The impact of concomitant medications on treatment outcomes in patients with cancer receiving immune checkpoint inhibitors

Abstract

During anticancer therapy, patients with cancer are often prescribed medications to combat concomitant health conditions and ameliorate cancer-associated side effects. Despite emerging evidence that many commonly prescribed medications have immunomodulating properties, surprisingly little is known about their interactions with immune checkpoint inhibitors (ICIs) in the treatment of cancer. This Review provides an overview of recent advances characterizing the reported impact of concomitant drug use on ICI-mediated therapeutic response and associated immune-related adverse events, and the potential to repurpose immunomodulatory drugs for other comorbidities to enhance ICI treatment efficacy.

Introduction

Cancer immunotherapy has transformed the treatment landscape for patients with a broad range of cancer types. Immune checkpoint inhibitors (ICIs) are a class of cancer immunotherapy which currently include 13 different FDA-approved monoclonal antibodies directed towards four targets: cytotoxic T lymphocyte-associated protein 4 (CTLA4) (ipilimumab, tremelimumab), programmed cell death protein 1 (PD1) (nivolumab, pembrolizumab, cemiplimab, dostarlimab, toripalimab, retifanlimab, tislelizumab), programmed cell death ligand 1 (PDL1) (atezolizumab, avelumab, durvalumab) and lymphocyte activation gene 3 (LAG3) (relatlimab)^1–9. Antitumour immunity elicited by ICIs is predominantly mediated by potentiation of T cell activity and reduction of immunosuppression in the tumour microenvironment (TME), leading to the elimination or control of cancer cells. A broad number of cell types and molecular mechanisms highlighting the complexity of the TME, and the impact on ICI treatment efficacy, have been extensively reviewed elsewhere^10–13.

Despite the success of ICIs in clinical trials, therapeutic resistance remains a considerable challenge for many patients¹⁴. Additionally, due to the non-specific nature of immune activation triggered by ICIs, patients are susceptible to ICI-mediated toxicities known as immune-related adverse events (irAEs)¹⁵. These irAEs can impact multiple organ systems and contribute to severe complications, risk of hospitalization, discontinuation of potentially cancer-curing treatments and even mortality^15–17. Although biomarker studies have identified predictors of response (tumour mutational burden^18,19, PDL1 status^20,21) and toxicity (human leukocyte antigen (HLA) type^22,23, autoantibody positivity²⁴), there is a growing need to understand modifiable exposures for patients with cancer that could improve the safety and efficacy of ICI treatment.

Many medications for concomitant health conditions display both direct and indirect immunomodulatory properties²⁵. However, formally testing the impact of concomitant drug use on ICI treatment is more difficult than traditional pharmacodynamic and pharmacokinetic drug–drug interactions, due in part to the underappreciation of these immune-related changes, and further compounded by the complexity of the clinical conditions for which they are used. Clinical trials also have strict exclusion criteria, including patient omission based on comorbidities and concomitant medication use, which may also limit current understanding in this area, due to selection of potentially non-representative patient cohorts^26–28. As ICI use becomes more inclusive in real-world settings, an improved understanding of the potential risks associated with concomitant medication exposure during ICI treatment is needed^28–30.

In this Review, we outline emerging evidence related to potential positive and negative interactions of commonly prescribed medications and supplements with ICI treatment (Supplementary Tables 1 and 2). We also highlight the ability of these medications to shape the TME, and the reported impact on survival outcomes of patients receiving ICI treatment. In distinction from other reviews, we provide a comprehensive perspective on a broad range of medications and their emerging effects on ICI treatment across cancer types, integrating both preclinical and clinical findings. Finally, we discuss the advantages and disadvantages of different resources for patient cohort analysis and strategize how both clinical annotations and preclinical models could be improved for greater utility in understanding ICI-related drug interactions that may lead to differences in treatment outcomes for therapeutic response and irAEs. We envision that this understanding may identify opportunities to leverage the immunomodulatory properties of concomitant medications to optimize outcomes for patients with cancer and prevent toxicity, particularly in real-world settings.

Impact of concomitant immunosuppression

More than 40% of patients with concomitant autoimmune diseases and the majority of patients with a history of solid organ transplantation are noted to be receiving immunosuppressive therapy (for example, corticosteroids, cytokine inhibitors, hydroxychloroquine and Janus kinase (JAK) inhibitors) at the start of ICI treatment³¹. Immunosuppressants have been shown to enhance cancer risk, likely through reduction of immunosurveillance enabling tumour progression^32,33. However, the impact of immunosuppressant drugs in the context of potent immune stimulation by ICI treatment remains unclear and has become an area of increasing interest with their expanding use for ICI-induced irAEs. Although multiple immunosuppressants may play a role in altering ICI efficacy^34–38, in this section we will focus on the impact of corticosteroids, given their broad clinical utility.

Corticosteroids

Corticosteroids are routinely administered to suppress an inflammatory immune response either locally, when given topically or via inhalation, or systemically. Initial studies suggested that there was no negative impact on response or survival outcomes for patients receiving ICI treatment requiring systemic corticosteroids to treat irAEs, which allayed concerns for their deployment in clinical care^39,40. However, with increasing patient numbers exposed to concurrent corticosteroid use alongside ICI treatment there has been an opportunity to parse out differences based on the timing, dose and reason for corticosteroid administration, leading to diversity in patient outcomes.

The majority of studies suggest that patients exposed to baseline systemic corticosteroid use of ≥10 mg prednisone equivalent daily, a dose frequently used for exclusion criteria in immunotherapy clinical trials^5,8,9, have worse outcomes from ICI treatment (Table 1 and Supplementary Table 1). In a study by Arbour et al.⁴¹, patients with non-small cell lung cancer (NSCLC) treated with corticosteroids (≥10 mg prednisone equivalent daily) at the initiation of anti-PD1 or anti-PDL1 treatment displayed decreased overall response rate (ORR), progression-free survival (PFS) and overall survival (OS). Common indications for corticosteroid use in this retrospective clinical cohort included dyspnoea, fatigue and brain metastases, which may have the potential to influence outcomes. Notably, the significant association between corticosteroid use and reduced ICI efficacy remained even after adjusting for covariates such as performance status and history of brain metastasis⁴¹. Additionally, in the pan-cancer Resistance Drivers for Immuno-Oncology Patients Interrogated by Harmonized Molecular Datasets (RADIOHEAD) cohort, systemic but not non-systemic corticosteroid exposure led to reduced real-world OS in immunotherapy-naive individuals receiving standard-of-care ICI treatment⁴². Consistent with these findings, early use of corticosteroids for irAE management, especially within the first 2 months of ICI treatment initiation, has also been increasingly associated with reduced treatment efficacy^43–45 (Supplementary Table 1).

Table 1 |.

Selected studies detailing associations between concomitant medications and survival outcomes in patients with cancer receiving immune checkpoint inhibitor treatment

Medication	Cancer type	ICI modality	Outcome	Refs.
Polypharmacy
Drug score	NSCLC (PDL1 ≥ 50%)	Anti-PD1	‘Poor’ drug score was significantly associated with reduced ORR, PFS and OS	255
Corticosteroids
Corticosteroids	NSCLC	Anti-PD1 or anti-PDL1	Baseline corticosteroids were associated with decreased ORR, PFS and OS	41
Corticosteroids	NSCLC	Anti-PD1 or anti-PDL1 ± anti-CTLA4	Baseline corticosteroid use was associated with shorter median PFS and OS	47
Corticosteroids	Pan-cancer	Anti-PD1 or anti-PDL1 ± anti-CTLA4 ±other cancer therapies	Baseline systemic corticosteroid use was associated with reduced real-world OS	42
Analgesics
Opioids	Pan-cancer	Anti-PD1 or anti-PDL1	Opioid use was associated with significantly worse PFS and OS	61
Acetaminophen	Pan-cancer	Anti-PD1 or anti-PDL1 ± anti-CTLA4	Detectable plasma acetaminophen was associated with worse OS and PFS	83
Antibiotics and PPIs
Antibiotics	Pan-cancer	Anti-PD1 or anti-PDL1	Antibiotics were associated with worse PFS and OS	118
Antibiotics	Melanoma	Anti-PD1	Antibiotic exposure in the 3 months prior to starting anti-PD1 therapy was not associated with OS	120
Antibiotics	NSCLC (PDL1 ≥ 50%)	Anti-PD1 or anti-PDL1	Antibiotics were associated with worse OS and PFS	122
Antibiotics, PPIs	NSCLC	Anti-PDL1	Antibiotics were associated with worse OS; PPIs were associated with worse OS and PFS	127
PPIs	Urothelial carcinoma	Anti-PDL1	PPI use was associated with worse OS with atezolizumab, but not chemotherapy	141
PPIs	NSCLC (PDL1 ≥ 50%)	Anti-PD1 or anti-PDL1 + chemotherapy	PPI use was associated with shorter PFS in the pembrolizumab monotherapy cohort compared with combination chemotherapy	142
Supplements
Probiotics, dietary fibre	Melanoma	Anti-PD1 ± anti-CTLA4, anti-CTLA4	Patients who reported sufficient dietary fibre intake had improved PFD; patients who reported probiotic usage had significantly worse PFD	116
Vitamin D	Pan-cancer	Anti-PD1, anti-PDL1, anti-CTLA4	Low vitamin D and VDR gene signature was associated with ICI resistance and more rapid tumour progression	117
Vitamin D	Pan-cancer	Anti-PD1 or anti-PDL1 ± anti-CTLA4	Vitamin D repletion showed longer OS, longer time to treatment failure and higher disease control rate	162
Fibre and omega-3 fatty acids	Melanoma	Anti-PD1 and anti-CTLA4	Poor response was associated with lower fibre and omega-3 fatty acid consumption	166,167
Cardioprotective and metabolic drugs
Statins	MPM, NSCLC	Anti-PD1	Statin use was associated with improved PFS and OS in MPM, and better PFS in NSCLC; high-intensity statins were associated with improved OS	175
RAASi	Pan-cancer	Anti-PD1 or anti-PDL1, anti-CTLA4, anti-PD1 or anti-PDL1 + anti-CTLA4	RAASi use was associated with improved OS, and there was a trend suggesting this was most pronounced in patients with genitourinary and gastrointestinal cancer	184
Beta-blockers	Pan-cancer	Anti-PD1 or anti-PDL1, anti-CTLA4, anti-PD1 or anti-PDL1 + anti-CTLA4	Beta-blockers were significantly associated with improved disease control; OS was significantly improved in the urothelial carcinoma cohort	208
≥3 medications
Multiple	Pan-cancer	Anti-PD1 or anti-PDL1	Aspirin, beta-blockers and statins were independently related to improved ORR; exposure to cancer-related corticosteroids, prophylactic antibiotics, gastric acid suppressants, PPIs, anticoagulants and opioids had significantly higher risk of disease progression and mortality	60
Antihistamines, aspirin, antibiotics	NSCLC and melanoma	Anti-PD1 or anti-PDL1	H1-Antihistamine and aspirin use were associated with a reduced death rate, whereas antibiotic use displayed a higher death rate	225
Multiple	NSCLC	Anti-PD1	Antibiotics were an independent predictor of poor OS and PFS; PPIs were associated with shorter OS, and corticosteroids were associated with worse PFS	92

Selected studies were chosen to represent a range of key findings. Studies were included based on relevance to the main text. Supplementary Table 1 provides additional references. CTLA4, cytotoxic T lymphocyte-associated protein 4; ICI, immune checkpoint inhibitor; MPM, malignant pleural mesothelioma; NSCLC, non-small cell lung cancer; ORR, overall response rate; OS, overall survival; PFD, progression-free disease; PFS, progression-free survival; PPI, proton pump inhibitor; RAASi, renin–angiotensin–aldosterone system inhibitors; VDR, vitamin D receptor.

In patients with cancer exposed to baseline corticosteroids, there may be nuance relating to both the dose and indication for use when considering their influence on ICI efficacy. A post hoc analysis of six clinical trials indicated that a higher corticosteroid peak dose is associated with worse survival across tumour types, whereas a cumulative dose does not have the same association⁴⁶. Furthermore, a retrospective study of patients with NSCLC receiving anti-PD1 therapy displayed significantly shorter median PFS and median OS in patients exposed to high-dose corticosteroids (≥10 mg prednisone equivalent daily) at the initiation of ICI treatment compared with patients treated with no or lower corticosteroid doses⁴⁷. Patients receiving high-dose corticosteroids for cancer-related indications (such as metastatic site, dyspnoea and anorexia) were shown to perform worse than patients receiving equivalent dosages for cancer-unrelated indications, including autoimmune or inflammatory conditions⁴⁷. Furthermore, in patients with cancer and pre-existing active rheumatoid arthritis, McCarter et al.⁴⁸ found that 24% of patients received corticosteroids (median dose 10 mg daily) prior to ICI treatment, and this did not appreciably impact either rheumatoid arthritis flares or OS. Although systemic corticosteroid use was not shown to influence recurrence of rheumatoid arthritis, there is evidence that prior corticosteroid exposure, within 3 months of initiating ICI treatment, is associated with greater risk of irAEs, and may portend a smouldering inflammatory immune response that can be triggered by ICI treatment⁴⁹ (Table 2). Future research is therefore needed to clarify distinctions in dosing, timing and indication for use, to inform clinical guidelines for the optimal management of corticosteroids when using ICI treatment.

Table 2 |.

Selected studies detailing associations between concomitant medications and immune-related adverse events in patients with cancer receiving immune checkpoint inhibitor treatment

Medication	Cancer type	ICI modality	irAE	Outcome	Refs.
Analgesics
Aspirin	Melanoma	Anti-PD1 and anti-CTLA4	Any, grade ≥3 (particularly hypophysitis and colitis)	Patients experienced an increased incidence of severe irAEs with concomitant aspirin compared with previous studies	96,97
Aspirin	Pan-cancer	Anti-PD1 or anti-PDL1, anti-CTLA4, anti-LAG3, anti-PD1 + anti-LAG3	Any	Aspirin use was associated with an increased risk of irAEs	100
Antibiotics and PPIs
Antibiotics	Pan-cancer	Anti-PD1 or anti-PDL1	Any	Antibiotic use is associated with greater irAE risk in patients with cancer	128
Antibiotics	Pan-cancer	Anti-PD1 or anti-PDL1, anti-CTLA4, anti-PD1 or anti-PDL1 + anti-CTLA4	Diarrhoea and/or colitis	Antibiotic use at any time was associated with reduced ICI-induced diarrhoea and/or colitis occurrence and recurrence rates, but exposure also increased diarrhoea and colitis severity	129
PPIs	Pan-cancer	Anti-PD1 or anti-PDL1, anti-CTLA4, anti-PD1 or anti-PDL1 + anti-CTLA4	AKI	PPI use was associated with a higher risk of ICI-induced AKI	155
PPIs	Pan-cancer	Anti-PD1 or anti-PDL1, anti-CTLA4, anti-PD1 or anti-PDL1 + anti-CTLA4	AKI	PPI use at baseline was associated with ICI-induced AKI lasting 3 days or more	156
PPIs	RCC	Anti-PD1	Any	Severe irAEs (grade ≥3) were more common among PPI users	160
Supplements
Vitamin D	Pan-cancer	Anti-PD1 or anti-PDL1 ± anti-CTLA4	Any	Vitamin D supplementation increased the incidence of irAEs (any grade), but decreased the risk of thyroid irAEs	162
Vitamin D	Melanoma	Anti-PD1 or anti-PDL1, anti-CTLA4, anti-PD1 or anti-PDL1 + anti-CTLA4	Colitis, grade ≥2	Vitamin D use (supplement or therapeutic administration) was associated with reduced ICI-induced colitis	165
Cardioprotective and metabolic drugs
Statins, corticosteroids	NSCLC	Anti-PD1 or anti-PDL1	Any	Statins and previous corticosteroid therapy increased the risk of irAEs	49
Statins	Pan-cancer	Anti-PD1 or anti-PDL1, anti-CTLA4, anti-PD1 or anti-PDL1 + anti-CTLA4	Skeletal myopathy	Statin use was found to be associated with increased risk of skeletal myopathy	179
GLP1RAs	Pan-cancer	Anti-PD1 or anti-PDL1, anti-CTLA4	Cardiovascular, any	Use of GLP1RAs was associated with fewer incidences of major cardiovascular events, but similar incidences of other irAEs	204
≥3 medications
Multiple	Pan-cancer	Anti-PD1 or anti-PDL1, anti-CTLA4, anti-PD1 or anti-PDL1 + anti-CTLA4 + other cancer therapies	Any	Use of multiple medications, PPIs and corticosteroids was associated with reduced incidence of irAEs overall and endocrine irAEs specifically; PPI use increased the incidence of gastrointestinal irAEs	157
Multiple	Pan-cancer	Anti-PD1 or anti-PDL1, anti-CTLA4, anti-PD1 or anti-PDL1 + anti-CTLA4	AKI	Use of ACEi, ARBs or corticosteroids was associated with AKI	187

Selected studies were chosen to represent a range of key findings. Studies were included based on relevance to main text. Supplementary Table 2 provides additional references. ACEi, angiotensin-converting enzyme inhibitors; AKI, acute kidney injury; ARB, angiotensin receptor blocker; CTLA4, cytotoxic T lymphocyte-associated protein 4; GLP1RA, glucagon-like peptide-1 receptor agonist; ICI, immune checkpoint inhibitor; irAE, immune-related adverse event; LAG3, lymphocyte activation gene 3; NSCLC, non-small cell lung cancer; PPI, proton pump inhibitor; RCC, renal cell carcinoma.

In preclinical mouse studies, concurrent corticosteroid administration with ICIs, but not delayed use, led to impaired antitumour immunity, consistent with clinical observations⁵⁰ (Supplementary Table 3). Similarly, preclinical studies using MC38 colon adenocarcinoma in autoimmune-prone C57BL/6-lpr mice, which are susceptible to multi-organ irAEs, identified that prednisone could limit toxicity but also attenuated ICI-mediated antitumour immunity^51,52. The impact of corticosteroids was shown to be most pronounced in low-affinity CD8⁺ T cells, due to the restriction of essential metabolic processes, such as fatty acid oxidation⁵⁰. Alternatively, an enzyme involved in glucocorticoid activation, 11-β-hydroxysteroid dehydrogenase-1 (11-β-HSD1), expressed by multiple cell types in the melanoma TME, was shown to provide a negative feedback mechanism in response to anti-PD1 treatment in vivo⁵³. Inhibition of 11-β-HSD1 in mouse melanoma improved ICI response through enhanced interferon-γ (IFNγ) production by CD8⁺ T cells, highlighting the diverse mechanisms by which corticosteroids may influence ICI treatment.

These studies indicate potential complications associated with concomitant use of corticosteroids or other forms of immunosuppressants and highlight a need to identify alternate strategies to enhance ICI efficacy while also balancing the autoimmune or autoinflammatory response. Notably, two recent early-phase clinical trials have demonstrated that JAK inhibitors, such as itacinib and ruxolitinib, can enhance anti-PD1 efficacy in metastatic NSCLC (NCT03425006)⁵⁴ and relapsed or refractory Hodgkin’s lymphoma (NCT03681561)^36,37,55. Traditionally used to treat autoimmune diseases, JAK inhibitors may offer a viable corticosteroid-sparing measure for consideration in both the concomitant setting for pre-existing conditions and to treat irAEs.

Analgesics influence immunity

Patients with cancer often experience pain due to underlying chronic illnesses, side effects from cancer treatment or cancer itself. Analgesic use follows a tiered approach in which non-steroidal anti-inflammatory drugs (NSAIDs) and acetaminophen are commonly used for mild pain, with escalation to opioids for greater pain severity^56,57. Although some forms of pain medication seem innocuous due to over-the-counter (OTC) accessibility, they may have the potential to play potent roles in modulating inflammatory responses (Fig. 1). Understanding the mechanisms by which analgesics modify systemic and tumour-specific immunity and exploring alternate forms of pain management will be critical in defining new opportunities to optimize ICI treatment.

Fig. 1 |. Mechanistic interactions between analgesics and the tumour microenvironment.

Proposed preclinical mechanisms by which analgesic medications sculpt the tumour microenvironment (TME). a, Aspirin inhibits cyclooxygenase (COX)-mediated prostaglandin E₂ (PGE₂) production, which enhances type I interferon production that is restricted by PGE₂ exposure and inhibits PGE₂-mediated production of cancer-promoting cytokines (interleukin-6 (IL-6), CXC-chemokine ligand 1 (CXCL1) and granulocyte colony-stimulating factor (G-CSF)) by dendritic cells (DCs) and other cell types, to promote T cell-dependent tumour control⁸⁸. b, Exogenous morphine administration increases PD1 expression in tumour-infiltrating CD8⁺ T cells, and decreases expression of pro-inflammatory cytokines, including IL-2 and interferon-γ (IFNγ) through μ-opioid receptor agonism⁷². c, Acetaminophen use in combination with anti-PD1 therapy is correlated with enhanced neutrophil extracellular trap (NET) formation⁸⁵. d, Acetaminophen use has also been associated with upregulation of IL-10 and expansion of LAG3⁺TIM3⁺ regulatory T cells (T_reg cells) in the periphery⁸³. e, Engagement of ^Δ9-tetrahydrocannabinol (THC) with cannabinoid receptor 2 (CNR2) impairs the function of tumour-specific T cells through inhibition of Janus kinase (JAK)–signal transducer and activator of transcription (STAT) signalling¹⁰⁵. LAG3, lymphocyte activation gene 3; TIM3, T cell immunoglobulin and mucin domain 3; TNF, tumour necrosis factor.

Opioids

Concomitant opioid use is associated with worse clinical outcomes (PFS, OS and ORR) in patients treated with ICIs across multiple cancer types (Table 1 and Supplementary Table 1). These effects appear to be dose-dependent, as separate cohorts of patients with NSCLC and patients with head and neck squamous cell carcinoma (HNSCC) displayed a relationship between higher opioid dose and worse outcomes^58,59. Interactions between opioids and worse outcomes were shown to persist even after adjusting for baseline characteristics such as platinum chemotherapy failure, neutrophil-to-lymphocyte ratio and PDL1 status⁵⁹. Although these studies reveal strong associations, the findings should be interpreted cautiously due to the potential for confounding relationships between opioid use, higher disease burden, aggressive cancer types, lower body mass index and poor performance status^58,60,61.

Opioids display pleiotropic roles towards the immune system by directly engaging immune cells^62,63, indirectly modulating the central nervous system through the neuro-immune axis⁶⁴ or disrupting gut-microbiome homeostasis^65–67 (Figs. 1 and 2). Opioids have also been associated with peripheral immunosuppression in patients with cancer treated for both perioperative^68,69 and cancer pain^70,71. Alterations in the TME have been observed, with high opioid use shown to correlate with significantly lower numbers of tumour-infiltrating CD8⁺ T cells, without reducing the presence of immunosuppressive regulatory T cells (T_reg cells)⁵⁹. Additionally, publicly available single-cell RNA-sequencing data indicated that tumour-infiltrating CD8⁺ T cells from patients with HNSCC displayed higher gene expression of μ-opioid receptor type 1 (OPRM1) compared with peripheral blood CD8⁺ T cells⁵⁹. Tumour-infiltrating OPRM1-expressing CD8⁺ T cells also displayed enrichment in immunoregulatory molecules, including PDCD1, CTLA4, T cell immunoreceptor with immunoglobulin and ITIM domain (TIGIT), LAG3 and ectonucleoside triphosphate diphosphohydrolase 1 (ENTPD1; also known as CD39)⁷². This highlights the potential of opioids for having a potent impact on the generation of a dysfunctional immune response in the TME.

Fig. 2 |. Medications and dietary supplements induce changes to the microbiome.

Emerging evidence indicates that medication and supplement-induced changes to microbial diversity can impact immune checkpoint inhibitor (ICI) response. a–c, Dietary factors and supplements that may promote microbial changes associated with improved response to ICIs. Interactions between vitamin D and the vitamin D receptor (VDR) on intestinal epithelial cells alter the gut microbiome in favour of the accumulation of Bacteroides fragilis; increased abundance of this species or its associated metabolites has been shown to enhance antitumour immunity¹¹⁷ (panel a). Omega-3 fatty acids can enhance intestinal production and secretion of short chain fatty acids (SCFAs) and intestinal alkaline phosphatase (IAP) in the gut, which may induce changes in microbial composition that enhance intestinal epithelial barrier integrity and decrease pro-inflammatory lipopolysaccharides (LPS)^166,257 (panel b). Similarly, dietary fibre enhances microbial diversity associated with increased SCFA production and the promotion of intestinal epithelial barrier integrity^116,258 (panel c). d–f, Proposed mechanisms by which commonly prescribed medications including proton pump inhibitors (PPIs), opioids and antibiotics may induce dysbiosis. PPIs inhibit parietal cell acid secretion by irreversibly inhibiting the H⁺/K⁺-ATPase; the resulting alkalinization of gastric pH enables potentially harmful microbes to survive and colonize the gastrointestinal microbiome^140,153 (panel d). Chronic opioid use has been linked to opioid receptor-mediated reduction in gastrointestinal motility, the coordinated movement of the digestive tract, which may lead to changes in the microbiome that disrupt intestinal epithelial barrier integrity and enable bacterial translocation and subsequent inflammation⁶⁷ (panel e). Antibiotics disrupt gut-microbiome homeostasis by reducing microbial diversity¹¹⁸ (panel f).

Beyond the immunomodulatory effects of opioids, opioid receptor upregulation has been reported in numerous solid tumour types, and in some studies this expression appears to correlate with cancer progression and metastasis^73–75. However, a large-scale pan-cancer analysis of The Cancer Genome Atlas (TCGA) identified differential expression of μ-opioid, κ-opioid and δ-opioid receptors across cancers, but found no association with patient OS or disease-free interval⁷⁶. Therefore, continued efforts to characterize the expression patterns of individual opioid receptors with specific cancer and treatment types are necessary. A preclinical study by Gondoh et al.⁷⁷ demonstrated that μ-opioid receptor agonism by the peripherally acting opioid loperamide increased Lewis lung cancer (LLC) tumour volume, whereas μ-opioid receptor antagonists methylnaltrexone and naldemine significantly reduced tumour burden, highlighting a reliance on opioid receptor signalling to promote tumour growth in this model.

Understanding the impact of opioid use in patients with cancer receiving ICIs may help establish new treatment options to overcome immunosuppressive effects without reducing the efficacy of pain management. A recent study by McIlvried et al.⁷² demonstrated that administration of the opioid morphine prior to anti-PD1 treatment reduced CD8⁺ and CD4⁺ T cell infiltration into the TME and diminished tumour control in an orthotopic model of oral squamous cell carcinoma (Fig. 1 and Supplementary Table 3). Use of an opioid receptor antagonist, axelopran, not only reversed the negative impact of morphine but led to a therapeutic advantage when combined with anti-PD1⁷². Naltrexone, another opioid receptor antagonist, is also being tested clinically with a combination of ICI treatment and the beta-blocker propanolol (NCT05968690)⁷⁸, indicative of the clinical interest in targeting opioid-mediated immunosuppression. Additionally, Scarpa et al.⁷⁹ identified a gene expression network implicated in triple-negative breast cancer pathogenesis and progression that was regulated in opposing directions by anti-PDL1 and opioids (Supplementary Table 3). This network, derived from publicly available RNA-sequencing data from tumour tissue of patients with triple-negative breast cancer and clinicopathological samples, was found to be associated with gene expression patterns derived from intratumoural CD8⁺ T cells. Morphine downregulated 54% of genes upregulated in the anti-PDL1 signature, indicating a shared, but opposing, nexus of interaction. Interestingly, ketamine, an emerging analgesic in cancer pain management⁸⁰, did not display overlap with this signature, suggesting that ketamine could provide a less immunosuppressive alternative to opioids, and may function without diminishing favourable molecular changes that could be associated with ICI-driven antitumour immunity. Taken together, these studies suggest potential benefits of limiting opioid use, and favouring lower opioid doses potentially combined with adjuvant co-analgesics such as ketamine or use of opioid receptor antagonists to reset the antitumour immune response in patients with cancer receiving ICIs.

Acetaminophen

Acetaminophen (also known as paracetamol) is a widely used OTC analgesic and antipyretic, which has historically been regarded as a non-immunomodulatory agent. However, evidence that the prophylactic use of acetaminophen limits vaccine responses in children^81,82 prompted interest in exploring its effect on additional immunomodulatory treatments, including response to ICI treatment (Table 1 and Supplementary Table 1). Retrospective studies that rely on medical use records have difficulty in discerning exposure to acetaminophen, given that it is primarily taken as needed and without prescription. Using mass spectrometry, the presence of acetaminophen or the downstream metabolite acetaminophen glucuronide in patient serum prior to nivolumab treatment was associated with significantly reduced OS in a cohort of patients with renal cell carcinoma (RCC), a finding that was also validated in patients with other cancer types receiving additional ICI agents⁸³.

The impact of acetaminophen on ICI treatment was also observed in patients with stage IIIA NSCLC enrolled in a prospective, phase II clinical trial involving neoadjuvant chemoimmunotherapy (NCT04422392)⁸⁴. Elevated plasma acetaminophen levels were associated with patients who failed to achieve a major pathologic response⁸⁵. The association between acetaminophen exposure and treatment response was particularly strong as age, smoking status, histological type or PDL1 status was not associated with major pathologic response by multivariate analysis⁸⁵. Additionally, the Eastern Cooperative Oncology Group (ECOG) performance status of these patients was uniform (between 0 and 1), indicative that it was unlikely to be due to differences in the severity of disease prior to treatment. The duration and dosage of acetaminophen may be pivotal for this effect. Nelli et al.⁸⁶ identified that high exposure, defined as therapeutic use lasting more than 24 h or more than 60 doses of 1,000 mg between 30 days prior to ICI initiation and 90 days after the first ICI infusion, was associated with shortened OS and PFS in a cohort of patients with NSCLC receiving ICI treatment either alone or in combination with platinum chemotherapy.

Mechanistically, patients with acetaminophen in circulation at the initiation of ICI treatment displayed significant upregulation of interleukin-10 (IL-10), often considered an immunosuppressive cytokine, and fms-related tyrosine kinase 3 ligand (FLT3L) 6 weeks after ICI treatment⁸³. In contrast, patients without acetaminophen exposure presented increased pro-inflammatory molecules, including CXC-chemokine ligand 11 (CXCL11), tumour necrosis factor (TNF), IL-12B, IL-15 receptor subunit-α (IL-15RA) and TNF receptor superfamily member 9 (TNFRSF9)⁸³. In preclinical tumour models, addition of acetaminophen alongside ICI treatment was associated with reduced MC38 rejection and accelerated LLC tumour growth^83,85 (Supplementary Table 3). Consistent with reduced tumour control, a significant increase in the frequency of immunosuppressive T_reg cells was identified in the TME⁸³ (Fig. 1). In the peripheral blood of healthy individuals, acetaminophen exposure for 24 h was shown to enrich expression of LAG3 and T cell immunoglobulin and mucin domain 3 (TIM3) in T_reg cells, indicative of potentially heightening their capacity for immunosuppression, along with increasing the abundance of arginase-expressing myeloid-derived suppressor cells and indoleamine 2,3-dioxygenase 1 (IDO1)-expressing plasmacytoid dendritic cells (DCs)⁸³. Additionally, acetaminophen may contribute to anti-PD1 resistance through enhanced neutrophil extracellular trap (NET) formation also observed in LLC tumours⁸⁵ (Fig. 1). These findings highlight the diversity of cellular and molecular changes initiated in both the periphery and the TME by acetaminophen, and the need to carefully approach analgesic use to minimize the potential impact on ICI treatment.

Non-steroidal anti-inflammatory drugs

NSAID use is highly prevalent among patients with cancer, not only for pain management but also for treating comorbidities, such as rheumatoid arthritis and coronary heart disease⁸⁷. The cyclooxygenase 1 (COX1) and COX2 pathways, inhibited by NSAIDs, have been implicated in promoting tumour growth, progression and immunosuppression^88,89. However, ICI-treated clinical cohorts have shown mixed responses in relation to clinical outcomes following NSAID exposure.

In a retrospective clinical cohort, patients with melanoma and NSCLC receiving first-line ICI treatment displayed significantly increased time to progression and ORR when exposed to a broad class of COX inhibitors (aspirin, NSAIDs or selective COX2 inhibitors) during the first two cycles of ICI treatment⁹⁰ (Supplementary Table 1). Notably, in the melanoma cohort, a high pretreatment neutrophil-to-lymphocyte ratio (>5) was associated with decreased time to progression, but this effect was negated in patients concurrently receiving COX inhibitors⁹⁰. Additionally, Cortellini et al.⁶⁰ identified that low-dose aspirin for prevention of cardiovascular events was associated with increased ORR and PFS in a pan-cancer cohort receiving anti-PD1 or anti-PDL1 treatment, although this effect was not observed with other NSAIDs (Table 1). This trend was further validated in a meta-analysis, where baseline aspirin was found to improve PFS; however, no impact on OS was evident, and other NSAIDs showed no significant survival benefit⁹¹. Additional retrospective studies have associated low-dose aspirin with improved survival, whereas general NSAID use has been linked to either neutral or slightly poorer outcomes (Supplementary Table 1). However, these findings may be cohort-specific, as no beneficial influence of aspirin was found in a retrospective cohort of patients with NSCLC with PDL1 expression ≥50% receiving first-line pembrolizumab monotherapy⁹².

Preclinical studies have shown that COX activity impedes antitumour immunity, leading to promising evidence that concomitant NSAID use could also improve ICI efficacy (Supplementary Table 3). Zelenay et al.⁸⁸ demonstrated that genetic ablation of COX signalling in BRAF-mutant melanoma cells promoted gene expression of Ifng, Il12b, Cxcl10, Tbx21 and Cd86, while reducing Il6 and Cxcl1 in the TME (Fig. 1). Consistent with this pro-inflammatory signature, multiple tumour types engineered to be deficient for COX signalling displayed improved tumour control, specifically in an immunocompetent mouse setting. This was mediated, at least in part, by type I interferon signalling, basic leucine zipper transcriptional factor ATF-like 3 (BATF3)-dependent DCs, major histocompatibility complex (MHC) class II^highCC-chemokine receptor 7 (CCR7)^negative DCs, natural killer cells as well as both CD4⁺ and CD8⁺ T cells^88,89,93. Heightened antitumour immunity observed in the absence of COX signalling led to the preclinical testing of NSAIDs, aspirin and the COX2 inhibitor celecoxib, in which they were all shown to combine efficiently with anti-PD1 treatment⁸⁸. Multiple COX inhibitors were also identified to limit macrophage-derived T cell suppression in a high-throughput screen while enhancing anti-PDL1-induced T cell proliferation⁹⁴. In particular, the NSAID indomethacin has also been demonstrated to improve anti-PDL1 efficacy in preclinical tumour models by promoting tumour-infiltrating CD8⁺ T cells, including those expressing T cell-specific transcription factor 1 (TCF1; also known as TCF7) or granzyme B⁹⁴. Some NSAIDs can also promote antitumour immunity through COX-independent mechanisms. For instance, diclofenac inhibits lactate transport by monocarboxylate transporter 1 (MCT1) and MCT4, which led to improved response to anti-PD1 in B16 mouse melanoma by impairing tumour glycolysis and reducing acidification of the TME⁹⁵.

The connection between aspirin use and improved ICI outcomes in retrospective and preclinical settings prompted a phase II clinical trial of pembrolizumab, ipilimumab and high-dose aspirin in advanced melanoma (NCT03396952)^96,97. Although there was no appreciable difference in survival outcomes compared with previous studies using a similar ICI regimen^98,99, those patients who were able to tolerate aspirin through the induction period displayed better survival outcomes⁹⁷. However, tolerability was a concern in this study as patients experienced more severe (grade ≥3) irAEs than previously observed for this ICI combination without aspirin, with hypophysitis and colitis being particularly prevalent (Table 2). It should be noted that there were possible limitations to this study including a small sample size, a lack of screening for COX enrichment in the TME and the concomitant use of proton pump inhibitors (PPIs) to alleviate potential aspirin-related toxicities, which may also have immunomodulatory effects⁹⁷. An analysis utilizing the FDA Adverse Event Reporting System (FAERS) database also found that aspirin use was linked to an increased risk of irAEs in a pan-cancer setting (Table 2). Elevation of tissue-specific irAEs, including anaemia, colitis, myocarditis, myositis, pancreatitis, pericarditis and pneumonia, and a reduced risk of rash, Stevens–Johnson syndrome and thyroiditis were observed¹⁰⁰. However, the FAERS is a voluntary, non-mandatory reporting system with inherent challenges including an absence of patient outcomes and the dose or duration of medication use; results should therefore be cautiously interpreted. Nonetheless, providers should be cognizant of potential enrichment for these tissue-specific toxicities when advising patients on aspirin usage.

Cannabis

Cannabis is increasingly used for medical indications, and has been legalized for recreational use in nearly half of the United States^101,102. Due to historical categorization as a Schedule I substance and accompanying restrictions on usage, both preclinical studies and clinical assessment of the use of cannabis have been limited¹⁰³. Despite this, cannabis and other cannabinoid-containing products, such as cannabidiol (CBD), are increasingly used for palliation of symptoms in patients with cancer. Several studies have demonstrated the efficacy of cannabis for cancer-related symptoms, such as chemotherapy-induced nausea and pain¹⁰¹. However, the lack of standardization of formulations and dosing has made medical use of cannabis by clinicians challenging¹⁰⁴. Furthermore, cannabis carries its own risks, including gastrointestinal, psychiatric and cognitive adverse effects¹⁰¹.

Preclinical data indicate that cannabis may reduce antitumour immunity and thereby impact ICI efficacy in the treatment of cancer. ^Δ9-Tetrahydrocannabinol (THC), which is the key psychoactive ingredient in cannabis, suppresses the frequency and function of tumour-infiltrating CD8⁺ T cells through cannabinoid receptor 2 (CNR2) engagement, which in turn has been shown to reduce MC38 and B16 tumour control and ICI efficacy in mice¹⁰⁵. CNR2 binds to JAK1 and inhibits downstream signalling of signal transducer and activator of transcription (STAT) in T cells, limiting the production of pro-inflammatory cytokines such as IFNγ¹⁰⁵ (Fig. 1 and Supplementary Table 3). The endogenous cannabinoid N-arachidonoylethanolamine (AEA), an analogue of THC and another selective agonist of CNR2, has also been shown to inhibit IFNγ and TNF production and proliferation of mouse CD8⁺ T cells both in vitro and ex vivo; accordingly, patients with lung cancer and high serum levels of AEA showed worse OS than patients with low levels of AEA, independent of treatment type¹⁰⁵. However, these results may not be consistent across tumour models, as a study by Waissengrin et al.¹⁰⁶ did not identify an association between THC exposure and reduced survival in CT26 colon carcinoma-bearing mice receiving anti-PD1 therapy.

Recent observational cohort studies have identified an association between the use of cannabis during ICI therapy across several cancer types and reduced efficacy (lower ORR, time to progression and/or OS)^107,108 (Table 1 and Supplementary Table 1). Two of these studies have come under scrutiny due to methodological concerns relating to the statistical analyses in which differences in baseline demographic and clinical variables were not originally identified¹⁰⁹. Furthermore, an observational cohort study of patients with NSCLC found no significant association between individuals with and without licences for cannabis use and ICI efficacy, despite a numerical reduction in OS in the cannabis group¹⁰⁶. Thus, current clinical evidence of the interaction between cannabis and ICI exposure is conflicting, and will require further evaluation with its growing use in both patients with cancer and the broader population. A systematic review for the Multinational Association of Supportive Care in Cancer (MASCC) guidelines found little evidence that cannabinoids are effective in the treatment of cancer-related pain and, based on the conflicting clinical data available, advises caution with the use of cannabinoids in patients receiving ICI treatment¹¹⁰.

Microbiome sculpting agents

Microbial diversity has been gaining strength as a prognostic indicator of both ICI response and toxicity^111–114. This connection has fuelled efforts to elevate the microbiome as a potential biomarker and therapeutic target^113,115. Several medications are known to reduce microbial diversity, including antibiotics and PPIs, whereas other dietary supplements can regulate the gut microbial composition, both of which can influence responses to ICI treatment^116–118 (Fig. 2).

Antibiotics

Antibiotics are commonly prescribed in patients with cancer to prevent and treat infections, especially prior to surgical resection of malignancy, for patients susceptible to chemotherapy-induced neutropenia and for patients with indwelling catheters. Multiple studies have linked antibiotics to reduced OS, PFS and ORR in patients receiving ICIs, either alone or in combination with chemotherapy, across various cancer types (Table 1 and Supplementary Table 1). These effects appear to be most pronounced within 30–60 days immediately prior to or following ICI administration^92,119–122, although negative interactions have been observed up to 1 year post exposure¹²¹.

The degree of impact may be dependent upon antibiotic class, as a study by Eng et al.¹²¹ found that patients treated with anti-PD1 who were aged 65 years or older and receiving fluoroquinolones — potent broad-spectrum antibiotics targeting aerobic, enteric gram-negative bacilli and common respiratory pathogens — had significantly worse outcomes compared with patients receiving no antibiotics or other broad-spectrum classes including penicillins and cephalosporins. This effect was associated with the timing of antibiotic exposure, in that any antibiotic use, defined by any exposure in the year prior to starting ICI therapy, was linked to worse OS. For fluoroquinolones specifically, this association was observed with exposure both within 60 days and the year prior to treatment, with an observed dose effect defined by weeks of cumulative exposure. Conversely, a sub-cohort analysis of patients treated with anti-CTLA4 identified that increased antibiotic exposure 1 year prior to initiating ICI therapy showed improved OS¹²¹. This suggests that different ICI agents, which have the potential to target different immune cell populations in both the TME and tissue, could be used dependent on the history of antibiotic exposure to maximize antitumour efficacy.

The detrimental effects of antibiotic use in patients with cancer appear specific to ICI treatment. A retrospective analysis of patients with NSCLC in the phase II POPLAR (NCT01903993)^123,124 and phase III OAK (NCT02008227)^125,126 trials found that baseline antibiotic use was associated with significantly worse OS in the atezolizumab-treated, but not docetaxel chemotherapy-treated, cohort¹²⁷. However, an important limitation of this analysis was that patients exposed to antibiotics had worse baseline ECOG performance. In a separate retrospective study of patients with NSCLC receiving first-line pembrolizumab monotherapy versus cytotoxic chemotherapy, systemic antibiotic use was associated with significantly reduced OS and PFS in the pembrolizumab cohort, but had no association with survival in the chemotherapy group⁹². Importantly, here exposure to antibiotics did not correlate with worse ECOG performance status. These findings may indicate a probable immunomodulatory mechanism by which antibiotics attenuate ICI efficacy. To clarify this relationship, future studies should stratify analyses by antibiotic class and indication in order to disentangle the role of antibiotics from underlying patient comorbidities.

Antibiotics have also been linked to enhanced risk of irAEs, in particular ICI-mediated colitis^128,129 (Table 2 and Supplementary Table 2). A retrospective study of a pan-cancer cohort of patients receiving ICI treatment observed that antibiotic use at any time with respect to ICI administration appeared to be protective against ICI-induced colitis onset and recurrence¹²⁹. However, of patients who developed colitis, antibiotic exposure after ICI initiation but before ICI-induced colitis, especially antibiotics with anaerobic activity, was associated with increased incidence of severe colitis, leading to more frequent and longer hospitalizations compared to patients with no antibiotic exposure or exposure preceding ICI treatment¹²⁹.

Preclinical studies have also identified specific contexts in which antibiotic repurposing can mitigate irAE occurrence without compromising ICI efficacy. For instance, clinical drug screening identified that clofazimine, an infrequently used antimycobacterial agent, potentiates the efficacy of combination anti-PD1 and anti-CTLA4 treatment by promoting TCF1 activation to increase the early memory and effector functions of CD8⁺ T cells, while simultaneously reducing irAE toxicity through suppression of pathogenic T helper 17 cells (T_H17 cells)¹³⁰. Antibiotics have also been demonstrated to reduce the dose-limiting toxicity of immune agonistic antibodies, anti-CD40 or anti-CD137, used in combination with anti-PD1 without impairing antitumour efficacy¹³¹. Antibiotic exposure led to differential gene expression related to TNF signalling, Toll-like receptor (TLR) cascades and interferon-stimulated gene pathways in the liver, a site susceptible to irAEs, 24 h following anti-CD40 or anti-CD137 treatment¹³¹. This indicates that in treatment settings where combinatorial partners are needed to augment a therapeutic response to ICI treatment, antibiotics could be considered to suppress irAEs without abrogating tumour control.

Faecal microbiome analyses in patients treated with ICIs suggest that increased species diversity correlates with improved ORR, although consensus on a specific microbial signature is still ongoing^112,113,132. Treatment with antibiotics reduces baseline microbial diversity and induces dysbiosis^133,134 (Fig. 2 and Supplementary Table 3). Mechanistically, Fidelle et al.¹³⁴ demonstrated that antibiotics cause recolonization of the gut with bacterial species of Enterocloster, which in turn can downregulate expression of the α4β7 integrin ligand mucosal addressin cell adhesion molecule 1 (MAdCAM1) in the ileum, observed in both patients and mice. This reduction in MAdCAM1 expression was found to induce the exodus of immunosuppressive regulatory T_H17 cells bearing α4β7 integrin from the gut and into tumours, thereby compromising cancer immunosurveillance and hindering ICI efficacy¹³⁴. Larger-scale clinical studies with longitudinal faecal microbial analysis paired with assessment of immune correlates will shed light on immunomodulatory effects of antibiotics, and whether the absence of a favourable microbial composition or rebound of an unfavourable one may limit ICI efficacy¹³⁵.

Comprehensive detailing of antibiotic type, dosage and duration is needed to clarify the interactions between ICIs and antibiotics in order to provide comprehensive guidelines regarding appropriate antibiotic use. Although prophylactic or concomitant antibiotics may be warranted for patients with cancer at risk of chemotherapy-induced neutropenia or diagnosed with an infection, unnecessary prophylactic antibiotic exposure in patients with cancer receiving ICI treatment may diminish efficacy and increase the risk of immunotoxicity. Providers may need to consider prior antibiotic exposure when determining the optimal timing and target for ICI treatment.

Proton pump inhibitors

PPIs are available by prescription or OTC and are used to relieve symptoms of gastro-oesophageal reflux disease and peptic ulcers by blocking H⁺/K⁺-ATPase on parietal cells in the stomach to prevent gastric acid secretion. PPI use is highly prevalent in patients with cancer¹³⁶, but can limit the bioavailability of some oral tyrosine kinase inhibitors¹³⁷ and chemotherapy¹³⁸, and thus have already been associated with worse clinical outcomes in some patient populations^92,139. Additionally, PPIs are frequently taken for extended periods of time without evidence-based indication and with no plan for discontinuation¹⁴⁰, and as such are receiving attention for their potential role in impairing ICI efficacy.

Retrospective and meta-analyses have linked concomitant PPI usage with diminished PFS, OS and ORR across various cancer types (Table 1 and Supplementary Table 1). A retrospective study by Hopkins et al.¹⁴¹ indicated that PPI use within 30 days preceding or following treatment administration was associated with significantly worse survival outcomes (OS and PFS) in patients with advanced urothelial carcinoma receiving ICI, which was not observed in patients receiving chemotherapy. The negative association between PPIs and ICI efficacy may be strongest in the context of ICI monotherapy and may not extend to individuals receiving ICIs in combination with chemotherapy. A retrospective study of patients with NSCLC receiving first-line ICI monotherapy, compared with those receiving ICIs with chemotherapy, found that PPIs were independently associated with shortened PFS in the monotherapy cohort, but not in the combination chemotherapy cohort¹⁴². However, this may be specific to certain chemotherapy regimens¹⁴³. It is possible that this association may be specific to certain cancer types, as PPIs have also been reported to have a neutral or potentially beneficial effect on patient survival^144–146. For instance, a post hoc analysis by Homicsko et al.¹⁴⁷ found no association between PPI use and survival outcomes (ORR, PFS and OS) in a pooled analysis of patients with melanoma receiving ICI treatment enrolled in the CheckMate 066, 067 and 069 clinical trials (NCT01721772¹⁴⁸, NCT01844505¹⁴⁹, NCT01927419¹⁵⁰). However, PPI use in patients treated with nivolumab and ipilimumab within the CheckMate 069 study was associated with reduced ORR and PFS and trended towards reduced OS. A notable limitation in pooling these studies was the geographical heterogeneity in study populations across clinical trials, as well as discrepancies in baseline characteristics including ECOG performance status, proportion of PDL1⁺ tumours and BRAF mutation status. Additional high-quality prospective studies will be necessary to provide clarification on the prognostic influence of PPI use in the setting of ICI treatment.

Emerging evidence suggests that PPIs display anti-commensal activity and influence gut microbial composition^{140,151–153} (Fig. 2). PPI-induced gastric alkalization leads to dysbiosis and increased risk of enteric infections by enabling bacterial survival and transit to the gut^140,151. Increased risk of Clostridium difficile infection, higher abundance of Streptococcaceae and Micrococcaceae, and reduced microbial diversity have also been observed following PPI use^140,152,153. These microbiome alterations have been linked to diminished ICI efficacy and increased toxicity in patients with melanoma receiving ICI treatment¹³². Interestingly, PPI-mediated alkalinization to reduce tumour acidity via daily intraperitoneal administration of the PPI esomeprazole to mice bearing B16 melanoma reduced TME acidity, restoring T cell function and improving tumour control¹⁵⁴. This suggests that in some settings, such as systemic rather than oral administration or in the absence of comorbidity, PPI use could be advantageous.

In a few studies, PPI use has been associated with an increased incidence and prolongation of ICI-induced acute kidney injury (AKI)^155,156 and risk of gastrointestinal irAEs^157,158 (Table 2 and Supplementary Table 2). A retrospective analysis of patients with RCC receiving nivolumab treatment in the GETUG-AFU 26 NIVOREN multicentre phase II trial (NCT03013335)¹⁵⁹ observed that although baseline PPI use was not associated with PFS or OS after adjusting for patient characteristics such as treatment line and pathological features, it did increase the incidence and severity of adverse events, including renal failure, fatigue and anaemia¹⁶⁰. A known side effect of PPIs, outside the setting of cancer, is acute interstitial nephritis. Among patients receiving ICIs, acute interstitial nephritis is one of the most common biopsy-proven AKI diagnoses¹⁵⁶. One potential mechanism for ICI-induced AKI involves PPIs acting as haptens bound to interstitial proteins in the renal tubules, which may provoke a T cell-mediated immune response upon initiation of ICI treatment¹⁴⁶. Although the mechanistic understanding of the correlation between gastrointestinal irAEs and PPI use remains incomplete, dysbiosis may promote a pro-inflammatory environment that increases susceptibility to irAEs¹⁶¹. Future studies are necessary to elucidate the immunomodulatory mechanisms and the specific microbial alterations by which PPIs may dampen ICI efficacy and promote tissue-specific irAEs in cancer-specific settings. Ultimately, clinicians should remain mindful of potential interactions and weigh the necessity of PPI therapy on an individual basis in patients receiving ICIs.

Vitamin D, probiotics and fibre

Beyond exploring the effects of medications that alter microbial diversity, recent studies have also emphasized the role of dietary habits and OTC supplements in modulating the gut microbiome and influencing ICI outcomes (Table 1 and Supplementary Table 1). In a prospective observational study, patients with advanced cancer receiving ICIs and with baseline vitamin D deficiency (hypovitaminosis considered as ≤30 ng ml⁻¹) who received vitamin D repletion with cholecalciferol had significantly improved OS as well as extended time to treatment failure and disease control rate compared with the non-repleted control cohort¹⁶².

Retrospective analysis of a pan-cancer cohort of more than 1,000 patients receiving ICI treatment demonstrated that increased tumour-derived expression of a vitamin D-induced gene signature, consisting of vitamin D receptor (VDR) target genes, correlated with improved response to ICI treatment¹¹⁷. A preclinically proposed mechanism for these findings is that vitamin D acts upon intestinal epithelial cells to shift the microbiota composition in favour of Bacteroides fragilis, which assists in potentiating cancer immunity¹¹⁷ (Fig. 2 and Supplementary Table 3). Vitamin D and VDR signalling have previously been recognized for their role at the gut–immune interface, which includes increasing species diversity and preserving tight junctions between intestinal epithelial cells^163,164. These actions collectively support both gut and systemic immune defences and may also underscore the benefits of vitamin D in antitumour immunity.

Recent studies have emphasized a link between vitamin D supplementation and the development of irAEs (Table 2 and Supplementary Table 2). Bersanelli et al.¹⁶² observed that the cumulative incidence of irAEs was increased when supplementing vitamin D in a prospective cohort, whereas the risk of thyroid irAEs was significantly decreased. Alternatively, Grover et al.¹⁶⁵ identified that vitamin D supplementation before starting ICI treatment was associated with significantly reduced odds of developing ICI-induced colitis in two distinct clinical cohorts. Multivariable logistic regression analysis also indicated that vitamin D supplementation outperformed other clinical covariates previously associated with colitis protection, such as pretreatment neutrophil-to-lymphocyte ratios¹⁶⁵. Large-scale studies will be necessary to more comprehensively define the relationship between vitamin D and outcomes in patients treated with ICIs.

Multiple studies have linked increased dietary fibre and omega-3 fatty acid intake with improved ICI outcomes^116,166 (Table 1 and Supplementary Table 1). In a prospective cohort of patients with melanoma receiving ICI treatment, Spencer et al.¹¹⁶ observed that after adjusting for other clinical factors such as age, sex, body mass index and prior antibiotic use, every reported 5 g increase in dietary fibre correlated with a 30% reduced risk of disease progression or death. Interestingly, probiotic supplementation appears to attenuate the benefits of increased dietary fibre, as patients who reported sufficient dietary fibre intake and probiotic usage had worse PFS compared with patients reporting sufficient dietary fibre but no probiotic consumption¹¹⁶. These results were recapitulated in preclinical models with a humanized microbiome from a complete responder, which found that those mice receiving a low-fibre or probiotic diet exhibited increased resistance to anti-PD1 therapy and a reduced frequency of IFNγ-producing CD8⁺ T cells within the TME¹¹⁶ (Supplementary Table 3). Similarly, Simpson et al.¹⁶⁶ performed faecal microbiome analysis of patients with melanoma receiving ICI treatment (NCT02977052)¹⁶⁷, and observed a correlation between omega-3 fatty acid and dietary fibre intake, alongside changes to the microbiome, with enhanced ICI response. These findings demonstrate the potential impact of diet and supplementation, and the purported benefits of incorporating gut health-promoting interventions for patients treated with ICIs.

Immunomodulatory cardioprotective drugs

Cardioprotective medications are frequently prescribed to patients with cancer to treat underlying comorbidities such as hypertension, hypercholesteraemia and diabetes. Emerging evidence suggests that cardioprotective and lipid-lowering medications that systemically modulate inflammation, including statins, angiotensin-converting enzyme inhibitors (ACEi), angiotensin II receptor blockers (ARBs), beta-blockers and metformin, can influence ICI efficacy (Table 1 and Supplementary Table 1).

Statins

Statins are commonly prescribed cholesterol-lowering agents that exert their effects through inhibition of the enzyme hydroxymethylglutaryl-coenzyme A (HMG-CoA) reductase. Multiple preclinical studies have demonstrated the ability of statins to improve ICI efficacy (Supplementary Table 3). In preclinical B16 melanoma and TC-1 human papillomavirus-associated tumour models, using simvastatin as an adjuvant in ovalbumin (OVA) vaccination prior to anti-PD1 administration induced a strong T_H1-type CD4⁺ T cell and cytotoxic CD8⁺ T cell response that enhanced tumour control¹⁶⁸. Likewise, in a preclinical model of oral carcinoma, combination therapy with anti-PD1 and simvastatin or lovastatin enhanced survival by promoting T cell activation and shifting macrophage populations from an immunosuppressive to pro-inflammatory state¹⁶⁹ (Fig. 3). Additional preclinical studies have proposed mechanisms by which statins may enhance antitumour immunity by modulating several key signalling pathways such as the MAPK and JUN amino-terminal kinase (JNK) pathways, which in turn has been shown to reduce PDL1 expression^170–173.

Fig. 3 |. Cardioprotective medications shape intratumoural hypoxia to promote antitumour immunity.

Representation of preclinically proposed mechanisms by which cardioprotective medications may induce beneficial changes to the tumour microenvironment (TME). a, Intratumoural hypoxia contributes to enhanced CD8⁺ T cell exhaustion and increased recruitment and activation of regulatory T cells (T_reg cells)¹⁸³. Preclinical studies have highlighted the role of both renin–angiotensin–aldosterone system inhibitors (RAASi) and metformin in reducing hypoxia and improving the antitumour T cell response^182,190. b, Activation of AMP-activated protein kinase (AMPK) by metformin induces endoplasmic reticulum (ER)-mediated PDL1 degradation in tumour cells, enhancing recognition for tumour cell killing by T cells¹⁸⁹. c, Statins have also been demonstrated to reduce PDL1 expression, through multiple mechanisms including inhibition of long non-coding RNA SNHG29, which facilitates Yes-associated protein (YAP) activation¹⁷⁰. d, Tumour-selective angiotensin receptor blockers (ARBs) reprogramme cancer-associated fibroblasts (CAFs) to a quiescent state¹⁸¹. e, Statins can promote antitumour immunity through promoting shifts in macrophage predominance from immunosuppressive to pro-inflammatory¹⁶⁹. CXCL13, CXC-chemokine ligand 13.

In a multicentre retrospective study of 390 patients with advanced NSCLC receiving anti-PD1 monotherapy, statins appeared to prolong OS but not PFS¹⁷⁴. Despite correlating with favourable outcomes, statins were not an independent prognostic factor after adjusting for clinical covariates such as PDL1 status and tumour histology (Table 1 and Supplementary Table 1). Other retrospective pan-cancer cohorts of patients receiving anti-PD1 and anti-PDL1 treatment have linked baseline statin use to improved overall response^60,175. However, these survival benefits have not been consistently observed in cohorts of patients with NSCLC^92,176 and metastatic urothelial carcinoma¹⁷⁷. These effects may be more pronounced according to the type of statin; for instance, Chiang et al.¹⁷⁸ observed that a pan-cancer cohort of patients treated with ICIs receiving lipophilic statins, such as atorvastatin and fluvastatin, had longer median OS and PFS compared with patients using hydrophilic statins or no statins. Additionally, a retrospective study observed that high-intensity statins, that is, statins that reduce low-density lipoprotein by 50% or more, were associated with improved OS, whereas low or moderate-intensity statins were not¹⁷⁵.

Statins have also been linked to the development of irAEs. In a large pan-cancer cohort, statin use was associated with an increased risk of skeletal myopathies¹⁷⁹ (Table 2 and Supplementary Table 2). Similarly, multivariable logistic regression analysis of a NSCLC cohort identified statin use as an independent predictor of the development of irAEs⁴⁹.

Renin–angiotensin–aldosterone system inhibitors

Renin–angiotensin–aldosterone system inhibitors (RAASi), including ACEi and ARBs, are widely used to treat hypertension, heart failure and diabetic nephropathy. Preclinical studies suggest that renin–angiotensin–aldosterone system activity may contribute to immunosuppression in the TME, through augmentation of pro-inflammatory tumour-associated macrophages¹⁸⁰ and exacerbation of tumour hypoxia¹⁸¹ (Supplementary Table 3). Selectively blocking intratumoural renin–angiotensin–aldosterone system signalling with low pH-activated nanoconjugate ARBs was shown to alleviate cancer-associated fibroblast (CAF)-mediated immunosuppression, increasing intratumoural CD8⁺ T cell to T_reg cell ratios, and favouring MHC class II⁺ pro-inflammatory macrophages over immunosuppressive CD206-expressing populations¹⁸¹ (Fig. 4). Renin–angiotensin–aldosterone system inhibition also reduces solid stress, a mechanical force compressing tumour blood vessels and leading to hypoxia and acidity in the TME which, in turn, may promote T_reg cell activity, increased expression of PDL1 and suppressed cytotoxic lymphocyte activity^182,183.

Fig. 4 |. Proposed mechanisms by which neuromodulatory medications augment antitumour immunity.

Medications that engage the neuro-immune axis have been observed to enhance immune checkpoint inhibitor (ICI) efficacy in the preclinical setting. Innervation by the sympathetic nervous system stimulates the release of catecholamines, such as adrenaline and norepinephrine, which then bind to β-adrenergic receptors. a, β1-Blockers antagonize β1-adrenergic receptors preventing catecholamine-induced progression of CD8⁺ T cell exhaustion²⁰⁵. b, Targeting monoamine oxidase A (MAOA) in exhausted CD8⁺ T cells with monoamine oxidase inhibitors (MAOIs) can improve anti-PD1 efficacy by modulating antigen-induced autocrine serotonin signalling to reduce T cell dysfunction²⁴⁸. c, The tricyclic antidepressant imipramine, as well as H₁-antihistamines, may induce immunosuppressive macrophage reprogramming via histamine H₁ receptor (H1R) inhibition to increase interleukin-1α (IL-1α) production alongside a decrease in IL-10 and arginase 1 (ARG1)²⁴⁹. d, Selective serotonin reuptake inhibitors (SSRIs) inhibit the serotonin transporter (SERT), which may enhance CD8⁺ T cell functionality through activation of an autocrine T cell serotonin–5-HT receptor (5HTR)–MAPK–T cell receptor (TCR) signalling pathway²⁵¹. e, Platelet-derived serotonin can contribute to intratumoural CD8⁺ T cell exhaustion²²¹. IFNγ, interferon-γ; LAG3, lymphocyte activation gene 3; MHC, major histocompatibility complex; NFAT, nuclear factor of activated T cells; NF-κB, nuclear factor-κB; TEX_eff cell, terminally differentiated T cell effector; TEX_prog cell, progenitor exhausted T cell; TEX_term cell, terminally exhausted T cell; TIM3, T cell immunoglobulin and mucin domain 3.

In a large, pan-cancer retrospective study, concurrent use of RAASi (ACEi or ARBs) was associated with significantly improved OS¹⁸⁴ (Table 1 and Supplementary Table 1). This effect was most pronounced in patients with gastrointestinal and genitourinary cancers, with a trend towards significance in these subcohorts¹⁸⁴. In patients with genitourinary cancer receiving ICI treatment, high doses of the ACEi lisinopril (daily dose > 20 mg) also trended towards improved survival compared with lower doses¹⁸⁴. A pan-cancer meta-analysis showed consistency with these findings, indicating that patients receiving ICIs in combination with RAASi had significantly improved OS, especially in patients with urothelial carcinoma and RCC¹⁸⁵. However, a smaller study by Cortellini et al.⁶⁰ found no significant survival benefit of baseline RAASi, although gastrointestinal and genitourinary cancers where survival benefit appears to be greatest were under-represented in this cohort. RAASi have also been implicated as a risk factor for AKI during ICI therapy¹⁸⁶. However, a retrospective study by Meraz-Muñoz et al.¹⁸⁷ found this association only in univariable analysis, with no significant effect after adjusting for other clinical covariates, such as hypertension (Table 2 and Supplementary Table 2).

Metformin

Metformin, a widely used and effective medication for type 2 diabetes, has garnered significant attention for its potential to promote antitumour immunity. As a biguanide, metformin decreases insulin sensitivity and hepatic gluconeogenesis through AMP-activated protein kinase (AMPK)-dependent and AMPK-independent mechanisms that are not completely understood¹⁸⁸. Nonetheless, metformin-induced activation of AMPK in the TME has been demonstrated to enhance anti-CTLA4 efficacy by triggering PDL1 degradation in the endoplasmic reticulum (ER) in 4T1 breast cancer mouse models, thereby increasing the recruitment of granzyme B-producing and IFNγ-producing intratumoural CD8⁺ T cells¹⁸⁹ (Fig. 3 and Supplementary Table 3). Additionally, in vivo studies of B16 melanoma and MC38 colon cancer indicate that metformin reduces intratumoural hypoxia, enhancing the efficacy of anti-PD1 by reversing hypoxia-mediated suppression, leading to increased CD8⁺ T cell proliferation and cytokine production¹⁹⁰. Improved tumour control was particularly apparent in the ICI-resistant B16 melanoma model with B16 melanoma cells shown to have a higher rate of oxygen consumption than MC38 tumour cells in vitro¹⁹⁰.

Retrospective studies in pan-cancer cohorts have yielded conflicting results regarding the potential survival benefits of concomitant metformin use with ICI therapy (Table 1 and Supplementary Table 1). This may be due to discrepancies in the dosage and duration of metformin exposure. For example, Ciccarese et al.¹⁹¹ observed a significant association between high doses of metformin (>1,000 mg day⁻¹) and improved PFS and OS. There are also multiple ongoing and completed phase I and phase II clinical trials investigating metformin in combination with ICIs in patients with melanoma, colon cancer, HNSCC and lung cancer (NCT03618654¹⁹², NCT04414540¹⁹³, NCT03311308¹⁹⁴, NCT03048500¹⁹⁵ and NCT03800602¹⁹⁶). Results of a phase II trial of nivolumab and metformin in patients with metastatic colon cancer (NCT03800602)¹⁹⁶ indicate that although combination therapy with metformin does not seem to have an appreciable influence on efficacy, it is well tolerated and associated with increased numbers of tumour-infiltrating lymphocytes in the majority of patients¹⁹⁷.

Glucagon-like peptide-1 receptor agonists

As new drugs for type 2 diabetes continue to emerge, such as the increasing use of glucagon-like peptide-1 receptor agonists (GLP1RAs)¹⁹⁸, it will be essential to follow their impact not only on cancer but also on ICI efficacy. Use of GLP1RAs in individuals with type 2 diabetes has been associated with reduced incidence of obesity-associated cancers compared with individuals receiving insulin¹⁹⁹. However, ICI treatment is linked with stronger immune responses and increased survival benefit in preclinical models and patient populations with obesity^200–202. Given the use of GLP1RAs for weight loss, it will be important to assess this paradox in treatment efficacy in patients receiving these drugs concurrently. The glucagon-like peptide-1 receptor (GLP1R) has also been identified to be expressed in T cells enriched with exhaustion markers, including PD1, with GLP1R antagonism shown to both improve tumour control and increase the frequency of T cells, particularly inducing activation of CD8⁺ effector memory T cells, in the TME²⁰³. These insights demonstrate a need for immediate efforts to determine the impact of GLP1RA use in the context of ICI treatment, including preclinical modelling in settings that are complementary to their clinical use.

Interestingly, cardioprotective effects that have been observed in individuals receiving GLP1RAs may also be maintained in patients with cancer receiving ICI treatment. In a propensity score-matched cohort of patients with cancer and type 2 diabetes receiving a GLP1RA both prior to and following ICI treatment initiation compared with an alternate anti-hyperglycaemic medication, those receiving GLP1RAs displayed reduced risk of major adverse cardiovascular events, with similar rates of other organ-specific irAEs²⁰⁴. This was particularly pronounced when comparing GLP1RA use with dipeptidyl peptidase-4 inhibitors, which also increase GLP1R signalling, highlighting a potential advantage of the GLP1RAs. All-cause mortality was also reduced in patients receiving GLP1RAs; however, it is unclear whether this was due to their impact on cancer survival and response to treatment or related comorbidities, including cardiac dysfunction and type 2 diabetes.

Beta-blockers

β-Adrenergic receptor blockers, also known as beta-blockers, treat a wide variety of cardiovascular and non-cardiovascular comorbidities, ranging from myocardial infarction to hypertension as well as anxiety and tremors. The prognostic relevance of beta-blockers towards shaping outcomes for patients treated with ICIs is receiving increasing attention, particularly given preclinical studies highlighting the interaction between β-adrenergic signalling and T cell exhaustion^205,206; however, the results of clinical studies have been varied. This is possibly due to the range of clinical conditions for which beta-blockers are used as well as differences in β-adrenergic receptor selectivity of the drugs (Table 1 and Supplementary Table 1). A retrospective study aimed at investigating the influence of cardiovascular comorbidities on ICI treatment outcomes observed that exposure to beta-blockers prior to ICI treatment for patients with lung cancer was associated with significantly worse OS, including after adjusting for underlying comorbidities such as hypertension, heart failure and history of stroke²⁰⁷. Conversely, a recent retrospective study consisting of a pan-cancer cohort of patients receiving ICI treatment observed that concomitant use of beta-blockers, especially cardioselective beta-blockers such as atenolol and bisoprolol, was associated with a significantly improved disease control rate²⁰⁸ (Table 1). Similarly, Cortellini et al.⁶⁰ demonstrated that concurrent beta-blocker use was an independent predictor of increased ORR. A phase I clinical trial (NCT03384836)^209,210 displayed positive results with good tolerability, and multiple clinical trials testing beta-blockers in combination with ICI treatment are now ongoing across solid tumour types (NCT05741164²¹¹, NCT05968690⁷⁸, NCT05451043²¹², NCT05651594²¹³, NCT05961761²¹⁴, NCT04848519²¹⁵). The majority of preclinical studies investigating the influence of beta-blockers on ICI efficacy have been evaluated in the context of stress, inducing differences in the immune response mediated by adrenergic signalling, which will be discussed in detail within the next section.

Altering the neuro-immune axis

Neuro-immune cross-talk is a key regulator of cancer progression^{205,216–218}. Tumour growth can be modulated by the expression of neuropeptide and neurotransmitter receptors, such as dopamine^219,220, serotonin^221–224, histamine²²⁵ and β-adrenergic receptors²¹⁶, either directly in cancer cells or in other cells in the TME. Depression, psychological stress and other mood disorders are highly prevalent in patients with cancer^226,227, and have been linked to poor ICI outcomes in both clinical^228–230 and preclinical settings²¹⁷. As the global health burden for both mental health concerns and cancer continues to rise, identifying the appropriate selection of therapies will be critical for optimizing ICI treatment in patients with cancer.

Immune dysregulation from emotional distress

As molecular biomarkers fail to sufficiently predict ICI outcomes and are often unable to be biologically altered or therapeutically targeted, approaches to quantify modifiable environmental, behavioural and psychological factors that influence patient outcomes have emerged. Patients with melanoma and NSCLC experiencing heightened emotional distress, which encompasses stress, anxiety and depression, were shown to have reduced survival outcomes following ICI treatment^228–230. Patients who report heightened emotional distress display elevated levels of stress-related serum markers such as cortisol²²⁹. Additionally, in patients with melanoma receiving neoadjuvant ICI treatment, there was no difference in prognostic markers such as the tumour mutational burden, BRAF mutation status, IFNγ signature or tumour-infiltrating immune cells, between patients with and without emotional distress²²⁸. This suggests that the observed differences in survival outcomes may be attributed to stress-induced alterations in the periphery or TME, which may lead to differential molecular and cellular changes that are not typically associated with poor treatment outcomes in patients with cancer receiving ICI treatment.

Preclinical studies have demonstrated that stress can drive tumorigenesis^231,232, progression²³³ and metastasis^218,234 through activation of the sympathetic nervous system or hypothalamic–pituitary–adrenal axis, leading to upregulation of β-adrenergic^205,216,235 and glucocorticoid²³⁶ signalling pathways^237,238. Dysregulation from signalling through these pathways has been linked to a reduction in the number and effector function of key mediators of antitumour immunity, including CD8⁺ T cells^{235,236,239,240}, DCs²¹⁷ and natural killer cells²¹⁸, while simultaneously expanding immunosuppressive compartments, such as myeloid-derived suppressor cells^241,242 and T_reg cells²⁴³. CD8⁺ T cells may be particularly vulnerable to the impact of stress hormones, due to their upregulation of the β1-adrenergic receptor, which when engaged leads to increased exhaustion and decreased cytokine production²⁰⁵. Yang et al.²¹⁷ demonstrated that stress elevated plasma corticosterone leading to increased expression of glucocorticoid-inducible factor TSC22D3, which in turn abrogates the type I interferon response in DCs, thereby preventing CD8⁺ T cell activation and hampering the efficacy of anti-PD1 therapy. Importantly, preclinical models indicate that utilizing pharmacological interventions such as beta-blockers or glucocorticoid antagonists to disrupt the stress response can improve ICI efficacy^205,217 (Fig. 4 and Supplementary Table 3). Although many of these studies highlight the influence of stress on the TME, emerging evidence suggests that this relationship may be bidirectional, with cancer also inducing neuroinflammatory changes that underscore psychological pathology^{238,244–247}. These studies highlight the potential biological impact of stress on antitumour immunity, and a growing need to clinically address emotional distress in order to improve outcomes for patients with cancer receiving ICI treatment.

Repurposing antidepressants and anxiolytics to improve ICI treatment

Medications used for mental health disorders, including antidepressants and anxiolytics, have shown potent therapeutic efficacy in combination with ICIs in preclinical models^{221,248–250} (Fig. 4 and Supplementary Table 3). Selection of these medications may also have the added benefit of addressing patients’ psychological comorbidities, which may contribute to ICI resistance. For instance, two preclinical studies demonstrated that genetic ablation or pharmacological blockade of the intratumoural expression of monoamine oxidase A (MAOA), an enzyme that breaks down neurotransmitters such as serotonin, dopamine and norepinephrine specifically in the immune cell compartment, could restrict MC38 and B16-OVA tumour growth^248,250. MAOA expression in tumour-infiltrating CD8⁺ T cells was also shown to limit the accumulation of proliferating effector-like and terminally exhausted cells, correlating with reduced granzyme B and IFNγ production. This was also reflected in patients with melanoma receiving anti-PD1 treatment, in which high intratumoural MAOA gene expression was able to negate the advantage gained by cytotoxic T lymphocyte enrichment observed in low MAOA gene expression settings, indicative of a correlation with T cell dysfunction²⁴⁸. Administration of phenelzine, an FDA-approved monoamine oxidase inhibitor (MAOI) used in the treatment of depression, provided improved tumour control in combination with anti-PD1 in B16-OVA melanoma by restoring T cell function through increased autocrine serotonin secretion via inhibition of MAOA-mediated serotonin degradation, thereby inducing heightened activation of CD8⁺ T cells characterized by increased IL-2 and IFNγ production²⁴⁸. Furthermore, Li et al.²⁵¹ demonstrated that inhibition of the serotonin transporter (SERT) with selective serotonin reuptake inhibitors (SSRIs) enhanced ICI efficacy in melanoma and colon cancer models, by activating CD8⁺ T cell responses via upregulation of an autocrine T cell serotonin–5-HT receptor (5HTR)–MAPK–T cell receptor (TCR) signalling pathway, and increasing CD8⁺ T cell production of effector molecules, including IL-2, IFNγ, TNF and granzyme B. Conversely, the restriction of serotonin signalling has also been demonstrated to enhance antitumour immunity. Schneider et al.²²¹ identified that platelet-derived peripheral serotonin reduces the infiltration and functionality of CD8⁺ T cells into both MC38 colon and Panc02 pancreatic tumour models. Tumour-derived PDL1 expression is also enriched in the presence of serotonin signalling, which may contribute to tolerizing CD8⁺ T cell activity. Pharmacological depletion of serotonin with the SSRI fluoxetine augmented the efficacy of anti-PD1 by enhancing intratumoural accumulation of CD8⁺ T cells, promoting durable tumour control²²¹.

The tricyclic antidepressant imipramine has also emerged as a potential combinatorial agent in the treatment of glioblastoma due to its demonstrated ability to enhance cancer cell autophagy²⁴⁹. When used in combination with anti-vascular endothelial growth factor (anti-VEGF) therapy, imipramine was shown to enhance autophagy-induced recruitment and proliferation of CD8⁺ and CD4⁺ T cells, and reprogramme tumour-associated immunosuppressive macrophages via inhibition of histamine H₁ receptor (H1R) signalling to sensitize glioblastoma to anti-PD1 therapy²⁴⁹. Interrogation of TCGA and electronic health record data indicated that patients with glioblastoma with low intratumoural expression of H1R or patients with reported antihistamine use, which may mimic the effects of imipramine, had improved OS²⁴⁹. These findings were recapitulated by Li et al.²²⁵, who demonstrated that concomitant use of H₁-antihistamines (which selectively target H1R) in patients with melanoma and lung cancer receiving treatment with anti-PD1 and anti-PDL1 was associated with improved OS. H₁-Antihistamines were also shown to enhance ICI efficacy in a B16 melanoma model engineered to produce granulocyte–macrophage colony-stimulating factor (GM-CSF) through inhibition of histamine binding to H1R in macrophages, leading to immunosuppressive macrophage reprogramming and a reduction in CD8⁺ T cell dysfunction²²⁵. These studies highlight the connection between modulation of the neurotransmitter histamine and immune cell function within the context of ICI-mediated antitumour immunity.

Although not in the setting of ICI treatment, evidence that individual neuroimmunomodulatory drugs within the same therapeutic class, such as benzodiazepines, a type of anxiolytic, may differentially shape the TME also highlights the need for greater selectivity when considering optimal treatment regimens for clinical use²⁵².

Future directions

Understanding the complex interplay between drug interactions is a challenge not unique to the ICI treatment setting. However, the multitude of factors that can modulate the immune response following exposure to concomitant medications creates difficulty in discerning the cellular and molecular mechanisms that underlie these changes and whether they cause direct or indirect effects on ICI treatment response and immunotoxicity. Additionally, the clinical history of patients with cancer is often highly diverse, especially relating to the timeline for concomitant drug exposure, the influence of inflammation from comorbidities and the potential for multiple drug interactions that could also impact the ICI response. To further advance understanding, we provide a series of suggestions to improve studies that aim to delineate the impact of drug interactions in ICI treatment and propose strategies to integrate these findings towards guidance in clinical practice.

Improving clinical studies

Clinical studies to assess the interactions between medications and ICI response often face limitations, including observational design, small sample size and heterogeneity in cancer types and ICI treatments. Many of these studies are retrospective, and may be biased by both measured and unmeasured confounding factors such as advanced disease and poor performance status as well as causes of competing mortality such as the underlying comorbidities for which these medications are prescribed²⁵³. Nonetheless, these studies play a crucial role in generating hypotheses, establishing platforms and the rationale for future clinical testing, as well as the identification of patient groups requiring specialized treatment to overcome deleterious effects from concomitant medications.

Large-scale prospective cohort studies and randomized clinical trials will further assist in identifying the strength of associations between drug interactions and ICI treatment outcomes. In particular, randomized clinical trials may be necessary before medications that have proposed improved efficacy in preclinical studies and retrospective cohorts can be deployed as a combinatorial partner for ICI treatment, with early-phase clinical trials already underway in some cases^97,197,210. For negative associations, prospective cohorts that are generalizable for drug exposures in the broader patient population will be particularly powerful⁴². Although these prospective approaches will provide clarity relating to the influence of individual drugs on ICI efficacy in a rigorous manner, they may be prohibitive in terms of cost and time, and may suffer from reduced patient enrolment compared with trials assessing novel agents.

For medication exposures that do not necessitate exclusion from randomized clinical trials, ad hoc assessment of medication use provides an opportunity to review outcomes, as outlined by Homicsko et al.¹⁴⁷ Additionally, existing datasets such as the Veterans Affairs Corporate Data Warehouse (CDW), TriNetX or MarketScan Research Databases offer opportunities to perform large-scale analysis of these trends^43,204. Using the longitudinal nature of these databases, it will be important to integrate information about the duration of and indication for medication exposure to further define conditions that influence ICI treatment. The impact of concomitant medications may be more pronounced in patients receiving ICIs alone rather than in combinations with chemotherapy, which may underscore the immunomodulatory nature of these interactions^142,254. Future studies should stratify analyses by immunotherapy type to provide a clearer picture of these interactions, and to define whether ICI regimens either alone or in combination display different outcomes. Ultimately, shifting towards pragmatic clinical trials with inclusive eligibility representative of real-world patients with multiple comorbidities and concomitant medication exposure will be critical for improving the evaluation of ICI interventions in the future.

One major limitation for many of these studies is that they often focus on a single agent or drug class. However, patients with cancer are often at risk for polypharmacy, highlighting the need for efforts that account for potential interactions and patterns of co-prescription using advanced statistical modelling²⁵⁵ (Box 1). Incorporating machine learning algorithms to enhance predictive modelling and simplify the complexity of multiple concomitant medication exposures will help determine their ability to shape ICI outcomes. This will be necessary to incorporate demographic information alongside medication exposures that delineate risks associated with ICI treatment, provide improved treatment options as both cancer therapies and concomitant medications continue to advance, and assist in drug counselling and consolidation for patients. These efforts underscore the goal of developing frameworks to guide treatment decisions based on medication exposure for patients with cancer.

Box 1 |. Interactions between concomitant medications should be evaluated within the context of polypharmacy.

Patients with cancer are especially at risk for polypharmacy, defined as exposure to five or more individual drugs. Polypharmacy often arises due to the accumulation of multiple symptoms and comorbidities over time for which multiple medications are prescribed²⁵⁹. Polypharmacy has been attributed to increased rates of hospitalization, treatment-related toxicities and poor survival outcomes in both patients with cancer and the general population^260,261. The risk of pharmacokinetic drug–drug interactions increases as a function of the number of medications prescribed, underscoring the need to evaluate medication toxicity not only individually but also within the context of specific medication regimens²⁶². Although guidelines have been established to mitigate polypharmacy in older populations, these methods can be difficult to apply to patients with cancer, as they do not account for the unique risk–benefit considerations of active cancer treatment and supportive care. However, emerging evidence indicates that concomitant use of multiple medications may be associated with worse survival outcomes for patients treated with immune checkpoint inhibitors (ICIs)²⁵⁵, and several studies have highlighted that a tailored reduction in the burden of polypharmacy in patients with cancer can improve quality of life and reduce adverse events²⁶³. However, the influence of polypharmacy in patients with cancer receiving ICI treatment is underexplored, and it remains unclear whether individual types of medications or their cumulative effect influence outcomes. Assessing the impact of polypharmacy, as well as that of individual medications within the context of ICI therapy, may be crucial to optimize patient outcomes for ICI treatment. Given the complexity of polypharmacy, machine learning algorithms may help enhance these research efforts.

Integrating clinical datasets with molecular information

An inherent difficulty in assessing the influence of concomitant medications towards ICI treatment is the lack of precise dosage and adherence data, particularly for OTC drugs or medications used as needed. Combining clinical records with the analysis of drug presence in patient samples from the periphery or tumour may assist to provide a more robust quantification for treatment exposure^83,85. Longitudinal cohorts that have both comprehensive data relating to medication use and ICI outcomes with sampling aligned at complementary time points will assist in building drug metabolite profiles and will help determine whether presence alone or a drug concentration threshold is necessary to impact ICI treatment⁴². These types of cohorts will also be strategically positioned to identify drug-induced signatures for ICI treatment resistance, which may assist in defining biomarkers for therapeutic response and rational treatment strategies to improve clinical care^72,79. Although the influence of concomitant drug exposures is often undermined by potential associations with confounding factors that may already represent risk for poor prognosis to ICI treatment, this still portends a patient group that requires more specialized treatment. It will be important to understand whether concomitant medication exposure drives differences in tumour development or reduces immune responsiveness to ICI treatment, and whether this can be reversed by discontinuing the concomitant medication or requires additional or alternative lines of treatment for comorbidity and cancer.

Developing multidimensional preclinical systems

Repurposing therapeutic strategies for other indications to augment antitumour immunity and ICI efficacy has become popularized. However, the benefit observed in the preclinical setting is often not reflected when comparing the impact of drug exposures in clinical cohorts. Greater emphasis in developing preclinical systems that are inclusive of multiple comorbidities will be essential for understanding these differing responses. Efforts are increasing to perform multidisciplinary studies that are inclusive of simultaneous immune-mediated conditions, which has stemmed from the need to define interactions between antitumour immunity and immunotoxicity in response to ICI treatment²⁵⁶. This is particularly important when considering drugs that can be used in multiple settings, such as beta-blockers to alleviate stress responses, which have been shown to provide benefit in preclinical mouse studies, whereas in clinical studies where they are predominantly used as a cardioprotective agent they have been identified to potentially reduce ICI efficacy^205,207. Understanding the underlying mechanism of discordant therapeutic responses through sophisticated, multilayered preclinical modelling will assist in tailoring treatment options for patients with cancer.

Advancing clinical guidance

Data suggesting the potential risk for drug interactions to modify the trajectory of ICI treatment responses continues to rapidly grow, which leads to the challenge of how to responsibly utilize these findings to develop frameworks for clinical practice. Although it remains too early to make definitive recommendations for many treatments, in the case where drugs have been associated with poor survival outcomes and increased risk of ICI-induced irAEs it may be appropriate to further evaluate the need for concomitant medication and, when possible, identify potential therapeutic alternatives with fewer or unknown immunomodulating properties. It will be important to consider patient-specific factors, such as comorbidity status and OTC medication use, when evaluating potential drug interactions, while acknowledging the limitations of the available data during patient education. This will strengthen opportunities for shared decision-making in treatment choice and to increase awareness for potential off-target effects, particularly in the context of susceptibility to irAEs. This evolving effort will require continued collaboration and review of guidelines for treatment management from oncologists, pharmacists, tissue-specific specialists and researchers to overcome these challenges and optimize clinical care.

Conclusion

Concomitant medications display immunomodulating properties with the potential to disrupt or enhance ICI efficacy and immunotoxicity. This Review highlights the diversity of therapeutic responses that have been observed in clinical trials, retrospective studies and preclinical models. Unifying our understanding of drug interactions to reach a consensus for their impact will assist in establishing coordinated care strategies to counsel patients about the potential for drugs and supplements to impact treatment outcomes. Improving our mechanistic understanding of concomitant medications will also enable tailored combination strategies that not only improve the response to ICIs but also address patients’ underlying comorbidities.

Supplementary Material

The online version contains supplementary material available at https://doi.org/10.1038/s41568-025-00890-z.

Key points.

• Concomitant medications represent a modifiable exposure associated with influences on immune checkpoint inhibitor (ICI) response.

• These medications exert their influence through non-conventional drug interactions, due to either direct or indirect immunomodulation, that may be underappreciated in current clinical settings.

• Increasing evidence indicates that medications such as proton pump inhibitors, antibiotics, corticosteroids and opioids may have detrimental influences on ICI response.

• Conversely, preclinical studies have highlighted the potential for other medications and supplements such as vitamin D and beta-blockers to augment ICI efficacy.

• Large-scale prospective studies, randomized control trials and layered preclinical modelling approaches are needed to improve our understanding of these immunomodulatory drug interactions.

Glossary

Acute interstitial nephritis

A condition that leads to declining kidney function, characterized by inflammation of the renal interstitium, the area surrounding the kidney tubules.

Analgesic

A medication used for pain management, examples include acetaminophen, non-steroidal anti-inflammatory drugs (NSAIDs) and opioids.

Antihistamine

A medication that blocks histamine receptors, commonly used to treat allergies.

Antipyretic

A medication that reduces fever, often by decreasing the production of prostaglandins; examples include acetaminophen and non-steroidal anti-inflammatory drugs (NSAIDs).

Anxiolytics

Medications used to treat symptoms of anxiety.

Benzodiazepines

Medications with depressant properties often used to treat anxiety disorders and insomnia; examples include lorazepam and alprazolam.

Beta-blocker

A medication that inhibits β-adrenergic receptors, often used for cardiovascular conditions, or non-cardiovascular conditions such as anxiety and tremor. Examples include propranolol or atenolol.

Concomitant drug use

Use of medications taken either concurrently or in succession, in this case alongside immune checkpoint inhibitor (ICI) treatment, often for non-cancer-related purposes.

Corticosteroids

A class of hormones either endogenously or synthetically produced, which can combat inflammation.

Disease control rate

A measurement that assesses the percentage of disease control in a patient population, defined as complete response, partial response or stable disease, after therapeutic intervention.

Disease-free interval

The duration of time between the end of cancer treatment and the first sign of tumour recurrence.

Dysbiosis

A disruption in gut microbial homeostasis, characterized by loss of beneficial bacteria and a reduction in overall microbial diversity.

Dyspnoea

A shortness of breath.

Gastro-oesophageal reflux disease

A condition characterized by frequent flow of stomach acid into the oesophagus, leading to irritation and inflammation.

Haptens

Small molecules that are not immunogenic alone but can elicit an immune response when coupled to a carrier, such as a protein.

Hypophysitis

Inflammation of the pituitary gland that can lead to pituitary hormone deficiency.

Immune checkpoint inhibitors

(ICIs). FDA-approved monoclonal antibodies that target immunosuppressive proteins expressed on immune cells, such as PD1 or cytotoxic T lymphocyte-associated protein 4 (CTLA4), to enhance the antitumour immune response.

Immune-related adverse events

(irAEs). Toxicities initiated by immune checkpoint inhibitor (ICI) treatment that can originate from an autoimmune or autoinflammatory response, and occur in a broad range of organs.

Interleukin

A protein that regulates immune responses, and is able to modify survival, proliferation, activation and inhibition of immune cells in response to ICI treatment.

Ketamine

An FDA-approved anaesthetic also used to treat depression and pain management.

Major pathologic response

A clinical measurement used in clinical trials, often defined as the presence of 10% or fewer viable tumour cells in the primary tumour following treatment intervention.

Multivariable logistic regression

A type of statistical analysis that models the relationship between a binary outcome, such as survival versus death, and multiple independent predictor variables.

Neutropenia

A condition characterized by low levels of neutrophils in the bloodstream.

Neutrophil extracellular trap

(NET). A web-like structure released by neutrophils to trap and neutralize pathogens that can also be present in the context of cancer.

Opioids

Medications that inhibit central and peripheral opioid receptors, often used for analgesic purposes; examples include morphine and oxycodone.

Overall response rate

(ORR). The percentage of patients in a treatment group or clinical trial that experience a therapeutic response. This includes either partial or complete responses often determined using Response Evaluation Criteria in Solid Tumours (RECIST) criteria.

Overall survival

(OS). The amount of time a patient remains alive following treatment initiation or disease diagnosis.

Performance status

A clinical measurement of a patient’s ability to perform daily activities and need for care. The Eastern Cooperative Oncology group (ECOG) scale is often used to assess performance status.

Polypharmacy

The use of multiple individual medications, usually five or more, simultaneously.

Progression-free survival

(PFS). The amount of time a patient lives with disease following disease diagnosis or treatment initiation without the disease worsening or death.

Propensity score-matched cohort

An observational study design that uses propensity scores, which are estimated probabilities of group assignment based on observed characteristics, to generate comparable groups from non-randomized data.

Renin–angiotensin–aldosterone system

A critical regulator of blood pressure and electrolyte homeostasis. Dysregulation of this system is implicated in many cardiovascular and renal diseases.

Stevens–Johnson syndrome

A rare but severe skin reaction often triggered by medications such as antibiotics and non-steroidal anti-inflammatory drugs (NSAIDs).

Time to progression

The time from treatment initiation to disease progression.

Tricyclic antidepressant

A sub-class of antidepressants that inhibit the reuptake of neurotransmitters such as serotonin and norepinephrine; examples include amitriptyline and imipramine.