Opioid medications: an emerging cancer risk factor?

Editor—In September 2020, consumption of opium was classified as ‘carcinogenic to humans’ by the International Agency for Research on Cancer (IARC), based on sufficient evidence for causing cancers in the lung, larynx, and bladder, and limited evidence for causing cancers in the oesophagus, pancreas, stomach, and pharynx.1IARC monographs on the identification of carcinogenic hazards to humans, volume 126. Opium consumption.IARC. 2021; 126: 1-253Google Scholar Opium is an addictive narcotic drug obtained from the poppy plant. The recognition of using opium as a carcinogen raises concerns about potential carcinogenicity of opioid medications, which are either derived from opium or synthesised to mimic its chemical structure and effects.2Pathan H. Williams J. Basic opioid pharmacology: an update.Br J Pain. 2012; 6: 11-16Crossref PubMed Google Scholar Based on their origin, opioid medications are categorised as natural opioids (opiates), which include opium and its derivatives (e.g. morphine, codeine); semi-synthetic opioids that are synthesised from natural opioids (e.g. hydromorphone, oxycodone); and synthetic opioids that do not contain opium but are synthesised to mimic its chemical structure and effects (e.g. fentanyl, methadone). The carcinogenic effects of opium have multiple potential mechanisms. Some of these are exclusive to opium; for example smoking opium exposes the user to polycyclic aromatic hydrocarbons (PAHs) and other carcinogens in opium smoke, which sometimes include adulterants and contaminants such as arsenic and lead.1IARC monographs on the identification of carcinogenic hazards to humans, volume 126. Opium consumption.IARC. 2021; 126: 1-253Google Scholar However, a recent study identified a specific mutational signature in oesophageal cancer tissues from opium users in Iran, which points to carcinogenic mechanisms beyond exposure to PAHs and other known carcinogens in opium smoke.3Moody S. Senkin S. Islam S.M.A. et al.Mutational signatures in esophageal squamous cell carcinoma from eight countries with varying incidence.Nat Genet. 2021; 53: 1553-1563Crossref PubMed Scopus (47) Google Scholar Other mechanisms, particularly the effects of opium alkaloids, are shared with natural opiates and semi-synthetic opioids. Opium alkaloids have shown genotoxic effects in experimental studies (chromosome damage, micronuclei, DNA fragmentation related to morphine, and sister chromatid exchanges related to codeine).4Li J.H. Lin L.F. Genetic toxicology of abused drugs: a brief review.Mutagenesis. 1998; 13: 557-565Crossref PubMed Scopus (51) Google Scholar They also impair the physiological functions of specific organs that can lead to increased exposure to carcinogens.5Kamangar F. Shakeri R. Malekzadeh R. Islami F. Opium use: an emerging risk factor for cancer?.Lancet Oncol. 2014; 15: e69-e77Abstract Full Text Full Text PDF PubMed Scopus (66) Google Scholar Finally, some potential mechanisms are related to activation of opioid receptors, and these are shared between natural, semi-synthetic, and synthetic opioids. Specifically, activation of opioid receptors can lead to the activation of angiogenesis and neovascularisation,6Grandhi R.K. Lee S. Abd-Elsayed A. Does opioid use cause angiogenesis and metastasis?.Pain Med. 2017; 18: 140-151Crossref PubMed Scopus (28) Google Scholar,7Kosciuczuk U. Knapp P. Lotowska-Cwiklewska A.M. Opioid-induced immunosuppression and carcinogenesis promotion theories create the newest trend in acute and chronic pain pharmacotherapy.Clinics (Sao Paulo). 2020; 75: e1554Crossref PubMed Scopus (10) Google Scholar impairment of immune function,7Kosciuczuk U. Knapp P. Lotowska-Cwiklewska A.M. Opioid-induced immunosuppression and carcinogenesis promotion theories create the newest trend in acute and chronic pain pharmacotherapy.Clinics (Sao Paulo). 2020; 75: e1554Crossref PubMed Scopus (10) Google Scholar and facilitation of tumour initiation, proliferation, and migration.6Grandhi R.K. Lee S. Abd-Elsayed A. Does opioid use cause angiogenesis and metastasis?.Pain Med. 2017; 18: 140-151Crossref PubMed Scopus (28) Google Scholar Evidence on opioid medications and cancer is primarily from ecologic and registry data linkage studies. Analyses of records from national health insurance or addiction registry programmes in Australia,8Randall D. Degenhardt L. Vajdic C.M. et al.Increasing cancer mortality among opioid-dependent persons in Australia: a new public health challenge for a disadvantaged population.Aust N Z J Public Health. 2011; 35: 220-225Crossref PubMed Scopus (44) Google Scholar Asia,9Sun M. Lin J.-A. Chang C.-L. Wu S.-Y. Zhang J. Association between long-term opioid use and cancer risk in patients with chronic pain: a propensity score-matched cohort study.Br J Anaesth. 2022; 129: 84-91Abstract Full Text Full Text PDF PubMed Scopus (9) Google Scholar,10Oh T.K. Song I.-A. Chronic opioid use and risk of cancer in patients with chronic noncancer pain: a nationwide historical cohort study.Cancer Epidemiol Biomarkers Prev. 2020; 29 (–7): 1962Crossref PubMed Scopus (4) Google Scholar and Europe11Mancuso P. Djuric O. Collini G. et al.Risk of cancer in individuals with alcohol and drug use disorders: a registry-based study in Reggio Emilia, Italy.Eur J Cancer Prev. 2020; 29: 270-278Crossref PubMed Scopus (5) Google Scholar all showed increased cancer incidence or mortality amongst opioid medication users. Importantly, opium-related cancers (lung, urogenital, and pancreatic cancers) were consistently increased in users of opioid medications, whereas opium-unrelated cancers (e.g. colon and breast cancers) showed inconsistent results.8Randall D. Degenhardt L. Vajdic C.M. et al.Increasing cancer mortality among opioid-dependent persons in Australia: a new public health challenge for a disadvantaged population.Aust N Z J Public Health. 2011; 35: 220-225Crossref PubMed Scopus (44) Google Scholar, 9Sun M. Lin J.-A. Chang C.-L. Wu S.-Y. Zhang J. Association between long-term opioid use and cancer risk in patients with chronic pain: a propensity score-matched cohort study.Br J Anaesth. 2022; 129: 84-91Abstract Full Text Full Text PDF PubMed Scopus (9) Google Scholar, 10Oh T.K. Song I.-A. Chronic opioid use and risk of cancer in patients with chronic noncancer pain: a nationwide historical cohort study.Cancer Epidemiol Biomarkers Prev. 2020; 29 (–7): 1962Crossref PubMed Scopus (4) Google Scholar, 11Mancuso P. Djuric O. Collini G. et al.Risk of cancer in individuals with alcohol and drug use disorders: a registry-based study in Reggio Emilia, Italy.Eur J Cancer Prev. 2020; 29: 270-278Crossref PubMed Scopus (5) Google Scholar Although most these studies assessed cancer risk in relation to using any type of opioid medications, in one study cancer risk was found to be higher in users of strong opioids than in users of weak opioids.10Oh T.K. Song I.-A. Chronic opioid use and risk of cancer in patients with chronic noncancer pain: a nationwide historical cohort study.Cancer Epidemiol Biomarkers Prev. 2020; 29 (–7): 1962Crossref PubMed Scopus (4) Google Scholar A recent ecological study in the USA showed that state-level opioid death rates significantly predicted the incidence and annual changes of pancreatic cancer over the subsequent 4-yr period.12Barlass U. Deshmukh A. Beck T. Bishehsari F. Opioid use as a potential risk factor for pancreatic cancer in the United States: an analysis of state and national level databases.PLoS One. 2021; 16e0244285Crossref PubMed Scopus (6) Google Scholar Although this evidence is suggestive, registry and ecological studies are subject to important potential biases. In particular, linkage studies are subject to confounding as the prevalence of known carcinogens (e.g. tobacco and alcohol use) is higher amongst people who use opioid medications. In addition, certain chronic health conditions are more common in users of opioid medications (e.g. diabetes mellitus, musculoskeletal diseases, and inflammatory bowel diseases), and some of these are also associated with risk for certain cancer types. Given that an RCT to evaluate the effects of opioid medications on cancer outcomes is not feasible, additional evidence in humans must come from observational studies. Furthermore, as opioids are the primary medications used to manage cancer-related pain, and the use of opioids to manage non-cancer pain is increasingly discouraged, retrospective case-control studies will be inherently prone to biases including reverse causality and social desirability, and response bias. Therefore, prospective population-based cohort studies provide the best possible design to study the long-term effects of opioid medications in humans. However, most cohorts lack data on opioid use, and in those that have this information, the number of long-term opioid users cannot provide sufficient power to evaluate rare outcomes such as specific cancer types. One approach to overcome these challenges is to leverage data from multiple sources to generate a single database with detailed individual-level information on the exposure (opioid use), outcomes (cancer incidence), and confounders. Well-characterised population cohort studies have already collected information on confounding factors (tobacco, alcohol, health conditions), and cancer outcomes. The key challenge is to collate these data with detailed individual-level information on opioid use. In many countries, there are increasingly robust databases that record medication dispensing. Linkage of cohort participants to these databases would allow an accurate assessment of opioid medication use at baseline and during follow-up, including details on the type, strength, duration, and dose of medications dispensed, which might differently affect any potential relationship between using opioid medications and cancers. Given that specific cancers are a relatively rare outcome, a consortium-based approach would be needed to assemble a sufficient number of participants who used opioids. Given the increasing challenge posed by data protection regulations, distributed analyses in several cohorts and meta-analysis of the results may be required to circumvent the need for data transfer. A second approach could be to use novel genomic epidemiologic methods facilitated by recent advances in high-throughput genotyping. Genome-wide association studies (GWAS) have now identified independent genetic variants that are associated with use of opioid medications.13Wu Y. Byrne E.M. Zheng Z. et al.Genome-wide association study of medication-use and associated disease in the UK Biobank.Nat Commun. 2019; 10: 1891Crossref PubMed Scopus (89) Google Scholar,14Rosoff D.B. Smith G.D. Lohoff F.W. Prescription opioid use and risk for major depressive disorder and anxiety and stress-related disorders: a multivariable Mendelian randomization analysis.JAMA Psychiatry. 2021; 78: 151-160Crossref PubMed Scopus (46) Google Scholar This could enable use of a Mendelian randomisation approach to study the potential causal association between using opioids and incidence of specific cancers. Mendelian randomisation uses single-nucleotide variants as unconfounded proxies (genetic instruments) for exposures to estimate their effect on outcomes of interest, which minimises the bias affecting observational epidemiologic studies. Conceptually, Mendelian randomisation has analogies to RCTs, with allele randomisation occurring at conception, and can strengthen causal inference when RCTs are impractical or unethical. As a preliminary analysis, we applied two-sample Mendelian randomisation methods to estimate the association between opioid medication use and risk of the four cancer types for which we have access to GWAS data (lung, breast, pancreatic, and colon cancers). The genetic instruments for opioid medication use were derived from a GWAS of medication use in the UK Biobank (first sample),13Wu Y. Byrne E.M. Zheng Z. et al.Genome-wide association study of medication-use and associated disease in the UK Biobank.Nat Commun. 2019; 10: 1891Crossref PubMed Scopus (89) Google Scholar,14Rosoff D.B. Smith G.D. Lohoff F.W. Prescription opioid use and risk for major depressive disorder and anxiety and stress-related disorders: a multivariable Mendelian randomization analysis.JAMA Psychiatry. 2021; 78: 151-160Crossref PubMed Scopus (46) Google Scholar and the outcomes were based on GWAS of lung, pancreatic, breast, and colorectal cancers (second sample).15McKay J.D. Hung R.J. Han Y. et al.Large-scale association analysis identifies new lung cancer susceptibility loci and heterogeneity in genetic susceptibility across histological subtypes.Nat Genet. 2017; 49: 1126-1132Crossref PubMed Scopus (305) Google Scholar, 16Klein A.P. Wolpin B.M. Risch H.A. et al.Genome-wide meta-analysis identifies five new susceptibility loci for pancreatic cancer.Nat Commun. 2018; 9: 556Crossref PubMed Scopus (147) Google Scholar, 17Michailidou K. Lindström S. Dennis J. et al.Association analysis identifies 65 new breast cancer risk loci.Nature. 2017; 551: 92-94Crossref PubMed Scopus (737) Google Scholar, 18Huyghe J.R. Bien S.A. Harrison T.A. et al.Discovery of common and rare genetic risk variants for colorectal cancer.Nat Genet. 2019; 51: 76-87Crossref PubMed Scopus (251) Google Scholar We found that genetic liability to opioid medication use was associated with significantly higher risk for lung cancer (odds ratio [OR]=1.16; 95% confidence interval [CI], 1.05–1.28) and pancreatic cancer (OR=1.24; 95% CI, 1.04–1.48), but showed no statistically significant association with breast or colon cancer (Fig 1). These results are compelling because they suggest potential causal associations between opioid medication use and risk of opium-related cancers. However, a limitation of our Mendelian randomisation analysis is that the genetic instrument for using opioid medications is based on self-reported data which can be prone to underreporting bias. This could have affected the precision of the GWAS estimates for opioid medication use. Furthermore, for some of the cancer sites such as pancreatic and colon cancers, conclusive Mendelian randomisation evidence will require larger outcome GWAS that will allow performing adequately powered sensitivity analyses to investigate potential violations of Mendelian randomisation assumptions. A third possible approach could be to use biomarkers as a proxy for opioid exposure in existing cohorts. Recent studies have identified unique fingerprints of proteins, metabolites, and DNA methylation markers that manifest after exposure to opioids, which could be used to identify exposure to low levels of opioids.19Dhummakupt E.S. Rizzo G.M. Feasel M. et al.Proteomic and metabolomic profiling identify plasma biomarkers for exposure to ultra-low levels of carfentanil.Toxicol Sci. 2019; 167: 524-535Crossref PubMed Scopus (6) Google Scholar Given that many population-based prospective cohort studies have collected biosamples from their participants at recruitment, and some also during follow-up, nested case-control studies could be performed for different cancer types by measuring biomarkers of opioid exposure before cancer diagnosis. This approach could theoretically minimise measurement error in opioid use, as dispensing records cannot capture unprescribed opioid use, nor can they identify when dispensed medications are taken by someone other than the patient of record. However, it remains unclear whether the currently identified biomarkers are specific to use of opioid medications, and further research is warranted to identify and validate biomarkers that are highly sensitive and specific to opioid use. The authors declare no competing interests. Where authors are identified as personnel of the International Agency for Research on Cancer/World Health Organization, the authors alone are responsible for the views expressed in this article and they do not necessarily represent the decisions, policy or views of the International Agency for Research on Cancer/World Health Organization. The presented analysis was supported by the International HundredK+ Cohorts Consortium (IHCC), a programme of the Global Genomic Medicine Collaborative, with funding from the National Institute of Health, the Chan-Zuckerberg Initiative, and the Wellcome Trust.