Advancing Cancer Interception

The opportunity for cancer interception is the moment after the first and earliest cellular steps toward tumorigenesis, before tumor invasion and long before metastasis. As a potential maneuver in medicine, interception is active, “to intercept a cancer development process before the damage is done,” as Nobel Laureate and molecular biologist Elizabeth Blackburn wrote in 2011 in her landmark essay (1), “before the full-blown advanced tumor presents in the clinic.”

Interception aims to leverage two key features of premalignancy as a therapeutic target: (i) a window of opportunity afforded by the extended time cancers take to transit from premalignancy to a complex and diagnosable cancer and (ii) the potential for relatively low complexity that might limit mechanisms of resistance well-described for advanced cancers. Thus, interception is neither primary prevention per se, such as lifestyle modifications to avoid carcinogenetic exposure (such as smoking, alcohol, or unhealthy diets), nor tertiary prevention, such as efforts to eradicate minimal residual disease after primary surgical resection, which is the purpose of adjuvant therapy (Fig. 1).

Clinically, interception is already well-established using “mechanical” means. Removal of benign adenomatous colon polyps at screening colonoscopy markedly decreases the risk for subsequent cancer (2). Likewise, detection and removal of cervical intraepithelial neoplasia (CIN3) at colposcopy with excision or ablation intercepts the development of precursor lesions that commonly progress to invasive cancer. Risk-reducing bilateral salpingo-oophorectomy removes tissue that might one day develop cancer and, in some cases, is found at pathology to have surgically intercepted early lesions in the ovary and fallopian tubes.

There are also numerous examples of “medical” cancer interception. In appropriately selected individuals, such as those with Lynch syndrome, protracted daily aspirin is indicated for secondary prevention of polyps and reduces the incidence and mortality of colorectal cancer. Women with atypical ductal hyperplasia of the breast significantly reduce their risk of developing invasive breast cancer by taking tamoxifen, raloxifene, or exemestane. The smoothened inhibitors vismodegib and sonidegib, approved for patients with advanced basal cell carcinoma, are extremely effective in intercepting basal cell neoplasia in patients with Gorlin syndrome with germline PTCH1 mutations. More recently, studies of belzutifan (an oral hypoxia-inducible factor 2α inhibitor) demonstrate that in individuals with von Hippel–Lindau Disease, treatment effectively intercepts the growth of renal cell carcinoma and other tumors by blocking a key biological pathway in oncogenesis. In the definitive clinical trial for belzutifan (3), patients with significant tumor burden saw not only regression of preexisting tumors but marked reduction in the appearance of new lesions in short-term follow-up once treatment was started. The examples of smoothened inhibitors and belzutifan demonstrate the power of understanding the pathway by which the target malignancies develop in interception approaches.

We propose “immuno-interception” as another strategy, namely, to eliminate neoplastic lesions at their earliest stages by mobilizing a specific immune response. Although cancer immunotherapies can be remarkably effective, leveraging the power of antitumor T cells, there has been only one therapeutic cancer vaccine approved (sipuleucel-T) and no clinically approved prevention vaccines for cancer other than those that prevent infection of a cancer-causing virus [e.g., human papillomavirus (HPV)]. Most experimental “cancer vaccines” have been given to tumor-bearing patients in the “therapeutic” which poses enormous immunologic obstacles that might thwart even the best vaccines, namely tumor cell–intrinsic immunosuppression, immune-suppressive tumor microenvironment (TME), and weakened patient immune systems in the face of chemotherapy and the disease itself. It is possible that fewer of these barriers exist in the adjuvant setting (when minimal residual disease would be the target); however, none of these obstacles exist to such an extent in premalignancy.

Thus, the prospect and promise of immuno-interception is to intervene and generate adaptive immune responses before negative effects of the tumor, the TME, and treatment gain an upper hand. A provocative example is a randomized trial of an MUC-1 peptide vaccine with adjuvant for the prevention of recurrent colorectal adenomas (NCT00773097) for which there was a 38% absolute reduction in adenoma recurrence versus placebo in the minority of participants who had an immune response (4). Other active studies include vaccinating individuals at high risk for pancreatic cancer with mutant Kras peptide and adjuvant (NCT05013216) or vaccinating otherwise healthy individuals with Lynch syndrome using an off-the-shelf, viral-vector vaccine targeting recurrently observed frameshift neopeptides associated with microsatellite unstable colon cancers (NCT05078866). In our own work, we are vaccinating healthy individuals who carry germline mutations in BRCA1 or BRCA2 using a plasmid DNA vaccine encoding the telomerase reverse transcriptase (TERT) and two other antigens, with or without IL12 DNA plasmid, and delivered intramuscularly with electroporation (NCT04367675). Telomerase reactivation is one of the earliest and most common molecular steps in carcinoma in situ and other premalignancies and can be recognized by cytotoxic T cells. In patients in remission after surgery and adjuvant therapy, TERT/IL12 DNA was immunogenic in 96% of participants without serious or limiting toxicity (NCT02960594; ref. 5).

Anti-inflammatory approaches may also have immune-interception effects. The human monoclonal anti–interleukin 1B antibody canakinumab was tested in the CANTOS trial for a role in dampening inflammation and abrogating atherosclerosis and cardiovascular events, but in a hypothesis-generating exploratory analysis, canakinumab was associated with a decreased risk of incident lung cancer in this large randomized clinical trial involving more than 10,000 participants (NCT01327846; ref. 6). Subsequent studies in the “therapeutic” setting, however, have shown no benefit of adding canakinumab to standard lung cancer therapy in either the metastatic (CANOPY-1) or adjuvant (CANOPY-A) settings. Because of distinctive biology, agents may have different effects as cancer therapy versus cancer interception, but unfortunately, the CANAL trial (NCT05725343) of canakinumab in those at high risk for lung cancer was terminated by the sponsor based on the data from CANOPY-1 and CANOPY-A.

The relevant biologic mechanisms in cancer interception are the earliest steps in cell transformation and the emergence of nanotissue microenvironments that support or allow progression of benign neoplasia to malignancy. Critical pathways and networks during this sequence may not necessarily be those driving primary tumor development, invasion, and metastasis. Little is currently known about whether most cancers develop in a linear fashion with sequential well-described steps, whether particular insults could instead occur at different times along the pathway, or whether exogenous exposures and the tumor microenvironment act as cancer promoters (7). Thus, standard discovery tools and databases (such as The Cancer Genome Atlas, which has mostly primary tumors) may not be completely relevant. Early lesions are tiny compared to invasive cancers. Nevertheless, novel technologies, such as single-cell sequencing and advanced tissue imaging, offer opportunities to operate at this nanoscale and reveal distinctive features—and potential interception targets—of early neoplastic lesions, associated immune microenvironments, and contributions from the microbiome. This is the strategy driving the highly curated precancer atlases from the Human Tumor Atlas Network and other research initiatives from the NCI regarding precancerous pathophysiology and interception (8). It is already being appreciated from this work how mechanisms driving precancerous lesion vary depending on the tissue, cell of origin, genetic background, and preexisting inflammation. Moreover, bona fide oncogenes, such as mutant Kras, can be found in benign lesions that never progress. When cancer does develop from a single lesion, “longitudinal trajectory” may be step-wise and linear, as one reads in textbooks, or it may actually be branching and plastic (9). It is possible that premalignancies, depending on the context, may lack a single, dominant pathway that can be drugged for complete resolution in the way inhibitors of mutant Braf or EGFR produce marked, if temporary, full regressions of advanced melanoma and lung cancer expressing these dominant and causative aberrations.

For immuno-interception, major biological issues remain to be understood from clinical trials and model experimental systems. For example, tissue inflammation in premalignancy is typically minimal compared with what can be observed for invasive disease, such that danger signals may be lacking for a full-blown adaptive immune response unless otherwise provided therapeutically. Moreover, it is not fully understood the extent to which nascent neoplastic lesions are either antigenic or immunogenic and whether a T-cell response can be as briskly generated to attack noninvasive cancers as they can be against highly immunogenic neoepitopes that are features of most advanced cancers.

Whether genetically engineered mouse models of cancer that reproduce progression of noninvasive lesions to invasive cancer over the course of weeks to months may offer tools and a roadmap for advancing cancer interception remains to be seen. Although costly and cumbersome, the promise of genetics models would seem superior to classic implantable mouse tumor models that do not reproduce premalignancy. Moreover, tumors developing in genetic models in the setting of a host with a competent immune system, potentially facilitating discovery of immune-interception approaches.

Designing and executing clinical trials of medical or immunologic approaches to interception is challenging to show efficacy. There are currently no equivalents to RECIST criteria in premalignancy. Although biomarkers can potentially substitute (e.g., HPV vaccines first shown to reduce rates of viral infection), there is no surrogate endpoint or unambiguous biomarker to detect early signals of efficacy with confidence. Definitive large, long, and expensive phase III trials with new incident cancer as the primary endpoint are often launched on the basis of preclinical evidence without obvious biomarkers and can have disappointing results (such as the SELECT trial; ref. 10). Interception trials can be terminated based on negative trials in the metastatic or adjuvant setting (such as CANAL).

Even when successful trials have been completed and medications are approved for prevention/interception (e.g., tamoxifen and raloxifene in breast cancer), there can be disappointingly low uptake clinically. This challenge is markedly different from prevention/interception strategies in cardiovascular disease. Although the purpose of treating hypertension and hypercholesterolemia is to reduce myocardial infarctions and strokes, blood pressure and serum lipids serve as effective surrogate endpoints for new therapies, particularly in early studies. In addition, patients are often motivated to continue taking such medications as they can see their effect (a lowering of blood pressure or cholesterol), which is reinforcing. Thus, it is critical in clinical trial development to capture measures, including patient-reported outcomes and the likelihood that participants would recommend the approach to others, that address these issues.

Emphasis needs to be placed on the use of preclinical models to identify putative surrogate biomarkers that can then be taken into clinical trials with relevant tissue biopsies pre- and postexposure. These smaller clinical trials can provide validation of the importance of the pathway before large phase III trials. An example of this is the preclinical identification of the RANKL pathway as important in the development of BRCA1 mutation–associated breast cancer as discussed below.

As efforts continue to elucidate and validate surrogate biomarkers in animal models and smaller studies, testing interception intervention will be best done by identifying groups at the highest risk of the relevant disease, which allows for appropriate identification of pathways, targets, and clinical trial designs. This approach also allows much smaller scale studies when the a priori risk of developing cancer is relatively high in the targeted population, enhancing statistical power to meet an endpoint of incident cancer rather than just a surrogate biomarker.

Lifestyle, histologic, and genetic factors are associated with increasingly more precise estimations of risk. Just as a history of heavy smoking or evidence for premalignant lesions, such as Barrett esophagus or atypical ductal hyperplasia of the breast, are associated with a quantifiable level of subsequent risk, focusing on germline genetics affords many practical advantages in clinical development for proof of concept. There are now more than 100 known cancer predisposition genes with varying degrees of risk and clinical relevance, some with extraordinarily high relative lifetime risks of >70%. Each of these germline genetic predisposition genes allows for an opportunity to understand the molecular pathway and could lead to interception.

As an example of this, individuals with germline pathogenic variants (PV) in BRCA1 and BRCA2 are at very elevated risk for the development of cancers, specifically breast (lifetime risk of approximately 70% in women), ovarian (up to 45%), high-grade prostate (>25%), and pancreatic cancer (up to 7%). In most cases, canonical tumors developing in the setting of a BRCA1 or BRCA2 PV have loss of heterozygosity of the corresponding allele, leading to absent protein expression in the tumor. This transition from the heterozygote, haplo-insufficient state to loss of protein function but prior to clinical manifestation of malignancy is a potential therapeutic opportunity for interception.

In women with known PV in BRCA1 and BRCA2, the risk of developing cancer prior to age 30 is low (<5%) and subsequently increases. Between ages 35 and 60 the risk of developing breast cancer is approximately 2% to 3% per year. Evidence suggests that many ovarian cancers in women with PV in BRCA start in the fallopian tubes with an estimated 7-year window between the development of serous tubal intraepithelial carcinoma (STIC) and clinically apparent ovarian cancer (11).

In addition to mechanical interception (salpingectomy), and cancer-specific approaches (oral contraceptive pills to decrease the risk of ovarian cancer and selective estrogen receptor modulators for breast cancer), there are several novel BRCA-specific approaches. The BRCA-P trial (NCT04711109) is a large international randomized phase III trial investigating the use of denosumab (a human mAb that blocks RANKL) to reduce the development of breast cancer in women with BRCA1 PV based on robust preclinical evidence and a biomarker-driven window-of-opportunity study demonstrating the importance of RANKL in preneoplastic luminal progenitor cells (12). Recruitment is ongoing to this important trial. The seminal OlympiA trial, a randomized phase III trial of the oral PARP inhibitor olaparib for 1 year versus placebo in high-risk, early-stage BRCA1/2-associated breast cancer, demonstrated an improvement not only in disease-free survival but overall survival, changing the paradigm for clinical care (13). Detailed quality-of-life studies suggest that the side-effect profile may allow for consideration of PARP inhibitors for disease interception, particularly if they could be given intermittently for short courses. Notably, PARP inhibitors have the potential to intercept the spectrum of canonical BRCA-associated tumors: breast, ovarian, prostate, and pancreatic. Continued follow-up of the OlympiA trial and other studies is needed to understand potential toxicity, whereas preclinical models are needed to understand dose and schedule.

Critical for advancing cancer interception is the need to improve the power of early detection to identify the earliest possible disease states. Although genetic and other clinical factors help define risk, an at-risk individual with a positive early detection test but “scope and scan negative” may be particularly appropriate for interception. Multicancer early detection (MCED) assays aim to detect cancer of multiple types by looking for circulating tumor cells, cell-free tumor DNA, and tumor-associated proteins by various technologies including DNA methylation assays, mutational analysis, and isolation of exosomes and extracellular vesicles. Although these assays are promising, results from ongoing large prospective trials (ISRCTN91431511) are still needed to guide clinical utility and optimal deployment. Specifically, more information in asymptomatic individuals is needed regarding (i) positive predictive value, (ii) concerns regarding over diagnosis [many of the stage I/II cancers in the Pathfinder study (14) were low-grade lymphomas], (iii) utility of these tests in currently “unscreenable” cancers (e.g., pancreatic cancer or ovarian cancer), and (iv) whether these can take the place of, or supplement, existing screening tests. Although the commercial entities developing these assays use different technologies, it is possible that a multianalyte solution would be best.

Moreover, it is not fully understood currently if an individual who is positive on an MCED test but found to be “scan and scope” negative reliably represents someone bearing only premalignant lesions rather than an MCED false positive (truly no cancer present) or a scan/scope false negative (cancer present but not detectable by current methods). Nor is it understood the natural history of such individuals. Rather than deploying MCED assays broadly across the population, there may be an opportunity for the purpose of advancing interception to focus MCED testing on those at high risk based on germline screening (e.g., MCED testing of healthy individuals who carry BRCA1/2 PV to determine eligibility for interception trials).

Cancer interception may not need to be constant in the way cancer therapeutic regimens for metastatic disease are typically used. After successful removal of benign precursor lesions (e.g., for adenomatous colon polyps), the indication for repeat screening is on the order of years, not weeks or months, given the slow time course of benign polyp formation and subsequent tumorigenesis. Although genetic or other risk factors might drive a steady accumulation of precursor lesions, even medical approaches to interception could be intermittent or periodic if successful enough to eliminate precursor lesions and return the individual back to baseline risk. Dose and schedule could be designed in preclinical models, which is particularly important for medical approaches that otherwise might be too toxic or intolerable if prescribed on a daily and/or chronic basis such as PARP inhibitors or EGFR inhibitors. It is possible that treatment could occur for a limited time every few years: ridding the tissues of premalignant clones akin to removal of colonic polyps. For immuno-interception, although we usually insist that successful vaccination must establish long-lasting and high-level immune memory, this may be neither necessary, if repeated vaccination separated by years is sufficient to clear early lesions, nor desirable from a toxicity standpoint if the target antigen is self or altered-self (such as MUC1 or TERT).

With advances in biological insights and precision technology, we can now imagine a scenario in which individuals are optimally risk-stratified based on their genetics (including monoallelic conditions and polygenic risk scores from single-nucleotide polymorphisms) or environmental and other clinical findings. On the basis of this risk stratification, individuals could be intermittently treated with biologically informed interception interventions and monitored by measuring appropriate biomarkers and optimal early detection.

Cancer drug development classically begins with phase I testing in the most advanced and refractory treatment, and then if safe with some evidence of activity, further testing and validation are accomplished in less-advanced or earlier-stage patients. Applied to advancing interception, this approach would take a great deal of time and assumes, possibly incorrectly, that the cancer pathways driving invasive disease (and being targeted therapeutically) are the same driving premalignancy or its progression.

There are numerous and formidable questions in moving drug development from a treatment model to one focused on cancer interception:

• What is the overlap of biological pathways driving invasive cancer versus premalignancy? New experimental systems such as those focused on intraepithelial neoplasia (rather than invasive cancers in humans or implantable cell line models in mice) are needed.

• What are novel medical and immunologic agents to inhibit premalignant lesions? These may be against known pathways, but there is now a technical opportunity to discover drivers and novel immune-relevant antigens in premalignancy.

• What is the right dose and schedule for interception agents? Like colonoscopy recommended every 5 to 10 years, cancer interception may best be intermittent.

• What biological features define an individual at highest risk in the immediate future for bearing premalignant lesions that progress to cancer? Combinations of new technologies such as germline testing and MCED offer an opportunity.

• What trial designs, biostatistical methods, and biomarkers will allow faster development, validation, and approval of new interception agents? If we must rely on very large or even population-wide studies, progress will be slow.

We are seeing an increasing call for cancer interception. Advances in understanding risk assessment, precancer biology, and early detection put this vision within reach. The impact could be enormous.

S.M. Domchek reports personal fees from AstraZeneca and personal fees from GSK outside the submitted work. R.H. Vonderheide reports personal fees from BMS, grants from Revolution Medicines, and personal fees from Children's Hospital Boston outside the submitted work; in addition, R.H. Vonderheide has a patent for cellular therapy issued, licensed, and with royalties paid; a patent for cancer vaccines issued; and a patent for Kras immune epitopes pending. No other disclosures are reported.

This work is supported by Basser Center for BRCA and the Basser Cancer Interception Institute at Penn, and funding grants are from the Breast Cancer Research Foundation (to S.M. Domchek and R.H. Vonderheide) and the Parker Institute for Cancer Immunotherapy (to R.H. Vonderheide). We thank many colleagues in the interception field whose primary work we are not able to cite due to space limitations.