Lung cancer is responsible for over one million deaths worldwide each year. Smoking cessation for lung cancer prevention remains key, but it is increasingly acknowledged that prevention strategies also need to focus on high-risk groups, including ex-smokers, and patients who have undergone resection of a primary tumor. Models for chemoprevention of lung cancer often present conflicting results, making rational design of lung cancer chemoprevention trials challenging. There has been much focus on use of dietary bioactive compounds in lung cancer prevention strategies, primarily due to their favorable toxicity profile and long history of use within the human populace. One such compound is curcumin, derived from the spice turmeric. This review summarizes and stratifies preclinical evidence for chemopreventive efficacy of curcumin in models of lung cancer, and adjudges the weight of evidence for use of curcumin in lung cancer chemoprevention strategies.
Lung cancer causes over one million deaths worldwide each year (Parkin et al., 2005; Lortet-Tieulent et al., 2014), and is the second most common cause of cancer in the UK (http://www.cancerhelp.org.uk/type/lung-cancer). Five-year survival remains at approximately 16%, with little improvement observed over the past 30 years (Gower et al., 2014), despite the advances made in surgical and therapeutic interventions. The most important risk factor for lung cancer is smoking, which accounts for up to 85% of all cases (Herbst et al., 2008). However, incidence of lung cancer in never smokers is also increasing, with estimates of disease within this cohort accounting for 15% of lung cancers in men and up to 50% in women (Parkin et al., 2005; Jemal et al., 2006; Couraud et al., 2012). Other environmental risk factors include secondhand cigarette smoke, cooking smoke, radon exposure, and exposure to environmental carcinogens, such as asbestos (Field and Withers, 2012; Hubaux et al., 2012). Susceptibility to lung cancer may also be increased by inflammatory conditions, including chronic obstructive pulmonary disease (COPD), pulmonary fibrosis (Sohal et al., 2013), scarring of lung tissues by infectious agents, such as tuberculosis (Liang et al., 2009), and by inherited cancer syndromes caused by germ line mutations in P53, retinoblastoma, and epidermal growth factor receptor (EGFR) (Oxnard et al., 2014).
Although lung cancers often present late, there are many possible avenues for intervention within both a surgical and chemopreventive context. Primary chemoprevention focuses on preventing the development of precancerous lesions particularly in high-risk populations, such as smokers; secondary prevention aims to prevent progression of preneoplasia to cancer; and tertiary prevention is concerned with prevention of spread or recurrence of primary disease (Keith, 2009).
Early disease can be classified into papillomas (squamous, glandular, or mixed) and adenomas (alveolar, papillary, salivary-gland-type). These preinvasive lesions can be further classified as squamous dysplasia/carcinoma in situ, atypical adenomatous hyperplasia, and diffuse idiopathic pulmonary neuroendocrine cell hyperplasia (Brambilla and Gazdar, 2009). There is currently ambiguity regarding the best way to treat early disease. Pulmonary nodules and atypical adenomatous hyperplasia are characterized by a radiologic ground-glass opacity, some of which may remain stable for years, and others which may be the precursors of adenocarcinomas (Nakahara et al., 2001; Jeremy George et al., 2007; Bettio et al., 2012). With improved radiology in ground-glass opacity detection, and the question as to whether surgical intervention is appropriate in all cases, these lesions may present an ideal target for chemopreventive intervention.
Lung cancer itself can be divided into two main categories: small-cell lung cancer and non–small-cell lung cancer (NSCLC). Small-cell lung cancer is found primarily in smokers, with approximately 60% of patients presenting with widespread metastatic disease (most commonly to the brain, liver, or bone) at the time of diagnosis. NSCLC represents approximately 85% of all lung cancers, and can be subdivided into squamous cell carcinoma, adenocarcinoma, and large cell carcinoma, with adenocarcinoma being the most common disease in never smokers (Sun et al., 2007). Adenocarcinomas themselves may be further subdivided based on specific histologic subtyping (mixed subtype, acinar, papillary, solid, micropapillary, lepidic) (Travis et al., 2013). Adding to the complexity of this disease has been the observation that molecular subtyping can have a real impact on treatment outcomes. NSCLCs are now routinely tested for EGFR mutations and anaplastic lymphoma kinase fusions, which are altered in approximately 25% of adenocarcinomas (Shtivelman et al., 2014). This, in conjunction with histologic characteristics, guides targeted therapy treatment decisions. Further oncogenic drivers common to NSCLC include BRAF, KRAS, PIK3CA, RET, HER2, and MET (Munoz et al., 2013; Gower et al., 2014), which may exist within a tumor as multiple molecular drivers, ultimately leading to therapeutic resistance.
The primary intervention for lung cancer is smoking cessation, but a variety of interventional strategies may play important roles in other high-risk groups. Such strategies may include enhanced screening programs, dietary and lifestyle interventions, and chemoprevention.
Chemoprevention in Lung Cancer
Chemoprevention for solid tumors has had some success, particularly in colorectal cancer, yet there is currently little favorable clinical evidence to suggest success in lung cancer across a variety of both dietary and targeted agents (Szabo et al., 2013). A summary of chemoprevention trials for lung cancer is shown in Table 1. The majority of trials fit into the primary prevention category, targeting individuals at high risk. Many useful lessons have been learned from some of the largest trials to date for lung cancer prevention, most notably the Carotine and Retinol Efficacy Trial (Omenn, 2007). From this evidence, we now know that it is essential that chemoprevention trials take into account all preclinical data, including appropriate selection of models, understanding the mechanisms of action for the agents under investigation, understanding relevance of dose within a chemopreventive setting, targeting populations correctly, and giving due consideration to the likelihood of compliance in long-term intervention regimens (Szabo, 2006; Dragnev et al., 2013).
For a compound to be considered for use as a chemopreventive agent, it must have a very favorable toxicity profile, well studied pharmacokinetics (including comprehensive evaluation of stability and metabolism), known mechanism of action, proven efficacy in preclinical models, and be acceptable for long-term human consumption. It is for these reasons that many of the lung cancer chemoprevention trials to date have been undertaken using chemicals that occur naturally within the diet. One such chemical that meets many of the aforementioned criteria, which has yet to be assessed clinically for chemopreventive efficacy in lung cancer, is the dietary agent curcumin.
Curcumin is derived from the spice turmeric, and has been extensively investigated for its chemopreventive potential. The putative mechanisms of action for curcumin are numerous, among which are its radical scavenging activity and anti-inflammatory, antiproliferative, antiangiogenic, proapoptotic, and immune-modulatory properties (Irving et al., 2011). In addition, it has been used safely in a number of early-phase clinical trials for colorectal cancer, pancreatic cancer, Alzheimers’ disease, breast cancer, multiple myeloma, ulcerative colitis, rheumatoid arthritis, and COPD (http://clinicaltrials.gov/ct2/results?term=curcumin&pg=3). Despite much promise as a putative chemopreventive agent, clinical utility of curcumin for indications outside of the gastrointestinal tract has been considerably hindered by its lack of bioavailability (Anand et al., 2007). There is accumulating research, however, into a variety of formulations and administration methods (reviewed in Flora et al., 2013; Subramani and Narala, 2013; Prasad et al., 2014) which may increase the potential of curcumin for use in a wider variety of disease states. These formulations include coadministration with piperine, curcumin nanoparticles, liposomal encapsulation, phospholipid complexing, structural analogs of curcumin, formulation with oligosaccharides, and more recently, the potential for curcumin to be administered via a pressurized metered dose inhaler (Subramani and Narala, 2013). The ability to deliver systemically concentrations of curcumin that have proven biologic activity in preclinical models opens up a real possibility for use of curcumin in lung cancer chemoprevention regimens.
This review will summarize and stratify preclinical evidence for chemopreventive efficacy of curcumin in models of lung cancer published over the past 10 years. Model and dose selection will be discussed in the context of primary and tertiary chemoprevention regimens, and the weight of evidence for use of curcumin as a lung cancer chemopreventive agent adjudged.
Models for Primary Lung Cancer Prevention
Despite the fact that many of the clinical studies investigating lung cancer prevention fit the primary prevention model, there are relatively few in vitro and in vivo studies which assess efficacy of curcumin in “high-risk” models (Tables 2 and 3). The favored in vitro model for assessment of primary intervention strategies using curcumin is the BEAS-2B bronchial epithelial cell line treated with environmental carcinogens including fine particulate matter (Zhang et al., 2012), cadmium (Rennolds et al., 2012), or cigarette smoke (Shishodia et al., 2003). Curcumin appeared able to protect against inflammation in response to these agents, most notably by inhibition of proinflammatory cytokines and prevention of translocation of nuclear factor-κB, a transcription factor which acts as a central mediator of the immune response, to the nucleus. Only one study has used primary human lung fibroblasts, in which curcumin downregulated markers of lung fibrosis (Tourkina et al., 2004). None of the concentrations of curcumin used in these studies (0.5–50 μM) would likely be achievable in human lung tissue following oral dosing of standard nonformulated curcumin. Tmax plasma levels following curcumin doses of 3.5–12 g daily have previously been observed to be in the order of 11 nM to 1.7 μM, with tissue levels in the colorectal mucosa equating to approximately 50 μM, but remaining at the limits of detection for tissues distant to the gastrointestinal tract (Garcea et al., 2004; Howells et al., 2007; Irving et al., 2013). In vivo models explored lung injury induced by a variety of mechanisms, which bear relevance to the type of insult that human lungs would be exposed to. Administration of tobacco-related carcinogens and lung injury–inducing insults [infectious agents and diesel exhaust particles administered intratracheally (Kalpana and Menon, 2004; Vanisree and Sudha, 2006; Xu et al., 2007, 2014; Suzuki et al., 2009; Bansal and Chhibber, 2010; Malhotra et al., 2011, 2012a; Sehgal et al., 2011, 2012; Hamdy et al., 2012; Nemmar et al., 2012; Suresh et al., 2012; Xiao et al., 2012; Avasarala et al., 2013; Cho et al., 2013)] resulted in pulmonary edema, alveolar wall thickening, inflammatory cell infiltration, and fibrosis. These symptoms were attenuated by curcumin (most often delivered orally) at doses typically in the region of 50 mg/kg, which would equate to a dosing regimen of approximately 3.5 g/d for a 70 kg human. However, there is little evidence within the primary chemoprevention setting to definitively equate promising changes at the molecular level with an ultimate decrease in tumor burden.
Models for Secondary Lung Cancer Prevention
Moghaddam et al. (2009) describe one of the few models which fit into the secondary prevention category. Oral curcumin (1% in the diet) was found to inhibit lung tumor formation in a conditional K-ras–induced mouse model, mimicking COPD-like airway inflammation induced by nontypeable Haemophilus influenzae (Moghaddam et al., 2009). Serum curcumin levels at this dose equated to approximately 2 μM, but parent curcumin was undetectable in lung tissue. Tissue levels of 40 ng/mg protein were observed for both demethoxy and bisdemethoxy curcumin.
Models for Tertiary Lung Cancer Prevention
The in vitro data for tertiary prevention using curcumin are presented in Table 4. Although use of established cancer cell lines has therapeutic inference, it also follows that clinical maintenance strategies postresection/chemotherapy would fit within this paradigm for prevention of recurrence. However, when contextualizing the presented data, it is likely that many of the studies described here used established lung cancer cell lines due to ease of obtaining and maintenance of these cell lines, rather than direct targeting and tailoring of preclinical research toward translationally relevant endpoints. Several studies presented in Tables 4 and 5 directly reflect a chemotherapeutic approach, as they combine curcumin with therapeutic drugs. The possibility of applying such combinations clinically may again have potential for use in adjuvant or maintenance regimens, and so for the purpose of this review, are included within the tertiary prevention section.
Of the 38 studies presented, 24 (63%) use the NSCLC adenocarcinoma cell line A549, and 18 studies present data derived from a single cell line only. Most of the studies again used curcumin at concentrations in excess of those likely to be achieved in the lung following oral dosing with standard curcumin, which has not been formulated to specifically enhance bioavailability. Overwhelmingly, mechanisms of action reported for the antitumor efficacy of curcumin in lung cancer cells are via mitochondrial-mediated cell death elicited by an increase in the Bax:B cell lymphoma-2 ratio or by an increase in intracellular reactive oxygen species (ROS) (Chanvorachote et al., 2009; Pongrakhananon et al., 2010; Saha et al., 2010; Wu et al., 2010; Wang et al., 2011, 2013b; Sahoo et al., 2012; Yang et al., 2012a,b; Li et al., 2013; Liu et al., 2013; Xiao et al., 2013; Chen et al., 2014). Migration and invasive capacity of lung cancer cells may be further decreased by inhibition of matrix metalloprotease expression, decreased nuclear factor-κB, EGFR, Akt, signal transducer and activator of transcription 3, and Cdc42 signaling (Chen et al., 2004, 2012, 2014; Lee et al., 2005; Lin et al., 2009, 2012; Puliyappadamba et al., 2010; Kaushik et al., 2012; Liu et al., 2013; Yamauchi et al., 2014; Zhou et al., 2013a; Li et al., 2014b). Although drug resistance in lung cancer is a primary cause for therapeutic failure, very few in vitro models of resistance have been used when investigating the potential of curcumin for tertiary interventional strategies. The A549/CDDP cell line was developed by intermittent administration of cisplatin to native A549 cells (Pan et al., 2009), and is one of the few artificially generated models of resistance that has been used for assessing utility of curcumin to overcome drug resistance. However, some cell lines (PC9, H1975, H1650) become resistant to targeted drugs such as the tyrosine kinase inhibitors (e.g., erlotinib, gefitinib) and have thus been used to expand the potential utility of curcumin to lung cancers harboring/gaining specific mutations. Cotreatment using standard chemotherapy drugs in combination with curcumin enhanced antitumor efficacy of erlotinib (Yamauchi et al., 2014; Li et al., 2014a), cisplatin (Chanvorachote et al., 2009; Ye et al., 2012; Zhou et al., 2013a), and vinorelbine (Chen et al., 2004) by sensitizing cells to drug-induced apoptosis.
In vivo data for tertiary prevention are shown in Table 5. Xenograft models bearing A549 or Lewis lung carcinoma cells provide the majority of data in which curcumin has observable antitumor efficacy, but there are few concurrent mechanistic data to evaluate. Orthotopic models created via intratracheal instillation of lung cancer cells (Lee et al., 2010; Rocks et al., 2012) provide better insight as to whether an antitumor effect of curcumin could be maintained at the correct pathologic site. In all but one of the studies cited, either formulated curcumin or curcumin analogs were used to enhance bioavailability. Curcumin/curcuminoids were also used in conjunction with chemotherapy agents doxorubicin (Wang et al., 2013a), phosphosulindac (Cheng et al., 2013), gemcitabine (Rocks et al., 2012), docetaxel (Yin et al., 2012), or radiation (Lee et al., 2010; Shi et al., 2012). No studies took advantage of testing these novel formulations of curcumin, or combinations of curcumin with cytotoxic agents, using therapy-resistant lung tumor models.
There has been both promise and failure in equal measure for chemoprevention of lung cancer. Numerous agents have been taken into clinical trials, and although many have not had any observable effects on lung cancer incidence, some have proven to be of harm. Dietary-derived agents offer great potential for use in chemoprevention regimens due to their favorable toxicity profiles and long history of use within the human populace. Curcumin, derived from the spice turmeric, has undergone extensive preclinical investigation in models of lung carcinogenesis, and so we sought to evaluate whether the evidence was sufficient to confidently apply this knowledge to clinical interventions with curcumin in lung cancer chemoprevention strategies.
In vitro studies (for both primary and tertiary chemoprevention) appear extremely limited in their choice of both model and concentration/dose selection. Mechanistic inferences are often made using a very limited choice of cell lines. This does little to represent the heterogeneity that is inherent in lung cancer, in which there are many distinct subclassifications with their own specific driver mutations, responding very differently to therapeutic intervention (Munoz et al., 2013). There is therefore a need to assess the effects of curcumin across more extensive cellular models. Such systems should include models of cellular resistance against the current individual and combinations of therapeutics used clinically; coculture approaches such as the organotypic model, which allows interaction of lung cancer cells with mesenchymal cells; and explant cultures using primary tissue and cell cultures from primary tissues. Use of such methodology would allow greater insight into the effects of curcumin across clinically relevant models of resistance, and would better take into account the interaction that tumor cells have with their microenvironment. Furthermore, explant cultures would allow short-term treatments of cells where the tumor architecture is maintained, and the use of primary tissue-derived cell lines would help to increase the heterogeneity of the sampling population. These approaches are now commonly used for investigation into molecular mechanisms for targeted agents, and should also be incorporated into preclinical strategies for chemopreventive agents. This would be a rational approach for improving the chances of translating observations of preclinical efficacy into the clinic.
A further issue for much of the in vitro data presented here is that of choice of dose. The extensive use of high concentrations of curcumin (>10 μM) to achieve positive effects on mechanistic endpoints is unlikely to provide meaningful insight regarding the potential in vivo mechanisms of action, particularly those that might relate to humans. Although high concentrations consistently induce apoptosis in lung cancer cell lines via mitochondrial-mediated mechanisms, there is little evidence to suggest that this can also occur with long-term exposure to low concentrations, which would bear greater relevance to clinical models of chemoprevention. To define the key mechanisms of curcumin’s action that might translate to humans, it is essential to use clinically relevant systems and dosing regimens, but to date, little consideration has been given to these issues. Determination of suitable biomarkers of efficacy presents a problem for any chemoprevention trial, and translating in vitro markers of efficacy to the clinic could be deemed questionable, particularly when the models used have been so limited.
There are few animal models which accurately reflect the evolution of cancer in humans, and lung cancer models are no exception. The same limitations observed for the in vitro data apply to the animal models presented here, particularly in the tertiary models of prevention. The majority are mouse xenografts of Lewis lung carcinoma, or A549 cells, giving limited insight into potential drug efficacy across a wide and varied mutational spectrum. Orthotopic models of lung cancer in mice are less often used, but may present a more favorable system, particularly when assessment of drug delivery to tumors at their target site is required. If suitable orthotopic models cannot be generated, then patient-derived xenografts can provide well characterized models of specific gene mutations, which would greatly enhance investigation of preclinical drug efficacy (Zhang et al., 2013; Malaney et al., 2014). Although animal models of primary prevention are credible in that they deliver common environmental insults directly to the lung, there is still a paucity of data regarding efficacy of and mechanisms by which curcumin prevents lung injury. The anti-inflammatory effects of curcumin are most widely cited as its likely mechanism of action, but there are few comparators with commonly used steroidal/nonsteroidal anti-inflammatories with a similar mechanism of action. A wide variety of curcumin analogs and formulations designed to enhance absorption and/or retard metabolism have been used in in vivo models, most of which suggest enhanced antitumor efficacy compared with nonformulated curcumin, due to their greater bioavailability. Doses of curcumin used in vivo have the potential to be recapitulated in the clinical setting, and it may well be that curcumin formulated for greater bioavailability could be delivered orally at far lower doses than would be required for the parent compound. Although there is also promise of efficacy for the new curcumin analogs, their potential for clinical utility in a chemoprevention setting is limited until extensive and lengthy safety evaluations have been undertaken.
There are very few models of secondary lung cancer prevention, most likely due to the ambiguities as to whether early lesions definitively lead to carcinogenic progression. This is compounded by the rarity of resection of premalignant lesions, meaning that few tissues are available for use to generate cell lines to model this clinical paradigm. Furthermore, propagation of cells derived from such lesions will likely present with greater technical challenges compared with their malignant counterparts. Although a large proportion of the data presented here may show positive advocacy for the potential of curcumin in lung cancer prevention, there are a number of studies to note suggesting a detrimental role for curcumin in some in vivo models of lung cancer. Dance-Barnes et al. (2009) observed increased tumor multiplicity in a transgenic model of lung cancer [K-ras(G12C)], following administration of dietary curcumin. It has recently been shown that autophagy may promote BrafV600E-driven lung carcinogenesis (Strohecker and White, 2014), with curcumin known to be an inducer of autophagy in the lung (Xiao et al., 2013). Furthermore, in Lewis lung tumor xenografts, dietary curcumin increased the cross-sectional area of metastases concurrent with increases in levels of proinflammatory cytokines (Dance-Barnes et al., 2009; Yan, 2013). Curcumin is a potent antioxidant, yet paradoxically, antioxidants may have the facility to enhance carcinogenic progression (Sayin et al., 2014). However, it should be noted that, although trials have shown higher lung cancer risk in participants taking β-carotene, this was not the case for those taking the similarly potent antioxidant α-tocopherol. Therefore, it is perhaps an unlikely prediction that all antioxidants would elicit procarcinogenic mechanisms in lung models (Kaiser, 2014). Mechanisms for the procarcinogenic antioxidant effect have been postulated to be due to decreased ROS, leading to downregulation of the ROS-induced activation of p53-regulated DNA damage response, although approximately 50% of all NSCLCs are p53 mutant (Mogi and Kuwano, 2011), which would discount this mechanism. Curcumin itself can undergo extensive oxidation, which functions as a pro-oxidant switch producing curcumin radicals (Heger et al., 2014), with potential for procarcinogenic oxidative damage. Despite these observations, much of the available evidence still favors curcumin as having anticarcinogenic potential in lung models. These ambiguities further highlight the absolute requirement that greater understanding of the pleiotropic properties of curcumin is required within the lung cancer chemoprevention setting.
Prior to large-scale chemoprevention trials being undertaken with curcumin, several factors need to be addressed. Here, we have reviewed the literature on efficacy data for curcumin in lung cancer, and stratified it according to which prevention model it could fit into. First, models for chemopreventive efficacy in the primary and tertiary setting are extremely limited and often inappropriate. Models are largely nonexistent for the secondary prevention setting. Second, most mechanistic data are obtained at curcumin concentrations that could never be achieved clinically, even when using formulations with greater bioavailability. Many models use cytotoxic endpoints only, but epigenetic modulation of carcinogenic pathways at subcytotoxic levels has had little consideration. The ultimate application of chemopreventive strategies is in the primary prevention setting. However, the tertiary chemoprevention setting provides a targeted, closely monitored cohort that has undergone extensive disease classification, allowing assessment of efficacy biomarkers to be correlated with progression-free and overall survival. This type of approach is necessitated, as the current lack of clinical efficacy data for curcumin in lung malignancy should preclude large-scale interventions within the healthy populace.
In summary, curcumin meets many of the criteria necessary to be successfully taken forward to the clinic, but there is much work still to be done to bridge the knowledge gaps prior to entering it into large-scale prevention studies for lung cancer. This is critical if the failings of other trials championing diet-derived agents are not to be repeated.
Wrote or contributed to the writing of the manuscript: Howells, Mahale, Sale, McVeigh, Thomas, Steward, Brown.
- Received May 7, 2014.
- Accepted June 16, 2014.
This work was supported by Cancer Research UK [C325/A6691] and the Leicester Experimental Cancer Medicine Centre [C325/A15575], funded by Cancer Research UK/UK Department of Health.
- chronic obstructive pulmonary disease
- epidermal growth factor receptor
- non–small-cell lung cancer
- reactive oxygen species
- Copyright © 2014 by The American Society for Pharmacology and Experimental Therapeutics