Circulating Tumor DNA Analysis Guiding Adjuvant Therapy in Stage II Colon Cancer.

BACKGROUND: The role of adjuvant chemotherapy in stage II colon cancer continues to be debated. The presence of circulating tumor DNA (ctDNA) after surgery predicts very poor recurrence-free survival, whereas its absence predicts a low risk of recurrence. The benefit of adjuvant chemotherapy for ctDNA-positive patients is not well understood. METHODS: We conducted a trial to assess whether a ctDNA-guided approach could reduce the use of adjuvant chemotherapy without compromising recurrence risk. Patients with stage II colon cancer were randomly assigned in a 2:1 ratio to have treatment decisions guided by either ctDNA results or standard clinicopathological features. For ctDNA-guided management, a ctDNA-positive result at 4 or 7 weeks after surgery prompted oxaliplatin-based or fluoropyrimidine chemotherapy. Patients who were ctDNA-negative were not treated. The primary efficacy end point was recurrence-free survival at 2 years. A key secondary end point was adjuvant chemotherapy use. RESULTS: Of the 455 patients who underwent randomization, 302 were assigned to ctDNA-guided management and 153 to standard management. The median follow-up was 37 months. A lower percentage of patients in the ctDNA-guided group than in the standard-management group received adjuvant chemotherapy (15% vs. 28%; relative risk, 1.82; 95% confidence interval [CI], 1.25 to 2.65). In the evaluation of 2-year recurrence-free survival, ctDNA-guided management was noninferior to standard management (93.5% and 92.4%, respectively; absolute difference, 1.1 percentage points; 95% CI, -4.1 to 6.2 [noninferiority margin, -8.5 percentage points]). Three-year recurrence-free survival was 86.4% among ctDNA-positive patients who received adjuvant chemotherapy and 92.5% among ctDNA-negative patients who did not. CONCLUSIONS: A ctDNA-guided approach to the treatment of stage II colon cancer reduced adjuvant chemotherapy use without compromising recurrence-free survival. (Supported by the Australian National Health and Medical Research Council and others; DYNAMIC Australian New Zealand Clinical Trials Registry number, ACTRN12615000381583.).

Revisiting the tumorigenesis timeline with a data-driven generative model.

Cancer is driven by the sequential accumulation of genetic and epigenetic changes in oncogenes and tumor suppressor genes. The timing of these events is not well understood. Moreover, it is currently unknown why the same driver gene change appears as an early event in some cancer types and as a later event, or not at all, in others. These questions have become even more topical with the recent progress brought by genome-wide sequencing studies of cancer. Focusing on mutational events, we provide a mathematical model of the full process of tumor evolution that includes different types of fitness advantages for driver genes and carrying-capacity considerations. The model is able to recapitulate a substantial proportion of the observed cancer incidence in several cancer types (colorectal, pancreatic, and leukemia) and inherited conditions (Lynch and familial adenomatous polyposis), by changing only 2 tissue-specific parameters: the number of stem cells in a tissue and its cell division frequency. The model sheds light on the evolutionary dynamics of cancer by suggesting a generalized early onset of tumorigenesis followed by slow mutational waves, in contrast to previous conclusions. Formulas and estimates are provided for the fitness increases induced by driver mutations, often much larger than previously described, and highly tissue dependent. Our results suggest a mechanistic explanation for why the selective fitness advantage introduced by specific driver genes is tissue dependent.

Assessing aneuploidy with repetitive element sequencing.

We report a sensitive PCR-based assay called Repetitive Element AneupLoidy Sequencing System (RealSeqS) that can detect aneuploidy in samples containing as little as 3 pg of DNA. Using a single primer pair, we amplified ∼350,000 amplicons distributed throughout the genome. Aneuploidy was detected in 49% of liquid biopsies from a total of 883 nonmetastatic, clinically detected cancers of the colorectum, esophagus, liver, lung, ovary, pancreas, breast, or stomach. Combining aneuploidy with somatic mutation detection and eight standard protein biomarkers yielded a median sensitivity of 80% in these eight cancer types, while only 1% of 812 healthy controls scored positive.

Evaluating the impact of multicancer early detection testing on health and economic outcomes: Toward a decision modeling strategy.

Emerging data provide initial support for the concept that a single, minimally invasive liquid biopsy test, performed in conjunction with confirmatory radiologic or other diagnostic testing, when indicated, could be deployed on a broad scale to screen individuals for multiple types of cancer. Ideally, such a test could do this in a way that yields a clinically important percentage of true-positive indications of cancer while minimizing false-positive signals. Modern decision modeling approaches can and should be deployed to investigate the health and economic consequences of such multicancer early detection (MCED) testing within defined at-risk populations. In this paper, through small-scale analyses involving 3 hypothetical MCED-detectible cancers, the authors illustrate the potential for MCED testing to be cost-effective, along with the pivotal role of test-induced stage shift on results. The time is ripe for additional, prospective investigations of the clinical value of MCED testing, the benefits versus the risks for screened populations, and the overall projected impact on health outcomes and costs over time.

Cell division rates decrease with age, providing a potential explanation for the age-dependent deceleration in cancer incidence.

A new evaluation of previously published data suggested to us that the accumulation of mutations might slow, rather than increase, as individuals age. To explain this unexpected finding, we hypothesized that normal stem cell division rates might decrease as we age. To test this hypothesis, we evaluated cell division rates in the epithelium of human colonic, duodenal, esophageal, and posterior ethmoid sinonasal tissues. In all 4 tissues, there was a significant decrease in cell division rates with age. In contrast, cell division rates did not decrease in the colon of aged mice, and only small decreases were observed in their small intestine or esophagus. These results have important implications for understanding the relationship between normal stem cells, aging, and cancer. Moreover, they provide a plausible explanation for the enigmatic age-dependent deceleration in cancer incidence in very old humans but not in mice.

Prognostic significance of postsurgery circulating tumor DNA in nonmetastatic colorectal cancer: Individual patient pooled analysis of three cohort studies.

Studies in multiple solid tumor types have demonstrated the prognostic significance of ctDNA analysis after curative intent surgery. A combined analysis of data across completed studies could further our understanding of circulating tumor DNA (ctDNA) as a prognostic marker and inform future trial design. We combined individual patient data from three independent cohort studies of nonmetastatic colorectal cancer (CRC). Plasma samples were collected 4 to 10 weeks after surgery. Mutations in ctDNA were assayed using a massively parallel sequencing technique called SafeSeqS. We analyzed 485 CRC patients (230 Stage II colon, 96 Stage III colon, and 159 locally advanced rectum). ctDNA was detected after surgery in 59 (12%) patients overall (11.0%, 12.5% and 13.8% for samples taken at 4-6, 6-8 and 8-10 weeks; P = .740). ctDNA detection was associated with poorer 5-year recurrence-free (38.6% vs 85.5%; P < .001) and overall survival (64.6% vs 89.4%; P < .001). The predictive accuracy of postsurgery ctDNA for recurrence was higher than that of individual clinicopathologic risk features. Recurrence risk increased exponentially with increasing ctDNA mutant allele frequency (MAF) (hazard ratio, 1.2, 2.5 and 5.8 for MAF of 0.1%, 0.5% and 1%). Postsurgery ctDNA was detected in 3 of 20 (15%) patients with locoregional and 27 of 60 (45%) with distant recurrence (P = .018). This analysis demonstrates a consistent long-term impact of ctDNA as a prognostic marker across nonmetastatic CRC, where ctDNA outperforms other clinicopathologic risk factors and MAF further stratifies recurrence risk. ctDNA is a better predictor of distant vs locoregional recurrence.

Circulating tumor DNA dynamics and recurrence risk in patients undergoing curative intent resection of colorectal cancer liver metastases: A prospective cohort study.

BACKGROUND: In patients with resectable colorectal liver metastases (CRLM), the role of pre- and postoperative systemic therapy continues to be debated. Previous studies have shown that circulating tumor DNA (ctDNA) analysis, as a marker of minimal residual disease, is a powerful prognostic factor in patients with nonmetastatic colorectal cancer (CRC). Serial analysis of ctDNA in patients with resectable CRLM could inform the optimal use of perioperative chemotherapy. Here, we performed a validation study to confirm the prognostic impact of postoperative ctDNA in resectable CRLM observed in a previous discovery study. METHODS AND FINDINGS: We prospectively collected plasma samples from patients with resectable CRLM, including presurgical and postsurgical samples, serial samples during any pre- or postoperative chemotherapy, and serial samples in follow-up. Via targeted sequencing of 15 genes commonly mutated in CRC, we identified at least 1 somatic mutation in each patient's tumor. We then designed a personalized assay to assess 1 mutation in plasma samples using the Safe-SeqS assay. A total of 380 plasma samples from 54 patients recruited from July 2011 to Dec 2014 were included in our analysis. Twenty-three (43%) patients received neoadjuvant chemotherapy, and 42 patients (78%) received adjuvant chemotherapy after surgery. Median follow-up was 51 months (interquartile range, 31 to 60 months). At least 1 somatic mutation was identified in all patients' tumor tissue. ctDNA was detectable in 46/54 (85%) patients prior to any treatment and 12/49 (24%) patients after surgery. There was a median 40.93-fold (19.10 to 87.73, P < 0.001) decrease in ctDNA mutant allele fraction with neoadjuvant chemotherapy, but ctDNA clearance during neoadjuvant chemotherapy was not associated with a better recurrence-free survival (RFS). Patients with detectable postoperative ctDNA experienced a significantly lower RFS (HR 6.3; 95% CI 2.58 to 15.2; P < 0.001) and overall survival (HR 4.2; 95% CI 1.5 to 11.8; P < 0.001) compared to patients with undetectable ctDNA. For the 11 patients with detectable postoperative ctDNA who had serial ctDNA sampling during adjuvant chemotherapy, ctDNA clearance was observed in 3 patients, 2 of whom remained disease-free. All 8 patients with persistently detectable ctDNA after adjuvant chemotherapy have recurred. End-of-treatment (surgery +/- adjuvant chemotherapy) ctDNA detection was associated with a 5-year RFS of 0% compared to 75.6% for patients with an undetectable end-of-treatment ctDNA (HR 14.9; 95% CI 4.94 to 44.7; P < 0.001). Key limitations of the study include the small sample size and the potential for false-positive findings with multiple hypothesis testing. CONCLUSIONS: We confirmed the prognostic impact of postsurgery and posttreatment ctDNA in patients with resected CRLM. The potential utility of serial ctDNA analysis during adjuvant chemotherapy as an early marker of treatment efficacy was also demonstrated. Further studies are required to define how to optimally integrate ctDNA analyses into decision-making regarding the use and timing of adjuvant therapy for resectable CRLM. TRIAL REGISTRATION: ACTRN12612000345886.

Cancer-Associated Mutations in Endometriosis without Cancer.

BACKGROUND: Endometriosis, defined as the presence of ectopic endometrial stroma and epithelium, affects approximately 10% of reproductive-age women and can cause pelvic pain and infertility. Endometriotic lesions are considered to be benign inflammatory lesions but have cancerlike features such as local invasion and resistance to apoptosis. METHODS: We analyzed deeply infiltrating endometriotic lesions from 27 patients by means of exomewide sequencing (24 patients) or cancer-driver targeted sequencing (3 patients). Mutations were validated with the use of digital genomic methods in microdissected epithelium and stroma. Epithelial and stromal components of lesions from an additional 12 patients were analyzed by means of a droplet digital polymerase-chain-reaction (PCR) assay for recurrent activating KRAS mutations. RESULTS: Exome sequencing revealed somatic mutations in 19 of 24 patients (79%). Five patients harbored known cancer driver mutations in ARID1A, PIK3CA, KRAS, or PPP2R1A, which were validated by Safe-Sequencing System or immunohistochemical analysis. The likelihood of driver genes being affected at this rate in the absence of selection was estimated at P=0.001 (binomial test). Targeted sequencing and a droplet digital PCR assay identified KRAS mutations in 2 of 3 patients and 3 of 12 patients, respectively, with mutations in the epithelium but not the stroma. One patient harbored two different KRAS mutations, c.35G→T and c.35G→C, and another carried identical KRAS c.35G→A mutations in three distinct lesions. CONCLUSIONS: We found that lesions in deep infiltrating endometriosis, which are associated with virtually no risk of malignant transformation, harbor somatic cancer driver mutations. Ten of 39 deep infiltrating lesions (26%) carried driver mutations; all the tested somatic mutations appeared to be confined to the epithelial compartment of endometriotic lesions.

Genomic landscape and evolutionary trajectories of ovarian cancer precursor lesions.

The clonal relationship between ovarian high-grade serous carcinoma (HGSC) and its presumed precursor lesion, serous tubal intraepithelial carcinoma (STIC), has been reported. However, when analyzing patients with concurrent ovarian carcinoma and tubal lesion, the extensive carcinoma tissues present at diagnosis may have effaced the natural habitat of precursor clone(s), obscuring tumor clonal evolutionary history, or may have disseminated to anatomically adjacent fimbriae ends, masquerading as precursor lesions. To circumvent these limitations, we analyzed the genomic landscape of incidental tubal precursor lesions including p53 signature, dormant STIC or serous tubal intraepithelial lesion (STIL) and proliferative STIC in women without ovarian carcinoma or any cancer diagnosis using whole-exome sequencing and amplicon sequencing. In three of the four cancer-free women with multiple discrete tubal lesions we observed non-identical TP53 mutations between precursor lesions from the same individual. In one of the four women with co-existing ovarian HGSC and tubal precursor lesion we found non-identical TP53 mutations and a lack of common mutations shared between her precursor lesion and carcinoma. Analyzing the evolutionary history of multiple tubal lesions in the same four patients with concurrent ovarian carcinoma indicated distinct evolution trajectories. Collectively, the results support diverse clonal origins of tubal precursor lesions at the very early stages of tumorigenesis. Mathematical modeling based on lesion-specific proliferation rates indicated that p53 signature and dormant STIC may take a prolonged time (two decades or more) to develop into STIC, whereas STIC may progress to carcinoma in a much shorter time (6 years). The above findings may have implications for future research aimed at prevention and early detection of ovarian cancer. Copyright © 2018 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

Combined circulating tumor DNA and protein biomarker-based liquid biopsy for the earlier detection of pancreatic cancers.

The earlier diagnosis of cancer is one of the keys to reducing cancer deaths in the future. Here we describe our efforts to develop a noninvasive blood test for the detection of pancreatic ductal adenocarcinoma. We combined blood tests for KRAS gene mutations with carefully thresholded protein biomarkers to determine whether the combination of these markers was superior to any single marker. The cohort tested included 221 patients with resectable pancreatic ductal adenocarcinomas and 182 control patients without known cancer. KRAS mutations were detected in the plasma of 66 patients (30%), and every mutation found in the plasma was identical to that subsequently found in the patient's primary tumor (100% concordance). The use of KRAS in conjunction with four thresholded protein biomarkers increased the sensitivity to 64%. Only one of the 182 plasma samples from the control cohort was positive for any of the DNA or protein biomarkers (99.5% specificity). This combinatorial approach may prove useful for the earlier detection of many cancer types.

Serial circulating tumour DNA analysis during multimodality treatment of locally advanced rectal cancer: a prospective biomarker study.

OBJECTIVE: For patients with locally advanced rectal cancer (LARC), adjuvant chemotherapy selection following surgery remains a major clinical dilemma. Here, we investigated the ability of circulating tumour DNA (ctDNA) to improve risk stratification in patients with LARC. DESIGN: We enrolled patients with LARC (T3/T4 and/or N+) planned for neoadjuvant chemoradiotherapy. Plasma samples were collected pretreatment, postchemoradiotherapy and 4-10 weeks after surgery. Somatic mutations in individual patient's tumour were identified via massively parallel sequencing of 15 genes commonly mutated in colorectal cancer. We then designed personalised assays to quantify ctDNA in plasma samples. Patients received adjuvant therapy at clinician discretion, blinded to the ctDNA results. RESULTS: We analysed 462 serial plasma samples from 159 patients. ctDNA was detectable in 77%, 8.3% and 12% of pretreatment, postchemoradiotherapy and postsurgery plasma samples. Significantly worse recurrence-free survival was seen if ctDNA was detectable after chemoradiotherapy (HR 6.6; P<0.001) or after surgery (HR 13.0; P<0.001). The estimated 3-year recurrence-free survival was 33% for the postoperative ctDNA-positive patients and 87% for the postoperative ctDNA-negative patients. Postoperative ctDNA detection was predictive of recurrence irrespective of adjuvant chemotherapy use (chemotherapy: HR 10.0; P<0.001; without chemotherapy: HR 22.0; P<0.001). Postoperative ctDNA status remained an independent predictor of recurrence-free survival after adjusting for known clinicopathological risk factors (HR 6.0; P<0.001). CONCLUSION: Postoperative ctDNA analysis stratifies patients with LARC into subsets that are either at very high or at low risk of recurrence, independent of conventional clinicopathological risk factors. ctDNA analysis could potentially be used to guide patient selection for adjuvant chemotherapy.

The potential role of circulating tumor DNA (ctDNA) in the further investigation of colorectal cancer patients with nonspecific findings on standard investigations.

Early detection of metastatic colorectal cancer, at initial diagnosis or during routine surveillance, can improve survival outcomes. Current routine investigations, including CEA and CT, have limited sensitivity and specificity. Recent studies of colorectal cancer cohorts under post surgery surveillance indicate circulating tumor DNA (ctDNA) evidence of recurrence can occur many months before clinical detection. Another possible role for ctDNA is in the further assessment of indeterminate findings on standard CEA or CT investigations. To further explore this potential, we undertook a prospective study. Further investigation, including FDG-PET imaging, was at clinician discretion, blinded to ctDNA analysis. Forty-nine patients were enrolled. Analyzed here are the 45 patients with an evaluable blood sample of whom 6 had an isolated elevated CEA, 30 had indeterminate CT findings, and 9 had both. FDG-PET scans were performed in 30 patients. Fourteen of 45 patients (31%) had detectable ctDNA. At completion of the planned 2 year follow-up, recurrence has occurred in 21 (47%) patients. Detectable ctDNA at study entry was associated with inferior relapse free survival (HR 4.85, p < 0.0001). Where FDG-PET scan was normal/equivocal (n = 15, 50%) 1 of 1 with detectable ctDNA versus 3 of 14 with undetectable ctDNA ultimately had recurrence confirmed. In summary, for colorectal cancer patients with indeterminate findings on routine investigations, ctDNA detection increases the probability that the findings indicate metastatic disease, including in a nonpredefined subset that also underwent FDG-PET imaging. Further studies of the value of ctDNA analysis during patient surveillance are warranted.

Incidence and distribution of UroSEEK gene panel in a multi-institutional cohort of bladder urothelial carcinoma.

Noninvasive approaches for early detection of bladder cancer are actively being investigated. We recently developed a urine- based molecular assay for the detection and surveillance of bladder neoplasms (UroSEEK). UroSEEK is designed to detect alterations in 11 genes that include most common genetic alterations in bladder cancer. In this study, we analyzed 527 cases, including 373 noninvasive and 154 invasive urothelial carcinomas of bladder from transurethral resections or cystectomies performed at four institutions (1991-2016). Two different mutational analysis assays of a representative tumor area were performed: first, a singleplex PCR assay for evaluation of the TERT promoter region (TERTSeqS) and second, a multiplex PCR assay using primers designed to amplify regions of interest of 10 (FGFR3, PIK3CA, TP53, HRAS, KRAS, ERBB2, CDKN2A, MET, MLL, and VHL) genes (UroSeqS). Overall, 92% of all bladder tumors were positive for at least one genetic alteration in the UroSEEK panel. We found TERT promoter mutations in 77% of low-grade noninvasive papillary carcinomas, with a relatively lower incidence of 65% in high-grade noninvasive papillary carcinomas and carcinomas in situ; p = 0.017. Seventy-two percent of pT1 and 63% of muscle-invasive bladder tumors harbored TERT promoter mutations with g.1295228C>T alteration being the most common in all groups. FGFR3 and PIK3CA mutations were more frequent in low-grade noninvasive papillary carcinomas compared with high-grade noninvasive papillary carcinomas and carcinomas in situ (p < 0.0001), while the opposite was true for TP53 (p < 0.0001). Significantly higher rates of TP53 and CDKN2A mutation rates (p = 0.005 and 0.035, respectively) were encountered in muscle-invasive bladder tumors compared with those of pT1 stage. The overwhelming majority of all investigated tumors showed at least one mutation among UroSEEK assay genes, confirming the comprehensive coverage of the panel and supporting its potential utility as a noninvasive urine-based assay.

Genome-wide quantification of rare somatic mutations in normal human tissues using massively parallel sequencing.

We present the bottleneck sequencing system (BotSeqS), a next-generation sequencing method that simultaneously quantifies rare somatic point mutations across the mitochondrial and nuclear genomes. BotSeqS combines molecular barcoding with a simple dilution step immediately before library amplification. We use BotSeqS to show age- and tissue-dependent accumulations of rare mutations and demonstrate that somatic mutational burden in normal human tissues can vary by several orders of magnitude, depending on biologic and environmental factors. We further show major differences between the mutational patterns of the mitochondrial and nuclear genomes in normal tissues. Lastly, the mutation spectra of normal tissues were different from each other, but similar to those of the cancers that arose in them. This technology can provide insights into the number and nature of genetic alterations in normal tissues and can be used to address a variety of fundamental questions about the genomes of diseased tissues.

Only three driver gene mutations are required for the development of lung and colorectal cancers.

Cancer arises through the sequential accumulation of mutations in oncogenes and tumor suppressor genes. However, how many such mutations are required for a normal human cell to progress to an advanced cancer? The best estimates for this number have been provided by mathematical models based on the relation between age and incidence. For example, the classic studies of Nordling [Nordling CO (1953) Br J Cancer 7(1):68-72] and Armitage and Doll [Armitage P, Doll R (1954) Br J Cancer 8(1):1-12] suggest that six or seven sequential mutations are required. Here, we describe a different approach to derive this estimate that combines conventional epidemiologic studies with genome-wide sequencing data: incidence data for different groups of patients with the same cancer type were compared with respect to their somatic mutation rates. In two well-documented cancer types (lung and colon adenocarcinomas), we find that only three sequential mutations are required to develop cancer. This conclusion deepens our understanding of the process of carcinogenesis and has important implications for the design of future cancer genome-sequencing efforts.

Half or more of the somatic mutations in cancers of self-renewing tissues originate prior to tumor initiation.

Although it has been hypothesized that some of the somatic mutations found in tumors may occur before tumor initiation, there is little experimental or conceptual data on this topic. To gain insights into this fundamental issue, we formulated a mathematical model for the evolution of somatic mutations in which all relevant phases of a tissue's history are considered. The model makes the prediction, validated by our empirical findings, that the number of somatic mutations in tumors of self-renewing tissues is positively correlated with the age of the patient at diagnosis. Importantly, our analysis indicates that half or more of the somatic mutations in certain tumors of self-renewing tissues occur before the onset of neoplasia. The model also provides a unique way to estimate the in vivo tissue-specific somatic mutation rates in normal tissues directly from the sequencing data of tumors. Our results have substantial implications for the interpretation of the large number of genome-wide cancer studies now being undertaken.

An argument for mechanism-based statistical inference in cancer.

Cancer is perhaps the prototypical systems disease, and as such has been the focus of extensive study in quantitative systems biology. However, translating these programs into personalized clinical care remains elusive and incomplete. In this perspective, we argue that realizing this agenda­in particular, predicting disease phenotypes, progression and treatment response for individuals­requires going well beyond standard computational and bioinformatics tools and algorithms. It entails designing global mathematical models over network-scale configurations of genomic states and molecular concentrations, and learning the model parameters from limited available samples of high-dimensional and integrative omics data. As such, any plausible design should accommodate: biological mechanism, necessary for both feasible learning and interpretable decision making; stochasticity, to deal with uncertainty and observed variation at many scales; and a capacity for statistical inference at the patient level. This program, which requires a close, sustained collaboration between mathematicians and biologists, is illustrated in several contexts, including learning biomarkers, metabolism, cell signaling, network inference and tumorigenesis.

Drug resistance.

Drug resistance is a fundamental problem in the treatment of cancer since cancer that becomes resistant to the available drugs may leave the patient with no therapeutic alternatives. In this chapter, we consider the dynamics of drug resistance in blood cancer and the related issue of the dynamics of cancer stem cells. After describing the main types of chemotherapeutic agents available for cancer treatment, we review the different mechanisms of drug resistance development. Various mathematical models of drug resistance found in the literature are then reviewed. Given the well-known hierarchy of the hematopoietic system, it is critical to focus on those cells that have the ability to self-renew, since these will be the only cells able to induce long-term drug resistance. Thus, a recent mathematical model taking into account the complex dynamics of the leukemic stem-like cells is described. The chapter closes with a few applications of this model to chronic myeloid leukemia.

Machine learning to detect the SINEs of cancer.

We previously described an approach called RealSeqS to evaluate aneuploidy in plasma cell-free DNA through the amplification of ~350,000 repeated elements with a single primer. We hypothesized that an unbiased evaluation of the large amount of sequencing data obtained with RealSeqS might reveal other differences between plasma samples from patients with and without cancer. This hypothesis was tested through the development of a machine learning approach called Alu Profile Learning Using Sequencing (A-PLUS) and its application to 7615 samples from 5178 individuals, 2073 with solid cancer and the remainder without cancer. Samples from patients with cancer and controls were prespecified into four cohorts used for model training, analyte integration, and threshold determination, validation, and reproducibility. A-PLUS alone provided a sensitivity of 40.5% across 11 different cancer types in the validation cohort, at a specificity of 98.5%. Combining A-PLUS with aneuploidy and eight common protein biomarkers detected 51% of the cancers at 98.9% specificity. We found that part of the power of A-PLUS could be ascribed to a single feature-the global reduction of AluS subfamily elements in the circulating DNA of patients with solid cancer. We confirmed this reduction through the analysis of another independent dataset obtained with a different approach (whole-genome sequencing). The evaluation of Alu elements may therefore have the potential to enhance the performance of several methods designed for the earlier detection of cancer.