Multiyear Clinical Outcomes of Cancers Diagnosed Following Detection by a Blood-Based Multicancer Early Detection Test

Guideline-endorsed standard-of-care (SoC) cancer screening tests are available for only a few cancer types (1–5); most incident cancers have no recommended SoC screening and are diagnosed in the course of clinical care (6). Many cancers are first diagnosed in symptomatic individuals with advanced disease, when treatment with curative intent is not possible and long-term survival is uncommon (6). Advances in bioanalytic technologies have facilitated the development of blood-based multicancer early detection (MCED) tests that may improve early cancer detection of multiple types, including uncommon cancers and those that lack SoC screening options (7–14). The combination of enhanced performance in novel MCED tests and continual improvements in cancer treatment offers the opportunity for significant reduction in cancer mortality that some prior cancer screening trials have failed to demonstrate (15, 16). However, concerns exist about the possibility of overdiagnosis and associated harms of overtreatment. Additionally, there is currently an absence of data on multiyear clinical outcomes for MCED-detected cancers in a prospective setting.

DETECT-A (Detecting cancers Earlier Through Elective mutation-based blood Collection and Testing), which enrolled participants from 2017 to 2019, is the first large prospective interventional trial to evaluate an MCED blood test (CancerSEEK). CancerSEEK was evaluated in 9,911 women aged 65 to 75 years with no personal history of cancer (13). Following a positive CancerSEEK test result, a clinical evaluation including full-body diagnostic IV contrast computed tomography and positron emission tomography (PET-CT) imaging and/or other clinically indicated procedures was used to identify and localize the suspected cancers. One of the primary aims of DETECT-A was to investigate whether an MCED test can be safely incorporated into routine care to detect cancers not previously identified through other means and whether the test may be used to intervene with therapy with the intent to cure (13). DETECT-A demonstrated that the MCED test could detect a variety of cancer types, including early-stage cancers without recommended SoC screening modalities, while maintaining high specificity. Ninety-six cancers were diagnosed, including 26 cancers first identified by CancerSEEK, 24 first identified by SoC screening, and 46 first identified by other means. As a single-arm feasibility study, DETECT-A was not designed to assess the mortality benefit from MCED testing relative to other screening modalities or no screening. Favorable clinical outcomes among participants with MCED-detected cancers are a prerequisite for determining clinical benefit and can be conclusively demonstrated with results obtained from large, randomized trials. The present observational study evaluated the resulting treatments and longitudinal clinical outcomes in participants with cancers diagnosed as a result of a positive CancerSEEK test, with a median follow-up time of more than 4 years from initial CancerSEEK testing. The characteristics of the stage I cancers identified by the CancerSEEK test were also evaluated to assess their clinical significance and the likelihood that a clinical benefit may have been derived by MCED detection and subsequent treatment.

The study was approved by the Institutional Review Boards for Human Research at the Geisinger Health System (Geisinger; #2017-0268) and the Johns Hopkins Medical Institutions (#00119844) and was compliant with U.S. Common Rule and The Health Insurance Portability and Accountability Act. Written, informed consent was obtained from all participants. Clinical follow-up data were restricted to a limited data set and were protected in adherence to applicable privacy laws and/or regulations and de-identified to the best extent possible to protect confidentiality.

In the DETECT-A study, nine cancer types were diagnosed in 26 participants whose cancers were first detected by CancerSEEK. Longitudinal follow-up data, including cancer diagnosis, treatment, treatment response, cancer status, and mortality (cancer-related and all-cause) were available for all participants and were extracted from electronic medical records through November 2022. Participants were determined to be cancer-free if records indicated achievement of full remission or absence of detectable cancer. Data collection took place at a median of 3.7 years following cancer diagnosis (IQR: 3.3–4.0) and a median of 4.4 years (IQR: 4.1–4.6) following study enrollment. All clinical follow-up data were restricted to a limited data set, including data protected under applicable privacy laws and/or regulations. To determine if the stage I cancers were clinically significant, their presenting histopathological, morphological, and clinical characteristics were examined.

The data generated in this study are available upon request from the corresponding author.

Primary cancer organ sites, stages at diagnosis, treatments, and clinical status (as of November 2022) of the 26 participants whose cancers were first detected by CancerSEEK are included in Table 1.

Ten (38%) participants received nonsurgical treatment; 14 (54%) underwent surgery alone or surgery with adjuvant/neoadjuvant chemotherapy, radiation, or hormone therapy. Treatment was offered but declined by one participant and was unknown for another.

Thirteen of 26 participants (50%) with cancers first detected by the blood test were alive and cancer-free through November 2022, including individuals with ovarian (4), uterine (2), thyroid (1), breast (1), colorectal (2), and lung (3) cancers. Seven of 13 [54%, 95% CI (29%, 77%)] participants with cancers without SoC screening modalities were alive and cancer-free, whereas 6 of 13 [46%, 95% CI (23%, 71%)] participants with breast, lung, or colorectal cancers (cancer types with available SoC screening modalities) were alive and cancer-free. All treated patients with stage I (5) and stage II (3) cancers remained alive and cancer-free, as did three (out of eight) patients with stage III cancers and two (out of eight) patients with stage IV cancers (Table 1). Eligibility for surgical treatment was significantly associated with favorable multiyear outcomes (P = 0.0002, Fisher’s exact test). Specifically, 12 out of 14 [86%, 95% CI (60%, 96%)] participants who underwent surgery alone or surgery with adjuvant/neoadjuvant chemotherapy, radiation, or hormone therapy remained alive and cancer-free, compared to 1 of 12 [8%, 95% CI (2%, 35%)] participants who did not receive surgical treatment. Review of the five stage I cancers diagnosed as a consequence of MCED testing revealed that the cancers [thyroid, uterine (2), lung, and ovarian] exhibited characteristics of clinically aggressive disease. These findings are summarized in Table 2. Notably, the thyroid cancer was a medullary carcinoma, commonly thought to be aggressive and treated by total thyroidectomy in the absence of distant metastases (17). Similarly, one of the observed uterine cancers was a leiomyosarcoma, a rare and typically aggressive cancer for which surgery is generally indicated (18). The other three stage I cancers were a grade 2 (moderately differentiated) endometrial adenocarcinoma, a grade 3 (poorly differentiated) non–small cell lung cancer, and a grade 1 (well-differentiated) ovarian cancer of large size (19.1 cm greatest dimension).

In this study, multiyear follow-up of patients with MCED-detected cancers in DETECT-A revealed that 50% were alive and cancer-free more than 4 years after study enrollment. Multiyear cancer-free survivors included patients diagnosed with cancers with and without SoC screening options. While these data are preliminary, they reveal patterns with important implications for the development and clinical implementation of MCED tests. Favorable outcomes were observed in patients diagnosed with early-stage cancer and/or cancers eligible for surgical therapy, emphasizing the importance of early cancer detection of solid tumors. Favorable outcomes were also seen in three of eight patients diagnosed with stage III cancers, suggesting that while less common, favorable outcomes are achievable when cancers exhibit regional spread but are detected before the development of distant metastases.

One common concern about cancer screening is the possibility of overdiagnosis, the detection of indolent, potentially clinically insignificant cancers that results in unnecessary and potentially harmful care (19). Progress has been made to mitigate this challenge with prostate cancer screening (20, 21). For cancers without SoC screening, this has not been a prominent issue because these cancers are typically clinically apparent, although there are certain cancers for which aggressive therapy is not uniformly indicated. For example, kidney tumors less than 3 cm in diameter and papillary thyroid carcinomas are frequently monitored with active surveillance without immediate treatment (22, 23). It has been suggested that MCED tests may preferentially detect large and/or more aggressive cancers, which often exhibit increased levels of circulating tumor DNA associated with frequent mitotic activity, cell death, and vascularity (7, 24). Additionally, in the UK Collaborative Trial of Ovarian Cancer Screening, early-stage cancers detected by blood-based single-protein biomarker (CA-125) screening appeared more likely to be aggressive compared to early-stage cancers detected in the absence of screening (15, 16). Understanding whether CancerSEEK-detected early-stage cancers were neither unusually indolent and clinically insignificant nor unexpectedly aggressive and capable of evading successful treatment was therefore important. As accepted definitions for clinically significant cancers are not available for most cancer types, we examined the decisions made by treating physicians and the histologic and clinical characteristics of the stage I tumors. Using qualitative judgment, we concluded that the stage I cancers diagnosed as a result of a positive CancerSEEK test appear to have the features of cancers that would typically be treated with curative intent and likely represent clinically significant cancers. The observed favorable outcomes support the likelihood that interventions were timely, appropriate, and beneficial and that the cancers did not have inherently poor prognoses in consideration of available treatments. However, larger prospective studies will be required to broadly characterize the features of MCED-detected early-stage cancers.

The DETECT-A study included women aged 65 to 75 from a single healthcare system. Therefore, DETECT-A findings should be confirmed in multicenter studies more representative of a screening population through the inclusion of multiple sexes, a broader age range, and a more racially and ethnically diverse cohort. Additional limitations of this study include the relatively small number of cancers and the inability to definitively link the MCED test to a clinical benefit in these patients. As with any screen-detected cancer, it is not possible to know with certainty the natural history of disease in the absence of the MCED test. The MCED-detected cancers may have been detected by other means at a similar stage and outcomes may have been favorable even with detection at later stages. We therefore relied on the population-level associations between cancer type, stage, grade and common clinical treatments and outcomes to ascertain a reasonable likelihood of clinical benefit from MCED test detection (25). As DETECT-A was a single-arm feasibility study, without a comparison group, we were unable to determine whether there was a mortality reduction in relation to SoC screening. The results reported herein demonstrate the potential clinical value of MCED testing. These results, demonstrating that early detection via MCED testing can facilitate curative treatment for cancers without SoC screening, can be considered a prerequisite, necessary for the success of MCED, but not sufficient to definitively establish the clinical utility of MCED testing. Larger prospective studies have been planned to determine whether MCED tests can broadly shift cancer detection to earlier stages for improved clinical outcomes, while limiting overdiagnosis and overtreatment.

In this study, a significant proportion of patients with cancer detected by MCED testing were alive and cancer-free several years after treatment. Despite the absence of SoC screening modalities for 50% of the detected cancers, a majority of these patients remained alive and cancer-free more than 4 years following a positive MCED test result. In addition, early-stage detection was associated with favorable outcomes, with 8/8 treated patients with stage I and II cancers remaining alive and cancer-free. These findings provide preliminary evidence of the potential of MCED testing to be an effective screening tool for detecting cancers without SoC screening modalities and complementing SoC cancer screening.

A.H. Buchanan reports grants from Exact Sciences during the conduct of the study; grants from Freenome Holdings, Inc. outside the submitted work; and equity stake in MeTree and You, Inc. A.M. Lennon reports other support from Exact Science during the conduct of the study; personal fees from Exact Science outside the submitted work; in addition, A.M. Lennon has a patent for CancerSEEK licensed and with royalties paid from Exact Science. O.A. Choudhry reports employment and stock ownership at Exact Sciences Corporation. P.Z. Elias reports employment and stock ownership at Exact Sciences Corporation. S.P. Rego reports employment and stock ownership at Exact Sciences Corporation. J.R. Sadler reports employment and stock ownership at Exact Sciences Corporation. J. Roberta reports employment and stock ownership at Exact Sciences Corporation. Y. Zhang reports Y. Zhang reports employment and stock ownership at Exact Sciences Corporation. D.D. Flake II reports employment and stock ownership at Exact Sciences Corporation. Z.M. Salvati reports grants from Exact Sciences during the conduct of the study. E.S. Wagner reports grants from Exact Sciences Corporation during the conduct of the study; grants from Freenome Holdings, Inc. outside the submitted work. E.K. Fishman reports other support from Exact Science during the conduct of the study. N. Papadopoulos reports other support from Thrive/EXACT Sciences, Personal Genome Diagnostics, NeoPhore, CAGE Pharma, ManaTbio, Receive royalties from licensings, and Haystack/Quest during the conduct of the study; in addition, N. Papadopoulos has a patent for PCT/US2018/045669 with royalties paid from EXACT. T.M. Beer reports employment and stock ownership at Exact Sciences Corporation; grants from Delfi, Adela, Freenome, and Grail outside the submitted work.

A.H. Buchanan: Writing–original draft, project administration, writing–review and editing. A.M. Lennon: Conceptualization, formal analysis, writing–original draft, project administration, writing–review and editing. O.A. Choudhry: Conceptualization, formal analysis, writing–original draft, project administration, writing–review and editing. P.Z. Elias: Conceptualization, formal analysis, methodology, writing–original draft, writing–review and editing. S.P. Rego: Conceptualization, formal analysis, methodology, writing–original draft, writing–review and editing. J.R. Sadler: Formal analysis, writing–review and editing. J. Roberta: Conceptualization, formal analysis, investigation, methodology, writing–original draft, writing–review and editing. Y. Zhang: Formal analysis, investigation, project administration, writing–review and editing. D.D. Flake: Formal analysis, investigation, project administration, writing–review and editing. Z.M. Salvati: Formal analysis, investigation, writing–review and editing. E.S. Wagner: Investigation, project administration, writing–review and editing. E.K. Fishman: Formal analysis, investigation, project administration, writing–review and editing. N. Papadopoulos: Writing–review and editing. T.M. Beer: Conceptualization, methodology, writing–original draft, writing–review and editing.

The authors are grateful for the contributions of the Geisinger team, including Ashley Honushefsky, Kathleen Sheridan, and Carroll Walter, for project administration. The Johns Hopkins University team for the development of the CancerSEEK test: Joshua Cohen, Chrisopher Douville, Isaac Kinde, Allison Klein, Cristian Tomasetti, Ralph Hruban, Kenneth Kinzler, and Bert Vogelstein. Biostatistical support was provided by Nishitha Therala and Satish Seerapu, and writing/editorial assistance was provided by Carolyn Hall and Feyza Sancar (Exact Sciences). The authors would also like to thank the participants who generously participated in the original sample collection studies. This study was funded by Exact Sciences Corporation.