Metagenomics and Chemotherapy-Induced Nausea: A Roadmap for Future Research

Abstract

Uncontrolled chemotherapy-induced nausea and vomiting (CINV) can reduce patients’ quality of life and may result in premature discontinuation of chemotherapy. Although nausea and vomiting are commonly grouped together, research has shown that antiemetics are clinically effective against chemotherapy-induced vomiting (CIV) but less so against chemotherapy-induced nausea (CIN). Nausea remains a problem for up to 68% of patients who are prescribed guideline-consistent antiemetics. Despite the high prevalence of CIN, relatively little is known regarding its etiology independent of CIV. In this review paper, we summarize a metagenomics approach to the study and treatment of CIN with the goal of encouraging future research. Metagenomics focuses on genetic risk factors, encompassing both human (i.e., host) and gut microbial genetic variation. Little work to date has focused on metagenomics as a putative biological mechanism of CIN. Metagenomics has the potential to be a powerful tool in advancing scientific understanding of CIN by identifying new biological pathways and intervention targets. Investigation of metagenomics in the context of well-established demographic, clinical, and patient-reported risk factors may help to identify patients at risk and facilitate prevention and management of CIN.

Precis:

In this paper we propose a comprehensive, metagenomic approach to investigating genetic risk factors, encompassing both human (i.e., host) and gut microbial genetic variation. An overview of current knowledge about chemotherapy induced nausea and vomiting, including the neurobiology, treatment, and associated risk factors is described following a summary of genetic and gut microbial associations with chemotherapy induced nausea and other gastrointestinal toxicities.

Nausea is one of the most common and feared side effects of chemotherapy.¹ Up to 47% of patients receiving moderately- or highly-emetogenic chemotherapy report moderate to severe nausea.^2-5 Patients describe chemotherapy-induced nausea (CIN) as “terrible,” “constant,” and “like having a hangover and being strapped to a rollercoaster.”⁶ CIN is associated with significantly worse quality of life,^7,8 declines in nutritional status,⁹ and decreases in day-to-day functioning.¹⁰ Despite the high prevalence of CIN, up to 72% of physicians underestimate its incidence, particularly during the delayed phase (i.e., >24 hours after chemotherapy).¹¹ Notably, CIN is also associated with increased healthcare utilization^12-14 and greater cost of care.^15,16 The Centers for Medicare and Medicaid Services has identified nausea as a top 10 driver of potentially avoidable hospital admissions for patients with cancer.¹⁷ Consequently, CIN is a major clinical problem.^18,19

Despite the development of new anti-emetic agents that have improved CIV, many patients still experience CIN. For instance, 68% of patients report nausea after receiving guideline-consistent antiemetic prophylaxis.^2-5 While researchers have developed risk algorithms to tailor antiemetic prescribing approaches to CINV, current models have low predictive accuracy for CIN.^20-24 Therefore, new approaches are needed to tailor antiemetic regimens. Metagenomics may be a promising approach to addressing CIN. Together with well-known demographic, clinical, and patient-reported factors, metagenomic data has the potential to better identify patients at risk, gain new insights into the etiology of CIN, and determine new behavioral and pharmacologic targets of intervention.

In this paper we propose a comprehensive, metagenomic approach to investigating genetic risk factors, encompassing both human (i.e., host) and gut microbial genetic variation. We begin by providing an overview of current knowledge about CINV, including the neurobiology, treatment, and associated risk factors. We then summarize genetic and gut microbial associations with CIN and other gastrointestinal toxicities, as well as future suggestions for research. We conclude with potential clinical implications of a metagenomic understanding of CIN. The manuscript synthesizes a compelling rationale across a variety of different disciplines to make the case that this is an important area of research.

Neurobiology of Chemotherapy-Induced Nausea and Vomiting

There are two major, known pathways that affect CINV, the peripheral pathway that primarily affects acute CINV,^25-27 and the central pathway that primarily affects delayed CINV (Figure 1).^28,29 Acute CINV (i.e., within 24 hours of infusion) results when chemotherapeutic agents irritate enterochromaffin cells in the gastrointestinal (GI) tract, triggering release of serotonin (5-HT) signaling in the medulla in the brain via vagal nerve afferents.^25-27 Delayed CINV (i.e., within 2-5 days) is thought to occur when serotonin sensitizes the vagal nerve afferents to the neuropeptide substance P, which binds to neurokinin 1 (NK₁) receptors in the nucleus of the solitary tract, passing the signal via vagal afferent nerves to the chemoreceptor trigger zone and then on to the medulla.²⁸ Consequently, the 5-HT₃ antagonist class of antiemetics such as palonosetron, ondansetron, and granisetron are used to prevent acute CINV while the NK₁ antagonist class of antiemetics such as fosaprepitant target delayed CINV. Dopamine and inflammation also appear to play a role in CINV as demonstrated by the antiemetic properties of D₂ dopamine receptor antagonists, metoclopramide, olanzapine, and corticosteroids.³⁰ Knowledge about these two pathways has guided current approaches to CINV treatment.

Figure 1.

Current understanding of pathophysiology in chemotherapy-induced nausea²⁹

Current Treatment Approaches to CINV

Early identification of patients at risk can significantly reduce CINV through tailored antiemetic prophylaxis.³¹ High-risk patients benefit from stronger antiemetics such as netupitant-palenosetron plus dexamethasone or the addition of alprazolam, lorazepam, olanzapine, or metoclopramide to their antiemetic regimen.³² Currently, antiemetic medications are selected on the basis of the emetogenicity of the chemotherapy. A 5-HT₃ antagonist and dexamethasone are used in moderately-emetogenic regimens and an NK1 antagonist and olanzapine are added for highly-emetogenic regimens.³³ Patients experiencing CINV after the first cycle may receive additional antiemetics during subsequent cycles. However, this practice is problematic for several reasons. First, as described above, the burden of CIN is high. In addition, patients who experience CIN after their first infusion are three times more likely to experience it again after later infusions.³² CIN following the initial infusion can also cause later anticipatory nausea (i.e., a conditioned response which is typically unresponsive to antiemetics).^34,35 Good control of CIN starting at the first infusion reflects high-quality care, may reduce healthcare utilization and costs, and improves patient quality of life.³¹ For these reasons, there is high clinical interest in tailored antiemetic prophylaxis.^36,37 It is possible that high rates of CIN suggest that additional biologic pathways may not yet be identified, although it is possible it may share the same pathway as a primitive reflex.^2-5

Risk Prediction and Tailored Antiemetic Prophylaxis

A large body of literature has established that CIN is associated with demographic, clinical, and patient-reported factors. Demographic risk factors include younger age and female sex.^22,38-43 Clinical risk factors include greater chemotherapy emetogenicity and the absence of comorbidities.^22,38-43 Patient-reported risk factors include no alcohol use, less sleep the night before chemotherapy, a heavier meal before chemotherapy, greater anxiety, a history of morning sickness during pregnancy, a history of motion sickness, and greater expectations of nausea.^22,38-43

Several risk prediction algorithms have been developed based on these factors^{21-24,36,44,45} and risk can either be calculated manually^22,24 or entered into a web-based calculator.⁴⁶ However, the accuracy of existing risk algorithms based on clinical and patient-reported factors is fair to poor—50% specificity, 79% sensitivity, with area under the receiver operating characteristic (AUROC) 0.70 (acute CINV) and 0.75 (delayed CINV).^20-24 The efficacy of these algorithms in reducing acute and delayed CIN was evaluated in a Phase III randomized trial comparing risk-based antiemetic prophylaxis to physicians’ choice in breast cancer patients receiving highly-emetogenic chemotherapy.³¹ Complete control of CIN across all cycles was significantly higher in the intervention group compared to the control group (acute: 54% vs. 42%, delayed: 40% vs. 31%).³¹ However, absolute rates of acute and delayed CIN were still high in both groups. Additionally, risk algorithms had low predictive value as evidenced by the fact that 95% of patients deemed to be at low risk received more aggressive antiemetic prophylaxis during cycle 2.⁴⁷ The accuracy of the same algorithms was similar to chance in patients treated with moderately- or highly-emetogenic chemotherapy in an independent study in patients treated with guideline-consistent antiemetic prophylaxis.⁴⁸ Given that current risk prediction models demonstrate low predictive accuracy, further research is needed to develop more accurate models. One challenge is that current available models have been constructed from limited data (e.g., patient demographics) and may miss key factors associated with CIN, such as metagenomics. Although the microbiome is not collected or analyzed as part of current clinical care, there is a push towards germline genotype testing to predict toxicities and side effects of cancer. We anticipate as more data is published assessing genetic risk factors for toxicity and germline genotyping becomes less expensive, these metagenomic factors will become more routine in the context of cancer care. Adding genetic and microbiome data to existing algorithms and random clinical trials of risk-based prescribing antiemetics such as Clemons et al. could evaluate the clinical utility of these hybrid algorithms to determine if it improves accuracy.³¹

A Novel Metagenomic Model of Chemotherapy-Induced Nausea

Metagenomics focuses on genetic risk factors, encompassing both human (i.e., host) and gut microbial genetic variation. Genetic variation is likely an important contributor to CIN as evidenced by the high heritability of nausea (i.e., 58-70%) due to various causes (e.g., pregnancy, opioid use, motion sickness).^49-51 To date, only a limited number of candidate gene studies have been performed examining CIN.⁵² A more comprehensive approach is needed to identify additional genetic variants and novel biological pathways. In addition, the biology of nausea must be considered in the context of the complex cellular makeup of the human body. The 30 trillion human cells in the body coexist with approximately 40 trillion microbial cells existing in the gut/GI tract known collectively as the microbiome. Relationships between human (i.e., host) cells and the microbiome are only starting to be understood. Nevertheless, it is clear that the microbiome should be considered “our second genome.”⁵³ Consequently, a rigorous investigation of CIN risk and pathophysiology should consider both host and microbiome genomic factors (i.e., the metagenome) in addition to demographic, clinical, and patient-reported risk factors.

A three-pronged research approach is needed to determine the role of the gut microbiome in CIN. First, studies should determine associations of CIN with the abundance of gut microbial species known to secrete neurotransmitters involved in CINV pathogenesis (e.g., serotonin, dopamine). Second, because there may be as-yet-unidentified bacterial species that secrete these neurotransmitters, the relative abundance of microbial genes producing these neurotransmitters should be determined. Finally, the expression of these microbial genes should be evaluated in relation to CIN. Recent reviews of metagenomics research describes best practices and challenges in experimental designs and computational analyses.⁵⁴ Experimental challenges include ensuring an adequate sample size for statistical power and determining an appropriate control group with identical timing of sample collection. Analytical challenges in metagenomics research include quality control with contamination removal, choice of assembly algorithm, detecting false positives in gene prediction, contig grouping or binning, ensuring sensitivity of taxonomic classification, and tool choice for functional classification.^54,55

An important area of future research in the gut microbiome is exploring how microorganisms in the gut engage in reciprocal signaling with the brain, known as the microbiota-gut-brain axis. It is possible many of these same microbiota-gut-brain axis mechanisms are known to cause CIN.^56-59 Host genetic variation has been significantly associated with CIN in candidate gene studies. Studies have focused primarily on genes regulating blood-brain barrier permeability such as ABCB1 (e.g., rs1045642, rs2032582, and rs1128503 ^48,60-65); serotonin receptors targeted by the 5-HT₃ antagonist class of antiemetics such as HTR3B (e.g., rs45460698⁶⁶), HTR3C (e.g., K163N⁶⁷), and HTR3D (e.g., rs6443930⁶⁸); and 5-HT₃ antagonist metabolism such as CYP2D6 (e.g., ultra-rapid phenotype^37,69). Candidate gene studies such as these are based on a priori knowledge of the biological processes relevant to the phenotype of interest. Because current knowledge of the pathophysiology of CIN is incomplete, a candidate gene approach is likely to miss important variants. Genome-wide association studies (GWAS) provide a more comprehensive alternative. GWAS, while requiring large sample sizes, spans the entire genome in an unbiased way and are agnostic with regard to the state of knowledge relating to the biological processes being investigated. Thus, GWAS maximize the chance to uncover novel biology. Best practices have been established in GWAS to achieve replicable results⁷⁰ and GWAS findings are replicable.^71,72 GWAS discoveries have led to major advances in therapeutics for a large number of diseases including Crohn’s disease, rheumatoid arthritis, and macular degeneration.⁷³

The post-GWAS era has resulted in novel bioinformatic and analytic techniques to ascertain the biological and clinical meaning of loci identified through GWAS.⁷⁴ A metagenomic investigation of CIN should leverage these advances, such as pathway analysis. Individual single nucleotide polymorphisms (SNPs) exist in intricate reaction and interaction pathways by which genes carry out their functions.⁷⁵ By mapping GWAS output, it is possible to interrogate SNP pathways into extensive genomic and epigenomic data such as transcription factor binding sites, histone modifications, and expression quantitative trait loci (eQTL) data.⁷⁶ Pathway analysis can also characterize SNP functionality at the genomic region therefore reducing complexity while increasing interpretability of large-scale GWAS data.⁷⁷ Pathway analysis is agnostic and is particularly advantageous when investigating complex traits such as nausea, which likely results from numerous individual contributions of multiple variants in the same gene and multiple genes within the same functional pathway. Pathway analysis can also be used to examine the combined influence of SNPs that do not reach GWAS-level significance (i.e., p<5x10⁻⁸). For these reasons, pathway analysis is a powerful approach to identifying potential mechanisms of CIN, as it allows SNPs and genes to be placed into their functional context.

Numerous studies have shown that bacteria normally present in the gut (e.g., Streptococcus, Escherichia, Lactococcus, Lactobacillus) are able to directly synthesize neurotransmitters (e.g., serotonin^78-83) and produce metabolites that stimulate neurotransmitter production by host cells.⁸⁴ Studies have also shown that gut microorganisms can activate the vagal nerve^85,86 to influence autonomic nerve system function and behavior.⁵⁷ Furthermore, a growing body of literature demonstrates the important role of gut bacteria in the development of chemotherapy-induced toxicities, including diarrhea, inflammation, and mucositis.^87-92 For example, pre-clinical studies suggest that chemotherapy results in decreases in commensal bacteria (e.g., Lactobacillus, Bifidobacterium, Bacteroides), increases in pathogenic bacteria (e.g., Clostridium cluster XI, Escherichia coli, Staphylococcus), and reduced microbial diversity.^93-95 However, data are currently lacking regarding whether these changes are associated with CIN.

Clinical Implications of the Metagenomic Model

Risk Prediction.

While sequencing the gut microbiome is not a current part of standard clinical cancer care, and there are several technical and cost-associated considerations regarding the implementation of cancer genomics assays into clinical practice, clinical genotyping is becoming increasingly common and used in clinical diagnostics as a means of identifying therapeutic options.⁹⁶ The goals of clinical germline genotyping in the context of cancer are largely to guide therapeutic dosing and identify risk of treatment toxicity (e.g., Steven-Johnson syndrome).^97-99 Genetic risk factors for CIN identified through GWAS and pathway analyses may increase the accuracy and clinical utility of existing risk prediction algorithms.^{21-24,36,44,45} Thus, the metagenomic model anticipates routine integration of genetic variation into risk prediction models, a process that will likely be facilitated by new developments in machine learning and their application to risk prediction and clinical decision support.¹⁰⁰ Currently, a “one-size-fits-all” approach is used for antimetic prescribing, however, accurate risk prediction algorithms could allow for a more tailored approach. Furthermore, such algorthims could guide clinical interventions in medical nutrition therapy.

Clinical Intervention.

Medical nutrition therapy, or special diets prescribed and monitored by a registered dietitian, may augment the efficacy of antiemetic agents to prevent or manage CIN in patients receiving moderately- or highly-emetogenic chemotherapy.¹⁰¹ Medical nutrition therapy is especially relevant considering patients with CIN are at risk for inadequate dietary intake and reduced quality of life.¹⁰¹ They are also at risk for malnutrition as a direct consequence of nausea/vomiting and behaviors, such as food avoidance.^102,103 Broad medical nutrition therapy recommendations include avoiding spicy, fatty, and sweet foods; eating slowly; consuming small, frequent meals as compared to three large meals during day-time hours; avoiding strong odors; participating in activities to distract from nausea (e.g., exercise or hobbies); and eating outside in fresh air.^101,104,105 Many of these strategies have not been evaluated through clinical trials; however, medical nutrition therapy may be efficious in reducing measures of CIN independent of CIV.¹⁰¹ Only one study has been conducted examining medical nutrition therapy for CINV, in breast cancer patients, compared to a usual-care control group. Between the two groups, patients in the control group reported more fatigue, nausea and vomiting, pain, constipation, and diarrhea (p < 0.001) compared to the intervention group who reported less nausea and higher quality of life (p < 0.001).¹⁰⁶ Thus, further research exploring companion metagenomics studies including clinical studies with specific interventions using medical nutrition therapy could provide further data regarding CIN mechanisms, toxicity and side effects to decrease CIN, prevent medical complications, and improve quality of life.

Strategies to remodel the gut microbiome should also be further evaluated in the context of CIN.^107,108 It is becoming increasingly evident that an individual’s unique microbiome may be critical in determining response to treatment and toxicities.^109,110 Advancements in DNA technologies have shown that chemotherapeutic agents detrimentally alter the composition of the microbiome, decreasing overall microbial diversity^87,109 and interrupting the microbiome-gut-brain axis.¹¹¹ A recent study examining microbiome alterations in gene expression in participants who experienced CIN found that chemotherapy administration was associated with expression of genes in mucosal inflammation and intestinal immune network pathways.¹¹² Therefore, it is possible mucosal inflammation and mucosal damage may lead to nausea, and interact with microbiome changes. Preclinical and clinical pilot studies have demonstrated that probiotic supplementation (e.g., Lactobacillus acidophilus, Streptococcus thermophilus) is associated with reduced gastrointestinal toxicities such as diarrhea,^113-115 although evidence is mixed.^109,116 Notably, data regarding use of probiotic supplementation to remodel the gut microbiome in the context of CIN are lacking. In addition, caution is warranted regarding probiotic supplementation during chemotherapy, as evidence suggests the gut microbiota may play a significant role in the efficacy and toxicity of chemotherapeutic agents by altering the drugs’ bioavailability and bioactivity.^9,114,117 Further, there is a theoretical risk of infection/sepsis with probiotic administration in individuals whose gut mucosal barrier has been compromised by chemotherapy.¹¹⁸ Additional pre-clinical and observational clinical studies are needed to determine which commensal bacterial strains can be supplemented without a detrimental effect on chemotherapy efficacy.

An alternative option to probiotic supplementation is dietary modification, which has been shown to be one of the most consistent and predictable methods of remodeling the gut microbiome.¹¹⁹ An example is anti-inflammatory diets (e.g., the Mediterranean diet), such as those that focus on increased consumption of unsaturated fats and fiber, with limited intake of red meat, processed foods, and sugar. In clinical studies, anti-inflammatory diets are associated with greater abundance of commensal bacterial species including Faecalibacterium and Bacteroides.^120,121 The extent of microbial change in response to dietary intervention appears to be highly variable and affected in part by pre-intervention gut microbial composition.^122,123 Individual variability makes gut microbial response difficult to predict, with corresponding variability in improvement in health outcomes.¹²² Nevertheless, anti-inflammatory diets have demonstrated a variety of health benefits in clinical studies including cardiovascular outcomes (e.g., risk of stroke, reduced blood pressure and cholesterol),¹²⁴ cognitive outcomes (i.e., mild cognitive impairment, Alzheimer’s disease),^125,126 and frailty^127,128 in non-cancer populations at higher risk of these outcomes. Findings must be interpreted with caution however, as the dietary intervention literature is beset with methodologic limitations such as different assessments of adherence and short duration of follow-up.

Literature examining anti-inflammatory diets and gastrointestinal toxicity in the context of chemotherapy is limited. A small, observational study found that patients with gynecologic cancer who scored high on the Mediterranean Diet Serving Score questionnaire reported less gastrointestinal toxicity during platinum-based chemotherapy.¹²⁹ Dietary interventions to restructure the microbiome in cancer patients should also take into account changes in the metabolome. Fatty acids, amino acids, minerals, vitamins, and antioxidants that may be modulated by diet are thought to modulate GI toxicities by preserving normal tissues by controlling immune responses, modulating the cytokine/hormone network, and modifying signaling events involved in cell proliferation and death.^130,131

An alternative dietary intervention that may improve CIN is fasting. In preclinical studies, short-term fasting (i.e., glucose restriction for up to 48 hours) has been shown to protect healthy cells against chemotherapy toxicity, but not cancer cells. This phenomenon is due to differential stress resistance, in which healthy cells are shielded from chemotherapy toxicity in times of stress (i.e., fasting), as opposed to cancer cells due to differences in growth factors and nutrient-sensing pathways.¹³² A case series of cancer patients, short-term fasting (i.e., 48-140 hours) peri-chemotherapy was associated with decreased fatigue, weakness, and gastrointestinal side effects. Fasting did not prevent chemotherapy-induced reduction of tumor volume or tumor markers, although some patients reported lightheadedness and dizziness.¹³³ Short-term fasting may not be appropriate for some patients, such as those with cachexia and risk of malnutrition.¹³⁴ Moreover, large randomized trials with close patient monitoring are needed before short-term fasting can be clinically recommended.

A more practical and tolerable form of fasting may be prolonged nightly fasting (PNF), in which the nightly fast is extended at least 12 hours and food intake occurs during the day.¹³⁵ PNF is hypothesized to improve CIN via differential stress resistance as well as favorable modulation of gut microbiota and corresponding metabolites.^112,136 PNF has been associated with selective enrichment of gut microbiota with Bactreroides and Lactobacillus in multiple studies.^137-139 In pre-clinical studies, supplementation of the gut microbiota with these bacterial species has been shown to decrease production of inflammatory cytokines and GI mucosal inflammation.¹⁴⁰ However, we are unaware of studies of PNF in humans to reduce chemotherapy-induced toxicities. Further research is needed to determine the safety and efficacy of PNF for this indication.

Conclusions

Although advances in anti-emetic prophylaxis have resulted in better control of CIV, the problem of CIN is still not resolved. In this paper, we described how metagenomic profiling may be used in combination with demographic, clinical, and behavioral risk factors to improve identification of patients at risk and provide personalized supportive care. There is a great deal of heterogeneity in the gut microbiome, especially among individuals with cancer. Additionally, anticancer drugs effect the gut microbiome, therefore, important considerations for interventions moving forward should include addressing baseline heterogeneity and differential changes in the microbiome. A greater understanding of the pathophysiology of CIN may also contribute to the development of targeted interventions. Through reducing CIN, novel interventions may improve quality of life, prevent dose reductions and treatment discontinuations, reduce healthcare utilization and costs, and improve quality of care.