Multi-Cancer Early Detection Tests: Current Progress and Future Perspectives

Advances in cancer screening practices have allowed for detection of cancer precursors and earlier detection of cancer, which has led to improvements in patient outcomes and changes in prevention and treatment options. This is particularly true for cancers where precancerous lesions can be detected through routine screening and removed (e.g., colonic polyps, cervical intraepithelial neoplasia). The majority of cancers, however, do not have recommended screening modalities available, highlighting a significant gap in public health practice. A recent area of research interest, in an attempt to begin to close the screening gap, has been the development of multi-cancer early detection (MCED) tests, which produce a signal when a cancer is detected but do not provide a definitive diagnosis. Several MCED tests have appeared on the market. Briefly, different MCED tests measure different analytes (e.g., mutation profiles, methylation signatures) and have the potential to detect cancers in various tissues. Some MCED tests also claim the ability to predict the specific site of origin that can facilitate the diagnosis of cancer. Ideally, MCED tests would detect a patient's cancer early and while the patient's disease is still in an early and curable stage, thus improving patient outcomes.

Population dissemination of affordable MCED tests has the potential to reduce disparities in cancer screening and diagnosis. Compared with White individuals, Black individuals are more likely to present with a diagnosis of distant stage disease for a number of cancers for which screening is common, including breast, colorectal, lung, prostate, and cervical cancer (1). It is not clear if this represents disparities in access to screening or differences in the aggressiveness of the cancer itself. Racial disparities in cancer screening have been consistently noted. For example, Black patients are less likely to be eligible for lung cancer screening, despite being at higher risk for lung cancer, and this may not be addressed in the recent revision to the recommendations intended to expand eligibility (2, 3). Furthermore, though Black and White women undergo breast cancer screening at similar rates, Black women are more likely to be diagnosed with advanced disease (4). Black individuals also have elevated rates of several highly lethal cancers, compared with White individuals, that do not have recommended screening, including pancreatic, liver, and stomach cancer, and Asian and Hispanic individuals also have elevated rates, compared with Whites, of several cancers, including liver and stomach cancer (1, 4–6).

In this issue of Cancer Epidemiology, Biomarkers, and Prevention, an example of the potential benefits of early diagnosis with MCED tests are presented by Clarke and colleagues (7), who modeled the effect of MCED screening by race/ethnicity and gender. Specifically, they simulated cancer stage shifting by assuming that the stage IV metastatic cancers were detected as stage III (in model 1) and stage I, II, or III (in model 2) for 20 cancer types to estimate changes in patient outcomes using Surveillance, Epidemiology, and End Results Program Data. With this stage shifting, the models showed an estimated reduction in mortality of 13% to 14% or 21% to 23%, depending on model assumptions, and were similar across race/ethnicity and gender. However, because some groups, especially Black men, have elevated mortality rates, the same proportional increase would lead to greater absolute reductions in mortality for them as compared with other groups. Although a reduction in cancer-related disparities is clearly a laudable goal, these data raise questions about the magnitude of the stage shift achieved with MCED tests, the level of potential benefit associated with staging shifting, and the looming unknowns associated with the use of new screening modalities, such as MCED tests.

The majority of the data informing MCED test performance and stage shift is derived from case–control and simulation studies. The magnitude of the impact on outcome from stage shift may depend on the cancer type, and the impact of stage shift may not be equivalent across all cancers. To truly evaluate sensitivity and a subsequent stage shift, an understanding of test performance in an asymptomatic population is needed, but this is not yet known. A prospective study assessing an asymptomatic population, comparing recommended screening modalities with MCED test performance, may inform the ability of an MCED test to shift cancer stage. The Clarke and colleagues article (7) assumed all stage IV cancers were detected at earlier stages with the MCED test, which is likely overly optimistic. For example, in the NLST, during the annual screening phase, there were still 14% of cancers in the LDCT arm that were stage IV, compared with 36% in the post-screening phase. Furthermore, compared with the chest radiography arm, the reduction in stage IV disease during the trial was only about 33%. This was despite the fact that test sensitivity of LDCT was very high for all stages of cancer.

The model used by Clarke and colleagues (7) and similar models of the effect of stage shifts, are essentially guaranteed to show a mortality benefit associated with stage shifts, as the models use stage-specific population survival rates to estimate the effect on mortality. For screening tests that detect cancer early but fail to detect cancer precursors, a stage shift has been thought to be a necessary condition for a mortality reduction, but not a sufficient one. Within a target organ, tumors that present in advanced stage may differ by grade, histology or other factors than those that clinically present in early stage. If those advanced-stage tumors were detected early, they may not have the same survival rates as clinically detected early-stage cases.

This question can be evaluated by examining existing screening practices. For example, a study of the Norwegian Breast Cancer Screening Program found that, following the institution of biennial mammography screening for women ages 50 to 69, incidence of both localized and more advanced disease increased, from 63.9 per 100,000 to 141.2 per 100,000 and 86.9 per 100,000 to 117.3 per 100,000, respectively (8). These data suggest that screening may identify a higher frequency of early disease without reducing advanced disease. Furthermore, even if cancers are detected earlier, cancer-specific mortality may not change in response to screening (9). Real-world data from breast cancer screening data may be in conflict with simulation studies, which may be similarly reflected in other screening practices. Refining and verifying simulation modeling to real-world applications is needed to better evaluate the potential outcomes of stage shifting cancers.

Several different MCED tests are currently under development (10–14); however, there is little clarity on the best practices for use for cancer screening. Studies have demonstrated that MCED assays can detect a signal for cancer, but the extent of the diagnostic work-up and the clinical utility are not known. The lone example of MCED test follow-up comes from the PATHFINDER study. The patients and study investigators were returned the results of the MCED test and diagnostic resolution (e.g., follow-up) was measured. Interim results reported 92 positive tests among the study participants, for which 29 had no diagnostic resolution (15). At this time, there is insufficient follow-up for those patients who are test positive but have no diagnostic resolution. For those who are test negative and have no follow-up, it remains unknown when these individuals should be re-evaluated, and, if so, how frequently, or what this means for traditional screening practices. Until these barriers are overcome, the usefulness of MCED tests may be questionable.

In comparison with MCED tests, confirmation of a cancer diagnosis from a single screening modality can be relatively straightforward (e.g., follow-up CT scans and possible biopsy of lung nodules identified by LDCT). This can be the case with a positive result of an MCED test that has accurate tissue of origin (TOO) prediction. For example, the presence of an APC mutation is indicative of colorectal cancer, which can be confirmed with a colonoscopy. However, for many TOOs, the appropriate procedures for follow-up, how long to continue to probe the individual with diagnostic procedures, the likelihood of incidental findings (16, 17) and the course of action when no cancer is detected are not known. Finally, and most importantly, it is still completely unknown whether or not a mortality benefit, if any, exists with MCED-based screening. A randomized trial with an appropriate cancer mortality endpoint is needed to answer this question.

MCED tests hold promise for reducing the burden of cancer at the population level, but many unknown issues remain. The MCED test used for cancer screening needs more rigorous evaluation before widespread use, as their harms are not currently known. MCED tests could potentially fill a substantial gap in cancer screening, particularly in minority and underserved populations, but equal access to MCED tests for screening would need to be established.

No disclosures were reported.

The views presented here are that of the authors and should not be interpreted as representing the viewpoint of the U.S. Department of Health and Human Services, the National Institutes of Health, or the National Cancer Institute.