did not demonstrate significant differences (Garufi et al., 2006; Qvortrup et al., 2010). Some data indicate that chronomodulation might be relevant in the context of immunotherapy. For instance, a recent provocative study reported inferior overall survival in patients who received more than 20% of immunotherapy infusions after 4:30 PM (Qian et al., 2021). However, these observations warrant verification in prospective randomised clinical trials. Data pertinent to radiotherapy comes from the REQUITE project, which disclosed novel serendipitous associations, for example, the interaction between time, circadian rhythm-related genes (CLOCK, PER3 and RASD1) and late radiation toxicity in breast cancer patients (Webb et al., 2022).

Microbiome heterogeneity

The human microbiome, comprising trillions of diverse microbial organisms, plays a significant role in modulating health and disease states, including cancer (Hou et al., 2022). Recent research indicates that microbiome heterogeneity can greatly influence response to anticancer treatment via drug-microbiota interactions.

Bacteria-derived enzymes target chemical compounds, including drugs used in systemic treatment. For example, approximately 40% of patients treated with irinotecan experience severe mucositis, sometimes leading to treatment cessation. Irinotecan is converted into its active form, SN38, which is later reverted back to an inactive form, SN38G, in the liver. Bacterial β-glucuronidases can then convert SN38G in the gastrointestinal tract back to its toxic form (Wallace et al., 2010). Additionally, Bifidobacterium longum, Collinsella aerofaciens and Enterococcus faecium abundancy in stool was found to be associated with increased response to anti–PD-1 treatment in patients with melanoma (Matson et al., 2018).

Additionally, the gut microbiome generates numerous metabolites, with short-chain fatty acids (SCFAs) being among the most prevalent and crucial. Significantly, SCFAs act as secondary messengers that facilitate signal transmission and influence the onset and progression of various diseases. Radiotherapy can modify the populations of bacteria that produce SCFAs, leading to changes in SCFA levels, which are linked to several conditions, including radiation-induced intestinal injury (Li et al., 2021).

The microbiome studies shape oncologic outcomes and are now being leveraged for the development of novel personalised therapeutic approaches in anticancer treatment. However, this topic exceeds the scope of this paper and has been addressed elsewhere (Chrysostomou et al., 2023; Yi et al., 2023).

Conclusions

Precision medicine has made remarkable progress in oncology by promising to administer therapy to “the right patient at the right time” (Abrahams, 2008). This review has discussed the impact of cancer heterogeneity as a major challenge facing precision oncology development. Apart from affecting treatment outcomes, heterogeneity can also be employed in the context of prevention and early detection. The results of the first large-scale observational cohort study evaluating methylation-based multicancer early detection diagnostic test (SIMPLIFY) have demonstrated the feasibility of this approach (Rebbeck et al., 2018; Nicholson et al., 2023; Tie, 2023).

Genomic makeup has been shown to impact the effectiveness and toxicity of systemic treatments and radiotherapy. Thus, genomic testing can identify pathogenic gene variants and polymorphisms affecting drug metabolism or mechanism of action, thus increasing the risk of treatment failure or toxicity. Spatial and temporal tumoural heterogeneity is a complex phenomenon linked to resistance to therapy, disease progression and adverse prognosis. There are substantial genetic and molecular differences across various tumour regions and between primary and metastatic foci. A better understanding of this phenomenon allowed for the development of novel strategies, for example, targeting with systemic therapies or radiation-specific tumour regions or populations of cancer cells.

Implementing advanced technologies, such as NGS, liquid biopsies and imaging modalities, has fostered precision oncology, accounting for both genomic and tumoural heterogeneity. NGS is routinely used to examine mutations in, for example, EGFR, BRAF and ALK, which are molecular targets for modern therapies. Liquid biopsies, which involve analysing circulating tumour cells or circulating tumour DNA, offer a noninvasive way to identify genetic alterations and monitor tumour progression. Several clinical trials, including the ongoing NCI-MATCH and MASTER trials and the previously completed MyPathway and MPACT trials, have been designed to identify genetic mutations associated with specific targeted therapies and develop novel treatment strategies for overcoming therapy resistance. These trials are expected to prompt further development of precision oncology.

As precision oncology continues to evolve, the future holds great promise for overcoming current challenges. Advanced tools should be more accessible and affordable. There is a need for routine, more comprehensive genomic profiling of different regions of primary and metastatic tumours to fully understand cancer heterogeneity. In addition, integrating machine learning algorithms and artificial intelligence would allow better identification of new therapeutic targets and the development of even more personalised treatment strategies. An exciting area of future research in precision oncology is the use of combination therapies that simultaneously target multiple pathways and molecular targets; this approach has the potential to overcome heterogeneity-led resistance to single-agent targeted therapies. Another area for improvement is integrating precision oncology into clinical practice and expanding access to new technologies for community oncologists and patients. This will require the development of user-friendly platforms and tools that are easily integrable into clinical workflows.

Overall, precision oncology holds great promise for improving cancer treatment efficacy by enabling personalised treatment strategies based on unique cancer and patient characteristics. Although challenges remain to be addressed, ongoing research and emerging developments create real hope for breath-taking therapeutic approaches and improved patient outcomes.