Cancer screening with multicancer detection tests: A translational science review

Abstract

Multicancer detection (MCD) tests use a single, easily obtainable biospecimen, such as blood, to screen for more than one cancer concurrently. MCD tests can potentially be used to improve early cancer detection, including cancers that currently lack effective screening methods. However, these tests have unknown and unquantified benefits and harms. MCD tests differ from conventional cancer screening tests in that the organ responsible for a positive test is unknown, and a broad diagnostic workup may be necessary to confirm the location and type of underlying cancer. Among two prospective studies involving greater than 16,000 individuals, MCD tests identified those who had some cancers without currently recommended screening tests, including pancreas, ovary, liver, uterus, small intestine, oropharyngeal, bone, thyroid, and hematologic malignancies, at early stages. Reported MCD test sensitivities range from 27% to 95% but differ by organ and are lower for early stage cancers, for which treatment toxicity would be lowest and the potential for cure might be highest. False reassurance from a negative MCD result may reduce screening adherence, risking a loss in proven public health benefits from standard-of-care screening. Prospective clinical trials are needed to address uncertainties about MCD accuracy to detect different cancers in asymptomatic individuals, whether these tests can detect cancer sufficiently early for effective treatment and mortality reduction, the degree to which these tests may contribute to cancer overdiagnosis and overtreatment, whether MCD tests work equally well across all populations, and the appropriate diagnostic evaluation and follow-up for patients with a positive test.

INTRODUCTION

Approximately 1.9 million people face a new cancer diagnosis and 600,000 will die with cancer each year in the United States.^1,2 Detecting cancer at earlier stages can potentially reduce cancer-related morbidity and mortality¹ and is a vital component of President Biden’s Cancer Moonshot goal to reduce cancer mortality rates by at least 50% over the next 25 years.^3,4 Public^5,6 and private investment in cancer research has led to a new generation of cancer screening tests, known as multicancer detection (MCD) tests, which measure biologic molecular analytes (e.g., DNA, RNA, and/or proteins) in readily accessible body fluids to signal the possible presence of cancer in asymptomatic individuals. MCD tests may provide early detection for cancers in organs that currently lack effective early cancer detection methods, and these cancers account for about 65% of cancer deaths.¹ At this time, no MCD tests have received marketing authorization by the US Food and Drug Administration (FDA). Some health plans provide MCD tests as part of their screening programs; however, in a recent study of 19 major private payers, none reported covering MCD testing.⁷

The usefulness of a cancer screening test to reduce cancer morbidity and mortality is interrelated with test performance measures, such as sensitivity (correctly detecting a cancer signal in those with cancer) and specificity (correctly identifying a cancer signal only in those with cancer), the acceptability and affordability of the test, the potential burden and costs of the diagnostic workup after a positive MCD screening result, and effective cancer care delivery that leads to improved patient outcomes, including survival. To date, cancer detection tests recommended for widespread population screening (breast, cervix, colon, and lung) have demonstrated improvements in cancer-specific mortality.⁸ At this time, however, there is a lack of data to inform the benefits of MCD testing (e.g., reduced cancer morbidity or mortality) or its harms (e.g., unnecessary procedures and exposure to radiation during a diagnostic workup, overdiagnosis, false reassurance, anxiety, and worsening of health disparities). This review summarizes current evidence on MCD tests and large-scale randomized controlled trials (RCTs) in early development that aim to fill evidence gaps on their clinical utility.

Overview of multicancer detection tests

Cancer screening aims to detect cancer at earlier and, presumably, more treatable stages, thereby reducing cancer mortality. Contemporary cancer screening methods are organ-specific, aim to detect common cancers at early stages, and are designed to have high sensitivity. Some screening tests (e.g., cervical and colorectal) primarily reduce mortality by detecting and treating precancer, which subsequently prevents incident cancer. Other cancer screening tests reduce mortality through early cancer detection (e.g., breast and lung). Adherence to recommended screening in the United States is about 80% for breast and cervical cancer,⁹ <70% for colorectal cancer,⁹ and about 13% for lung cancer.¹⁰ Over time, an individual undergoing recommended periodic screening has a high cumulative probability of having an invasive diagnostic workup ultimately because of a false-positive screening result.¹¹ For example, women undergoing breast cancer screening have an approximate 50% cumulative probability of being called back for at least one false-positive result over 10 years (not all of which lead to a diagnostic workup).¹²

MCD tests are an appealing new paradigm in cancer screening in that they use a single, readily obtainable biospecimen to screen for more than one cancer concurrently, including lethal cancers for which there is no screening test available. Single organ-based cancer screening in an asymptomatic population at average risk, in which the prevalence of individual cancers is relatively low, requires a large number needed to screen to detect one invasive cancer. Therefore, in this situation, a high test sensitivity is required to achieve a clinically acceptable positive predictive value (PPV) (i.e., the probability that a patient with an abnormal test result actually has the disease). Although the diagnostic performance of both single-organ and MCD tests is driven by their PPV, the PPV of MCD tests is strongly dependent on specificity and less so on sensitivity. This is because many low-prevalence cancers are aggregated into a single MCD screen.¹³ Consequently, the number needed to screen can be markedly reduced by MCD testing. In addition, because MCD testing can potentially screen for relatively uncommon cancers at the same time as more common ones, they represent a potential improvement over recommendations that are comprised of multiple organ-specific screening tests.¹³ MCD testing may also improve screening adherence because of the ease of sample collection.

The laboratory community generally uses the term, multicancer early detection tests (MCEDs). However, it is unclear whether asymptomatic cancers are detected sufficiently early to improve health outcomes. Therefore, in the absence of evidence that such tests detect early stage cancer at significantly high enough rates and/or markedly decrease the incidence of advanced-stage disease, which may be surrogates of a reduction in cancer-specific mortality, the National Cancer Institute (NCI) refers to MCEDs as MCD (multicancer detection) tests.

MCD tests are not simple blood tests. Rather, MCD tests are screening tests that assess the probability of the presence of cancer without making a definitive diagnosis. Further diagnostic workup is necessary when results indicate an increased likelihood of cancer, posing potential harms from invasive diagnostic procedures in multiple, healthy organs in the process of further diagnostic evaluation.

Evaluation of this new technology is a challenging task. There are many current knowledge gaps, and the frameworks for evaluating these tests are relatively new¹⁴ or are still under development.^15,16

Role of federal regulations

MCD tests are commercially available even while scientific investigation to evaluate their test performance is ongoing. Promotional campaigns directed at patients, clinicians, and health care systems are ongoing for many MCD tests in various stages of development, although the evidence supporting their widespread use is currently limited. These trends, reflected in the popular¹⁷ and medical press,¹⁸ raise a critical need to quantify risks and benefits of MCD test use.

Federal regulation of in vitro diagnostic tests does not require demonstration of clinical utility, such as a reduction in morbidity or cancer-related mortality through the use of MCD tests. Table 1 describes federal regulatory standards.¹⁹ Under federal regulations, MCD tests may be introduced into clinical care as laboratory-developed tests (LDTs) without FDA review.²⁰ LDTs are in vitro diagnostic products that are intended for clinical use and are designed, manufactured, and used within a single clinical laboratory,²⁰ i.e., they cannot be distributed for use in other laboratories. LDTs are primarily regulated by the Clinical Laboratory Improvement Amendments of 1988, which focus on laboratory operations and require that laboratories assess the analytical performance of new tests. MCD tests that are now clinically available as LDTs are not required to have demonstrated clinical validity (e.g., accurately predict the presence of cancer at early or advanced stages) or to show that the tests can be safely used without prompting unnecessary or harmful diagnostic evaluations. The Clinical Laboratory Improvement Amendments of 1988 do not regulate product labeling or marketing claims for LDTs.

TABLE 1.

Regulatory standards applied by the US Food and Drug Administration and the Clinical Laboratory Improvement Amendments of 1988.

	FDA-reviewed in vitro diagnostics	Laboratory-developed tests overseen by CLIA
Demonstration of analytical performance (accurate and reproducible detection of the analytes of interest) for new tests	Yes	Yes
Demonstration of clinical validity (accuracy of identifying, measuring, or predicting a clinical condition, such as the presence of cancer at early or advanced stages)	Yes	No
External review of moderate-risk and high-risk tests before use on patients	Yes	No
Public reporting of adverse events (e.g., false results leading to unnecessary diagnostic workups)	Yes	No
Review and approval of product labeling to ensure comprehensiveness and accuracy	Yes	No
Review and approval of marketing claims based on supporting evidence	Yes	No
Registration of tests in a public database	Yes	No
Oversight that can recall faulty tests	Yes	No
Evaluation of a test’s clinical utility, such as an effect of a test on reducing mortality	No	No

Note: Adapted from Figure 1 in the 2021 article by The Pew Charitable Trusts.¹⁹ Laboratory-developed tests are tests that are designed, manufactured, and used within a single clinical laboratory and are primarily regulated by CLIA, which focuses on laboratory operations and requires that laboratories assess the analytical performance of new tests (FDA 2024,²⁰ Graden 2021²¹).

Abbreviations: CLIA, Clinical Laboratory Improvement Amendments of 1988; FDA, US Food and Drug Administration.

Biologic measurements

Technologies to detect and measure cancer-related changes in circulating tumor DNA (ctDNA) (liquid biopsy) and other analytes have supported MCD assay development (Table 2).^22–52 Proof of principle that ctDNA can detect occult malignancies was first documented by the incidental detection of cancer in women undergoing noninvasive prenatal screening for fetal aneuploidies.⁵³ Most MCD tests are designed to evaluate multiple features of ctDNA, such as mutation profiles, methylation signatures, and fragmentation patterns. The amount of ctDNA shed into the bloodstream is correlated with cancer stage.⁵⁴ Although mechanisms and dynamics behind the release of ctDNA and other tumor components into the circulation have been subjects of intense investigation, whether they merely reflect total tumor burden, tumor biology, or other biologic factors remains unclear.

TABLE 2.

Analytes and technologies used to develop multicancer detection tests.

Analyte	Subanalyte	Measurement technology	Extracted cancer-related features
Cell-free DNA, circulating tumor DNA	DNA sequence	Whole-genome sequencing	Somatic copy number alterations of the DNA sequence, fragment end points, fragment length, allelic imbalance (Jamshidi 2022,22,a Wan 201923,b)
		Whole-genome sequencing followed by targeted next-generation sequencing (Keller 202124,c)	Short and long fragment coverage characteristics, chromosomal arm copy number changes, mitochondrial copy numbers (Cristiano 201925,a), small somatic variants (Jamshidi 202222,a)
		Targeted quantitative real-time polymerase chain reaction	Single nucleotide variants/small insertions and deletions (Cohen 2018,26,a Stackpole 202227,a)
	Methylation of the analyte	Whole-genome bisulfite sequencing	Cancer-specific hypomethylation and hypermethylation and tissue-specific hypomethylation and hypermethylation (Stackpole 202228,a), whole-genome methylation pattern (Jamshidi 202222,a)
		Reduced representation bisulfite sequencing followed by next-generation sequencing	Sequenced methylation profile and copy number profile (Van Paemel 202129,a)
		Sodium bisulfite treatment followed by quantitative real-time polymerase chain reaction	DNA methylation signal (Vrba 202230,a)
		Targeted sodium bisulfite treatment enrichment followed by next-generation sequencing	Methylation block score (Liang 202131,a) Hypomethylation and/or hypermethylation states for cancer and/or tissue-specific methylation patterns (Liu 202032,a)
			Methylation fraction (Liang 2021,31,a Oxnard 2019,33,a Chen 202034,a)
		Immunoprecipitation enrichment followed by quantitative real-time polymerase chain reaction or next-generation sequencing	Methylated fragment sequence (Song 2017,35,a Shen 201836,a)
	DNA copy number aberrations	Next-generation sequencing	Copy number variants (Chan 2013,37,a Keller 202124,c)
Cell-free RNA	miRNA	Quantitative real-time polymerase chain reaction with a gene panel	Gene expression (Vykoukal 202238,b)
Protein	Plasma proteins	Immunoassays with a protein panel	Quantity of protein (Cohen 2018,26,a Lennon 2020,27 a Fahrmann 201939,a)
	Serum proteins	Immunoassays with a protein panel	Quantity of protein (Wen 2015,40,a Wang 201841,a)
Metabolites		Ultra performance liquid chromatography, quadrupole time-of-flight mass spectrometry	Feature annotation based on custom libraries of standards (Fahrmann 201939,a)
Glycans	Glycosaminoglycans	Capillary electrophoresis with laser-induced fluorescence (Gatto 201842,b) ultra highperformance liquid chromatography-tandem mass spectrometry (Bratulic 202243,a)	GAGome features
Extracellular vesicles (EVs)	Total EV RNA miRNA	Quantitative real-time polymerase chain reaction Quantitative real-time polymerase chain reaction	Gene expression (Alen 202344,a) Gene expression (Tengda 201845,b)
	Proteins	Aptamer-based proteomics	Protein expression level (Fahrmann 202046,a)
		Immunoassay for targeted protein	Quantity of protein (Hinestrosa 202247,a), quantification of protein (Hinestrosa 202247,a)
	Proteins of serum EVs	Flow cytometry on isolated exosomes for presence of targeted protein	Count of EVs with target protein (Melo 201548,b)
	mRNA	Targeted nanoString nCounter (nanoString Technologies, Inc.)	Gene expression (Fortunato 202249,b)
Tumor-educated platelets	mRNA	Next-generation sequencing	Differential level of expression (Best 201850,a
Cancer stem cells	mRNA	Quantitative real-time polymerase chain reaction	Gene expression (Tripathi 202151,a)
Circulating tumor cells	Chromosomal aberrations	Locus-targeted fluorescence in situ hybridization assay	Circulating tumor cells and total number of abnormality quantities (Katz 202052,b)

Note: This table summarizes information on technologies and cancer‐related features included in the development of MCD tests. Analytes are derived

from the following blood components: plasma (cell‐free DNA, circulating tumor DNA, cell‐free RNA, protein, metabolites, glycans, extracellular vesicles), serum (protein, extracellular vesicles), platelets (tumor‐educated platelets), red blood cells (cancer stem cells), and peripheral blood mononuclear cells (circulating tumor cells).

Abbreviations: GAGome, a cell‐based library of displayed glycosaminoglycans; miRNA, microRNA; mRNA, messenger RNA.

^aPrimary research article detecting multiple cancers.

^bPrimary research article detecting a single cancer.

^cReview article.

One limitation of the mutational spectrum in ctDNA as biomarkers (measurable analytes found in body fluids that are indicative of an underlying disease, such as cancer) is the biologic signal from aging cells, which interferes with diagnostic interpretation. In people without cancer, normal human cells release DNA into the bloodstream, and these cells may have genetic variants associated with clonal hematopoiesis of indeterminate potential, a common premalignant, albeit low-risk, condition for hematologic malignancy that increases with age, as well as with inflammatory or other premalignant conditions that are cleared through immune surveillance and/or programmed cell death.⁵⁵

MCD assays use other biomarkers in addition to ctDNA, such as circulating tumor cells, cell-free RNA, proteins, metabolites, glycans, extracellular vesicles (most notably exosomes), tumor-educated platelets (platelets modified by tumors), and cancer stem cells (Table 2), in blood and other biospecimens, such as urine, saliva, and cerebrospinal fluid. These biomarkers are components of several tumor-related processes. For example, extracellular vesicles, which are secreted by living cells and may contain DNA, microRNAs, proteins, or lipids, are mediators of intercellular communication.^44,45

MCD assay technologies and algorithms

MCD tests leverage the shared biology of cancer cells of different tissues. Tumors shed various components into the blood or other body fluids that can be used as cancer biomarkers.¹³ MCD tests are complex tests that measure multiple molecular analytes and determine thresholds for positive and negative test results by using sophisticated statistical algorithms that include artificial intelligence-based/machine learning-based modeling. These multidimensional analyses of multicomponent measurements are used to set a threshold for a cancer-positive result based on the concentration and combination of analytes. A diagnostic result (e.g., cancer or not cancer) is applied to clinical samples that are used to train and validate the MCD assay. Artificial intelligence and machine learning may also be used to generate a tissue-of-origin prediction.^26,32

Assay developers continue to modify the biologic and algorithmic components of their tests. To categorize the readiness of MCD testing for clinical use, we mapped publicly reported MCD assays^{22,23,25,27,28,30–32,34–36,39–43,47,52,54,56–74} to their phase of development (Figure 1)²⁶–^61,73,74 based on a five-phase cancer biomarker framework developed by the NCI’s Early Detection Research Network (EDRN).^5,75 Most MCD tests are in early stages of development (Figure 1); their diagnostic performance is trained and validated based on retrospective studies using cohorts predominantly comprised of specimens obtained from individuals with a definitive diagnosis of cancer or from individuals with no known cancer (EDRN phase 2 or 3). Two MCD tests are in the prospective screening phase (EDRN phase 4) to evaluate the accuracy of the test for use in a cancer screening program. Data from prospective clinical trials (EDRN phase 5) are considered essential for evaluating the clinical utility of a tumor biomarker test; such data have not yet been reported for any MCD tests.⁷⁶ Because none of the currently available tests have completed phase 5, it is our collective assessment that MCD tests do not have demonstrated evidence of readiness for routine use in population-based cancer control.

FIGURE 1.

Translation of an MCD assay from the laboratory to clinical practice: stage of development and examples of MCD tests at each stage. This assessment of stage of development is limited to publicly available information. Some studies may not be represented here. The preclinical exploratory phase is an active area of assay development, and we did not attempt to comprehensively represent this work here. Information last reviewed April 24, 2023. MCD indicates multicancer detection.

Early stage cancer detection in nonrandomized prospective studies and RCTs

MCD assay developers report that these tests can detect several types of cancers, as summarized in Table 3. Performance characteristics of these assays to date have primarily relied on banked specimens from diagnosed patients, which provide an optimistically biased assessment of their sensitivity and specificity and are best characterized by RCTs of prospectively screened, asymptomatic individuals.

TABLE 3.

Cancers targeted by multicancer detection assays and their Early Detection Research Network phases of development.

Company	Assay	Technology	EDRN phase	Targeted cancers
				Lung	Colorectal	Breast	Pancreas	Liver	Esophagus	Stomach	Ovary	Prostate	Bladder	Kidney	Uterine	Head and neck	Lymphoma	Leukemia	Plasma Cell	Brain	Anus
Adela Bio	Adela	Cell-free methylated DNA im mu no precipitation-sequencing, cell-free DNA fragmentomics	2	✓	✓	✓	✓						✓	✓				✓
Biological Dynamics	TR(ACE)	Extracellular vesicle proteins, artificial intelligence	2	✓			✓				✓	✓	✓	✓	✓
Bluestar Genomics	Bluestar MCED	Ceii-free DNA 5- hydroxymethylcytosine- sequencing, cell-free DNA fragmentomics	2	✓	✓	✓	✓	✓	✓	✓	✓
Burning Rock	OverC	Deep-targeted bisulfite sequencing (ELSA sequencing)	2	✓	✓		✓	✓	✓		✓
Caris Life Sciences	MiGPSai	Cell-free DNA/cell-free RNA next- generation sequencing, artificial intelligence	3	✓	✓	✓				✓				✓	✓
Datar Cancer Genetics	TriNetra- Pragma	Circulating tumor cell multiplex im mu nocytochemistry	2	✓		✓	✓				✓	✓	✓
Delfi Diagnostics	DELFi	Cell-free DNA fragmentomics	2	✓	✓		✓			✓	✓
Early Diagnostics	cfMethyl- Seq	Cell-free DNA-methylation next- generation sequencing	2	✓	✓			✓		✓
Eiypta	MIRAM- SKY	Ultra high-performance liquid chromatography—tandem mass spectrometry glycosaminoglycans/spectral karyotyping	3			✓				✓		✓	✓	✓		✓	✓	✓		✓
Exact Sciences	CancerSeek	Cell-free DNA polymerase chain reaction—next-generation sequencing, protein markers	4	✓	✓	✓	✓	✓	✓	✓	✓
Grail	Galleri	Targeted cell-free DNA-methylation next-generation sequencing	4	✓	✓		✓	✓	✓	✓	✓		✓			✓	✓		✓		✓
LungLifeAl	LungLB	Circulating tumor cell fluorescence in situ hybridization, imaging artificial intelligence	3	✓		✓		✓						✓
MD Anderson	McaST	Liquid chromatography with tandem mass spectrometry	3	✓	✓	✓	✓				✓
Precision Epigenomics	Sentinel-10	Targeted cell-free DNA-methylation quantitative real-time polymerase chain reaction sequencing	2	✓	✓	✓	✓	✓
Rivela Diagnostics		Extracellular vesicle nucleic acids, proteins, metabolites	2	✓	✓	✓	✓	✓	✓	✓	✓	✓	✓	✓	✓	✓	✓	✓	✓	✓	✓
Singlera Genomics	PanSeer	Targeted cell-free DNA-methylation next-generation sequencing	3	✓	✓			✓	✓	✓
20/20 Gene Systems	OneTest	Circulating cancer antigens, artificial intelligence	3	✓	✓	✓	✓	✓			✓	✓

Note: Checkmarks indicate cancers that are targeted by assays.

Abbreviations: EDRN, the National Cancer Institute’s Early Detection Research Network; ELSA sequencing, enhanced linear-splinter amplification sequencing.

Two prospective screening studies (EDRN phase 4) have been reported.^27,56 The Pathfinder study (ClinicalTrials.gov identifier NCT04241796)⁵⁶ enrolled 6662 individuals from seven US health care systems who were older than 50 years and had no clinical suspicion of malignancy and no history of malignancy or cancer treatment within the past 3 years, although one quarter of participants had a prior history of cancer. All participants were screened with the MCD test in addition to receiving standard-of-care screening. A positive MCD signal was detected in 92 individuals (1.4%), of whom 57 had a false-positive result and 35 had a true-positive result (0.5% of 6621 analyzable results). Among the cancers diagnosed in the 35 volunteers with a positive MCD test, 19 were solid tumors, and 17 were hematologic cancers. Seven of the diagnosed cancers were recurrent disease (20%), and 14 were detected at early (I or II) stages, including seven organ sites for which currently there is no population-based screening: oropharyngeal, pancreas, liver, small intestine, uterus, bone, and hematologic. Hematologic malignancies comprised 57% of early stage diagnoses and thus are a major contributor to overall early stage sensitivity. Most of the 121 volunteers diagnosed with cancer had a negative MCD result (i.e., a false-negative test). Taken together, 228 MCD tests were required to detect one new cancer through the workup triggered by true-positive results, and 473 were required to detect cancer at an early stage.

The DETECT-A study²⁷ enrolled 10,006 women from a US health system aged 65–75 years who were not previously known to have cancer and had high adherence to standard-of-care screening, all of whom received MCD screening. Of 26 cancers (0.26%) first detected by MCD testing (381 tests to detect one new cancer using 9911 confirmed results), eight were detected at early stages (1239 tests to detect one new, early stage cancer). Early stage cancers were detected in three organ sites for which currently there is no recommended screening: ovary, uterus, and thyroid. Three early stage lung cancers were detected, although whether these volunteers were also candidates for lung cancer screening was unclear.

The only RCT is the UK National Health Service Galleri trial (International Standard Randomized Controlled Trial Number 91431511), which has not yet reported primary outcome results. This trial has enrolled more than 140,000 individuals with the goal to assess whether population-based MCD screening, compared with standard-of-care screening for colorectal, breast, and cervical cancer (offered in both arms), reduces the incidence rate of stage III and IV cancers 3–4 years after randomization.⁷⁷ Note that mortality is a secondary end point.

The NCI’s plans to evaluate the clinical utility of MCD tests

Although key trials are ongoing, none have the primary goal of evaluating the impact of population-based MCD screening on cancer mortality. The NCI launched a new Cancer Screening Research Network in January 2024 focused on cancer screening clinical interventional and observational studies.⁷⁸ The first study will be the Vanguard Study,⁷⁹ a pilot feasibility study to test multiple different design assumptions before initiating a definitive RCT to evaluate the clinical utility of MCD tests, likely using cancer-specific mortality as its primary end point.

The Vanguard Study will enroll and randomize asymptomatic individuals to a control arm versus one of two intervention arms, with approximately 8000 participants in each of the three arms. Each intervention arm will evaluate a different MCD assay. Participants in all three arms will be offered standard-of-care screening. This pilot study is expected to inform a future RCT by helping identify the most effective recruitment approaches (particularly for populations that have been historically underrepresented in clinical trials), the diagnostic process after a positive MCD test, and adherence to the RCT protocol for MCD testing and standard-of-care screening.

Unique challenges of MCD tests

Although MCD tests may offer important improvements in screening, they also create new challenges. For example, a positive MCD signal alone is not diagnostic of cancer. Additional procedures are required to localize the source of the positive test signal and make a pathologic diagnosis. Such procedures can include whole-body imaging or other tests, which have imperfect clinical performance. Diagnostic pathways that minimize unnecessary testing have not been determined.

Clinical evidence development frameworks to evaluate MCD test performance are still emerging,^15,16 and the FDA has not yet issued guidance for industry. In a screened population of asymptomatic individuals, the identification of true positives—defined here as pathologically confirmed cancers—depends on the effectiveness of the diagnostic workup. Some true-positive MCD results may not be confirmed because the follow-up diagnostic procedures may not be sensitive enough to detect underlying cancers, thereby creating ambiguity about the clinical meaning of a positive MCD result (Figure 2). That is, the failure to detect cancer after a positive test may represent a false-positive MCD test or a false-negative diagnostic workup. Furthermore, there is no standard definition for a complete diagnostic workup or how much time might elapse from the initial MCD result before a cancer diagnosis might be attributable to the emergence of a new, unrelated cancer. There is uncertainty about how to follow patients who have a negative diagnostic workup after a positive MCD test and concern about the potential for missing cancers that may emerge later.

FIGURE 2.

Possible outcomes from an MCD screening test. Description of possible outcomes from positive and negative MCD signals and diagnostic workup, including benefits (e.g., reduced mortality and morbidity) and harms. Decreased adherence to recommended screening and decreased healthy behaviors may be mediated through false reassurance for true-negative and false-negative MCD screening results in addition to false-negative diagnostic workups. Delayed treatment may result from false-negative MCD screening results. Patient follow-up is unclear for false-negative diagnostic workups. MCD indicates multicancer detection.

In addition, sensitivity relies on the definition of disease, which varies across published studies. For example, disease can be defined as all cancers that the test claims to detect across all stages of disease, or single cancers across all stages, or early stages of all detectable cancers, or early stages of single cancers. Definitions of early stage cancer can range from premalignant disease to presymptomatic stage III cancer. Lack of standardized definitions limits the ability to assess MCD test performance.

Most MCD assays are designed to have high test specificity to reduce the diagnostic evaluations performed because of false-positive results. For example, the DETECT-A study²⁷ chose a 16-fold higher positivity cutoff for the cancer antigen 125 biomarker than was used in the Prostate, Lung, Colorectal and Ovarian (PLCO) screening trial.⁸⁰ Retrospective case–control studies have shown high MCD test specificities of 98%–99%,⁸¹ which compare better with single-organ screening.^82–84 Nonetheless, physical and psychological risks from diagnostic procedures at multiple organ sites may be considerable (Figure 2); and, even with high specificities, repeated testing and increasing adoption of MCD screening by large practices and health care systems over time will lead to large numbers of patients undergoing diagnostic evaluations.

To guide the diagnostic evaluation after a positive screening result, some MCD tests provide a tissue-of-origin prediction on the most probable organ site(s) (origins) of the cancer signal.^26,32 Preliminary reports suggest it may be possible to predict the tumor site accurately within the top two predictions of tissues of origin.⁵⁶ In Pathfinder,⁵⁶ a prospective study evaluating the clinical feasibility of MCD testing with a tissue-of-origin prediction component, the median time to diagnostic resolution was 79 days. Greater than 25% of the 33 participants with a true-positive result needed more than 3 months to receive a definitive cancer diagnosis. Participants with false-positive results—comprising 62% of those with positive results—had a longer time to receive a final diagnosis (median, 162 days for those with a false-positive test vs. 57 days for those with a true-positive result).⁵⁶ Greater than 90% of participants with a positive MCD result underwent imaging, primarily positron emission tomography-computed tomography, computed tomography, and magnetic resonance imaging; and 53% had more than one imaging study. The proportion of patients diagnosed with cancer after a diagnostic workup for a positive MCD result (i.e., the PPV) is about 40%, regardless of whether the MCD test included a prediction for the tissue of origin.^27,56

MCD tests are particularly complex because of the biologic features evaluated, the combination of multiple analyte types and methodologies (Table 2), and the dependency of test thresholds and diagnostic performance on proprietary algorithms. Unlike conventional cancer screening tests, the meaning of a test may change over time, because MCD test companies continue to refine the testing algorithm and/or biologic measurements. Assay refinement may affect study results if changes are made during a clinical trial, thus prospective planning may be needed to support innovation. For example, changes to biologic measurements can be assessed by storing samples and comparing results with a later test version, but sample degradation over time must be taken into consideration in this assessment. Changes to the testing algorithm can be evaluated by comparing clinical performance between an initial version (results returned to patients) and a refined test.⁵⁶

Biases can be introduced by using data to train algorithms that come from patients with newly diagnosed, late stage cancers; oversampling of cancers that do not reflect the underlying population; and controls that are not matched to the cancer cases. As a rule, algorithms must be trained and validated on data sets that include diverse populations to have generalizability to these groups. Data on performance evaluated by or adjusted for risk categories, such as sex, age, race and ethnicity, or inherited germline variants, have not been available to date, limiting understanding of whether specific MCD tests are better suited for specific populations.

Challenges for all cancer screening tests that may be magnified for MCD tests

Effect of screening and early detection on mortality

Clinical trials with adequate statistical power have the potential to show reduced cancer mortality for organ-specific screening tests (e.g., lung cancer [using low-dose computed tomography],^85,86 breast cancer [mammography],⁸⁷ and colorectal cancer [flexible sigmoidoscopy]^88,89) or not (e.g., ovarian cancer [cancer antigen 125 and ultrasound^80,90] and lung cancer [using chest x-ray]⁹¹). For cancer screening to show meaningful population health benefits, an effective treatment for a premalignant or earlier stage cancer must be available.⁹²

A reduction in the incidence of advanced-stage disease (i.e., a stage shift) is a potential end point with which to demonstrate effective screening. The Tomosynthesis Mammographic Imaging Screening Trial (TMIST; ClinicalTrials.gov identifier NCT02616432) is using a reduction in advanced cancer diagnoses as a trial end point, but this end point was justified based on many previous breast cancer screening trials.^93,94 This is important because a reduction in advanced-stage diagnoses may not necessarily translate into a mortality reduction for all cancer types. For example, a patient diagnosed with early stage pancreatic cancer has a lower likelihood of surviving for 5 years than a patient diagnosed with late-stage lymphoma.⁹⁵ Survival gains for individual cancers at a given stage do not equate to survival gains for other cancers, posing a challenge to using stage shift for a large set of cancers as an end point.

For cancers without a current screening test, there is no evidence that reduction in late-stage diagnoses is related to mortality reduction. Some trials have demonstrated reductions in late-stage incidence but not mortality. For example, the UK Collaborative Trial of Ovarian Cancer Screening (International Standard Randomized Controlled Trial Number 22488978) demonstrated a 10% reduction in incidence of late-stage (III and IV) ovarian cancer and a concomitant 9.7% increase in the fraction of early stage (I and II) ovarian cancer in the multimodal screening arm compared with the no-screening control arm, yet it failed to demonstrate that this had any impact on cancer mortality.⁹⁰ Performance metrics based on cancer stage shift may artificially inflate the value of cancer screening using MCD tests.

Furthermore, statistical modeling of stage shift shows that mortality reduction varies considerably by cancer type and stage-specific survival, reinforcing that stage shift across all cancer types may not serve as a reliable surrogate to suggest a mortality benefit for MCD tests.⁹⁶ Although stage shift data can be obtained relatively quickly, a direct assessment of whether MCD tests truly reduce mortality through an RCT is crucial. Such a trial may provide the necessary evidence that a reduction in advanced-stage cancer for certain organs is sufficient to impute mortality benefits.

Sensitivities and specificities of MCD tests

Reported sensitivities of MCD tests for all types of detected cancers across all stages have ranged from 27% to 95%^{25,27,56,81,97} at specificities of 95%–99%. Although the goal of MCD tests is to identify more cancers at an early stage, lower sensitivity has been reported for early stage cancers (approximately 27%–62% for stage I) compared with late-stage cancers (approximately 60%–87% for stage III).^{25,26,32,81,97} Moreover, MCD tests that rely on tumor shedding are unlikely to detect cancer precursor lesions, such as colorectal adenomas and cervical intraepithelial neoplasia grade 3, that would reduce the incidence of those cancers if treated.

Sensitivity typically declines as assays move through stages of development (Figure 1). It is increasingly challenging to detect cancer as tests move through the development and validation stage (in which samples are from individuals with known cancers; EDRN phase 2), to the retrospective/longitudinal stage (in which samples are collected before clinical presentation and the population tested may include more early stage cancers; EDRN phase 3), to the prospective screening stage (in which an imperfectly sensitive clinical diagnostic workup may impair MCD test sensitivity; EDRN phases 4 and 5). Caution is advised when comparing the performance of tests at different stages of development.⁹⁷

In addition, for cancers that have recommended screening, MCD early stage test sensitivity (e.g., approximately 5% for stage I breast cancer and approximately 30% for stage I colorectal cancer³²) is far lower for some MCD tests than for standard-of-care screening, and many cancers would be missed using MCD screening alone.^27,56,81 Standard-of-care screening is recommended by test developers regardless of MCD results. However, patients may feel falsely reassured by a negative result overall or for an organ for which there is standard-of-care screening, which may lead to lower adherence to recommended screening or other preventive health care behaviors (Figure 2).⁷⁹ Nonadherence to standard-of-care colorectal and cervical cancer screenings will reduce their primary benefit of decreasing cancer incidence through detection and timely treatment of precancer abnormalities, which MCD tests have not been shown to detect. False reassurance from a false-negative MCD result may also lead to delayed treatment (Figure 2). Patients have reported anxiety after receiving both true-positive and false-positive MCD results.⁹⁸ The effect of anxiety on future health behaviors like cancer screening is unknown.

Limitation of MCD screening benefit by overdiagnosis

Screening asymptomatic individuals may result in overdiagnosis—the detection of slow-growing cancers that would not lead to morbidity or death yet expose patients to ineffective or harmful treatments. Overdiagnosis can artificially lengthen population-level cancer survival statistics by inflating the length of time patients live with known disease, creating lead-time bias.⁹⁹ The implementation of ultrasound screening for thyroid cancer in South Korea, for example, led to a 1500% increase in incidence, driven by the slower growing subtype—papillary thyroid cancer¹⁰⁰—without a corresponding drop in mortality from thyroid cancer, yet many were exposed to harms from invasive surgery.¹⁰¹

Assay developers have suggested that the detection of ctDNA identifies more aggressive tumors that may benefit from treatment, whereas undetected tumors are more indolent.⁵⁴ Alternatively, those may be the aggressive tumors destined to be lethal regardless of stage at diagnosis and receipt of treatment. Prospective outcome data on mortality are required to draw firm conclusions about the curability of cancers detected through MCD tests.

MCD screening and health care disparities

If shown to be effective in clinical trials, MCD assays for cancer screening have the potential to reduce health care disparities because standard-of-care cancer screening tests have disproportionately lower uptake among medically underserved populations.¹⁰² However, although blood tests are easy to implement, lack of access to diagnostic testing after a positive screening test has the potential to worsen cancer disparities. For example, women with lower socioeconomic status have a lower rate of completing the recommended workup when breast abnormalities are seen on mammograms.¹⁰³

To improve population health, an MCD screening program will need to be administered in a way that all population groups—including underrepresented racial and ethnic groups, sexual and gender minorities, socioeconomically disadvantaged populations, and underserved rural populations—have equal access to both MCD testing and to appropriate diagnostic evaluation and treatment of detected cancers. Achieving health equity is a major challenge for even well established medical interventions and requires overcoming cultural, social, economic, and structural barriers to health care delivery. Within the US health care system, lower cancer screening rates are associated with factors such as lack of health insurance, lower income, greater travel distance to sites of cancer screening, and lower educational attainment.⁹ If MCD tests are made widely available without addressing these barriers, MCD testing may worsen—rather than improve—existing health disparities.

Translation into clinical practice

MCD tests offer a potential advancement in cancer screening by using a single biospecimen to screen for multiple cancers, including many that lack recommended screening. However, before implementing MCD tests in clinical practice, clinicians, health care systems, and patients need evidence that these tests have a meaningful health benefit. To date, no clinical practice guidelines for MCD use have been published, the FDA has not authorized marketing of any MCD tests, and coverage by health plans is limited. Nonetheless, clinicians and patients are faced with decisions about using MCD tests now.

Clinicians and the public at large should appreciate the many complexities and uncertainties about using MCD tests for cancer screening (Table 4) and should carefully evaluate claims about the use of these tests for early detection and reduction of morbidity and mortality. In the meantime, clinicians can help resolve uncertainties about MCD-based cancer screening by enrolling patients into research studies, and caution is advised for the use of MCD tests outside of research settings. At a minimum, clinicians and patients contemplating MCD testing should discuss its potential benefits, harms, and uncertainties through a process of shared decision making.¹⁰⁴ However, much more evidence is needed to inform these discussions and to determine the most effective strategies for implementing MCD testing in routine clinical care.

TABLE 4.

Opportunities, evidence gaps, and risks using multicancer detection tests for cancer screening.

• Multicancer detection (MCD) tests offer the potential for using a single, noninvasive blood test to screen for multiple cancers, including cancers without guideline-recommended screening tests (e.g., pancreatic cancer).

• Currently, there is insufficient evidence to support MCD testing as a cancer screening tool that can reduce mortality.

• MCD testing poses several risks, including harms from multiple diagnostic procedures in multiple organs in patients with a positive test and potential harms from overdiagnosis and overtreatment.

• MCD screening has the potential to lower patient adherence to recommended, standard-of-care screening.

• The most effective strategies for evaluating positive MCD screening results and for implementing MCD testing in routine clinical care remain unknown.

• The effect of MCD-based cancer screening on health care disparities is unknown.

• There is no clinical trial evidence to support MCD screening in clinical care.

Future directions

The Vanguard Study and future RCT will rigorously assess the clinical utility of MCD technologies as cancer screening tools. Through these studies and others,⁷⁷ as well as those that leverage real-world data,¹⁰⁵ the medical community can determine whether various MCD tests may be appropriate to implement and receive guidance on how to approach the diagnostic workup. If MCD-based screening can successfully identify patients at earlier stages than currently possible, new treatments or refinements to existing approaches may be needed to reduce mortality. Detecting tumor components with MCD testing in patients without symptoms could have prognostic significance and possibly influence cancer staging, as has been suggested for liquid biopsy in patients diagnosed with cancer.¹⁰⁶ People at high risk of new primary malignancies, such as those with a hereditary cancer syndrome or cancer history, have the potential to benefit most from MCD screening. As new approaches for effective, individualized cancer risk assessment are developed, matching the optimal test for each person at the best time point with tailoring to their risk profile will help to optimize the use of this technology.

Limitations

This review has several limitations. First, an evidence-based review of MCD technologies was not performed. Second, some studies may have been missed. Third, assignment of MCD assays to a phase of development (Figure 1 and Table 3) was based on NCI expert knowledge, but there could be some disagreement by test developers about phase determinations. Fourth, because of space limitations, not all possible benefits and risks of MCD technologies were described.

CONCLUSIONS

MCD tests offer the potential for using a noninvasive screening test to reduce cancer-related morbidity and mortality. Before implementing population-wide use of MCD tests, evidence from prospective clinical trials is needed regarding the effectiveness of MCD tests for early cancer detection across different populations; the appropriate diagnostic evaluations for positive results and negative workups; and the potential harms of overdiagnosis, overtreatment, and lower adherence to recommended screening. The NCI is funding a new research network and will support studies to address the many evidence gaps and accelerate knowledge in this important area.^78,79