Clinical validation of a blood-based liquid biopsy test integrating cell-free DNA quantification and next-generation sequencing for cancer screening in dogs

Andi Flory Medical and Clinical Affairs, PetDx, La Jolla, CA

Search for other papers by Andi Flory in
Current site
Google Scholar
PubMed
Close
 DVM, DACVIM
,
Carlos A. Ruiz-Perez Information Technology, PetDx, La Jolla, CA

Search for other papers by Carlos A. Ruiz-Perez in
Current site
Google Scholar
PubMed
Close
 PhD
,
Ana G. Clavere-Graciette Research Programs, PetDx, La Jolla, CA

Search for other papers by Ana G. Clavere-Graciette in
Current site
Google Scholar
PubMed
Close
 PhD
,
Jill M. Rafalko Medical and Clinical Affairs, PetDx, La Jolla, CA

Search for other papers by Jill M. Rafalko in
Current site
Google Scholar
PubMed
Close
 MS
,
Allison L. O’Kell Medical and Clinical Affairs, PetDx, La Jolla, CA

Search for other papers by Allison L. O’Kell in
Current site
Google Scholar
PubMed
Close
 DVM, MS, DACVIM
,
Brian K. Flesner Medical and Clinical Affairs, PetDx, La Jolla, CA

Search for other papers by Brian K. Flesner in
Current site
Google Scholar
PubMed
Close
 DVM, MSc, DACVIM
,
Lisa M. McLennan Research Programs, PetDx, La Jolla, CA

Search for other papers by Lisa M. McLennan in
Current site
Google Scholar
PubMed
Close
 SCRA
,
Susan C. Hicks Analytical Production, PetDx, La Jolla, CA

Search for other papers by Susan C. Hicks in
Current site
Google Scholar
PubMed
Close
 MAS
,
Prachi Nakashe Research Programs, PetDx, La Jolla, CA

Search for other papers by Prachi Nakashe in
Current site
Google Scholar
PubMed
Close
 MS
,
Ashley Phelps-Dunn Medical and Clinical Affairs, PetDx, La Jolla, CA

Search for other papers by Ashley Phelps-Dunn in
Current site
Google Scholar
PubMed
Close
 DVM
,
Lauren R. DiMarzio Medical and Clinical Affairs, PetDx, La Jolla, CA

Search for other papers by Lauren R. DiMarzio in
Current site
Google Scholar
PubMed
Close
 DVM
,
Chelsea D. Warren Medical and Clinical Affairs, PetDx, La Jolla, CA

Search for other papers by Chelsea D. Warren in
Current site
Google Scholar
PubMed
Close
 DVM
,
Todd A. Cohen Medical and Clinical Affairs, PetDx, La Jolla, CA

Search for other papers by Todd A. Cohen in
Current site
Google Scholar
PubMed
Close
 DVM, DACVIM
,
Jason Chibuk Medical and Clinical Affairs, PetDx, La Jolla, CA

Search for other papers by Jason Chibuk in
Current site
Google Scholar
PubMed
Close
 MS
,
Ilya Chorny Information Technology, PetDx, La Jolla, CA

Search for other papers by Ilya Chorny in
Current site
Google Scholar
PubMed
Close
 PhD
,
Daniel S. Grosu Medical and Clinical Affairs, PetDx, La Jolla, CA

Search for other papers by Daniel S. Grosu in
Current site
Google Scholar
PubMed
Close
 MD, MBA
,
Dana W. Y. Tsui Research Programs, PetDx, La Jolla, CA

Search for other papers by Dana W. Y. Tsui in
Current site
Google Scholar
PubMed
Close
 PhD
,
John A. Tynan Research Programs, PetDx, La Jolla, CA

Search for other papers by John A. Tynan in
Current site
Google Scholar
PubMed
Close
 PhD
, and
Kristina M. Kruglyak Information Technology, PetDx, La Jolla, CA

Search for other papers by Kristina M. Kruglyak in
Current site
Google Scholar
PubMed
Close
 PhD

Click on author name to view affiliation information

Abstract

OBJECTIVE

To validate the performance of a novel, integrated test for canine cancer screening that combines cell-free DNA quantification with next-generation sequencing (NGS) analysis.

SAMPLE

Retrospective data from a total of 1,947 cancer-diagnosed and presumably cancer-free dogs were used to validate test performance for the detection of 7 predefined cancer types (lymphoma, hemangiosarcoma, osteosarcoma, leukemia, histiocytic sarcoma, primary lung tumors, and urothelial carcinoma), using independent training and testing sets.

METHODS

Cell-free DNA quantification data from all samples were analyzed using a proprietary machine learning algorithm to determine a Cancer Probability Index (High, Moderate, or Low). High and Low Probability of Cancer were final result classifications. Moderate cases were additionally analyzed by NGS to arrive at a final classification of High Probability of Cancer (Cancer Signal Detected) or Low Probability of Cancer (Cancer Signal Not Detected).

RESULTS

Of the 595 dogs in the testing set, 89% (n = 530) received a High or Low Probability result based on the machine learning algorithm; 11% (65) were Moderate Probability, and NGS results were used to assign a final classification. Overall, 87 of 122 dogs with the 7 predefined cancer types were classified as High Probability and 467 of 473 presumably cancer-free dogs were classified as Low Probability, corresponding to a sensitivity of 71.3% for the predefined cancer types at a specificity of 98.7%.

CLINICAL RELEVANCE

This integrated test offers a novel option to screen for cancer types that may be difficult to detect by physical examination at a dog’s wellness visit.

Abstract

OBJECTIVE

To validate the performance of a novel, integrated test for canine cancer screening that combines cell-free DNA quantification with next-generation sequencing (NGS) analysis.

SAMPLE

Retrospective data from a total of 1,947 cancer-diagnosed and presumably cancer-free dogs were used to validate test performance for the detection of 7 predefined cancer types (lymphoma, hemangiosarcoma, osteosarcoma, leukemia, histiocytic sarcoma, primary lung tumors, and urothelial carcinoma), using independent training and testing sets.

METHODS

Cell-free DNA quantification data from all samples were analyzed using a proprietary machine learning algorithm to determine a Cancer Probability Index (High, Moderate, or Low). High and Low Probability of Cancer were final result classifications. Moderate cases were additionally analyzed by NGS to arrive at a final classification of High Probability of Cancer (Cancer Signal Detected) or Low Probability of Cancer (Cancer Signal Not Detected).

RESULTS

Of the 595 dogs in the testing set, 89% (n = 530) received a High or Low Probability result based on the machine learning algorithm; 11% (65) were Moderate Probability, and NGS results were used to assign a final classification. Overall, 87 of 122 dogs with the 7 predefined cancer types were classified as High Probability and 467 of 473 presumably cancer-free dogs were classified as Low Probability, corresponding to a sensitivity of 71.3% for the predefined cancer types at a specificity of 98.7%.

CLINICAL RELEVANCE

This integrated test offers a novel option to screen for cancer types that may be difficult to detect by physical examination at a dog’s wellness visit.

Introduction

Cancer screening involves actions taken to look for cancer before signs or symptoms appear,1 and guidelines-driven routine screening tests are an important part of preventive care in human medicine for multiple cancer types in which early detection has been shown to reduce mortality.2 In veterinary medicine, professional organizations recognize the value of early cancer detection for optimal patient outcomes,3,4 but no formal screening guidelines currently exist and cancer in dogs is often diagnosed at advanced stages. In the absence of guidelines, the paradigm for early cancer detection in dogs is the annual or biannual wellness visit; however, 1 study5 found that only 12% of dogs with cancer had their disease detected prior to the development of clinical signs.

A thorough physical examination, including an oral and rectal examination, is an integral part of each wellness visit. Veterinarians receive extensive training in how to perform these examinations, which are routine tasks for the general practitioner; however, no matter how thorough the examination, cancer can be present in anatomic locations that may be difficult (eg, due to body habitus or patient temperament) or impossible (eg, intrathoracic) to assess without additional tools such as imaging. For anatomic locations where gross disease may evade evaluation or detection by physical examination, an alternative means of cancer screening has the potential to offer significant utility to clinicians and pet owners. Such a screening tool should not only be able to identify patients with cancers that may be difficult to detect but should also be at a cost that allows for broad and equitable access.

The authors of the current study previously developed and validated a novel, blood-based liquid biopsy test for canine multi-cancer early detection (MCED) using next-generation sequencing (NGS) of cell-free DNA (cfDNA).6 In short, cfDNA is the term to describe fragmented DNA that is released into circulation when cells die by apoptosis or necrosis. Both healthy cells and cancer cells (if present in the body) release cfDNA. Given that cancer is fundamentally a disease of the genome, the subset of cfDNA from cancer cells typically contains genomic alterations and is referred to as circulating tumor DNA or “ctDNA.”7 These DNA alterations are amenable to identification by NGS, an advanced genomic sequencing technology. At present, the tissue or organ of origin is not readily identifiable from NGS data in the majority of canine cancer types.

The clinical validation study for the NGS-based liquid biopsy test included 1,100 dogs with and without cancer and showed that the test was able to identify genomic alterations (a “cancer signal”) in 30 types of cancer out of 42 types evaluated. Though NGS-based liquid biopsy has the ability to detect a wide range of cancers, the cost of performing DNA sequencing on each sample submitted for testing could present a barrier to utilization in some cases.

Liquid biopsy using NGS involves a complex series of laboratory steps including blood centrifugation, DNA extraction, quantification, library preparation, sequencing, and bioinformatics analysis. The steps following DNA quantification account for the majority of the cost and turnaround time of NGS-based testing (Figure 1). Previously published studies of both humans and dogs have shown that the concentration of cfDNA in plasma is significantly higher in patients with cancer compared to healthy individuals8,9 and may also be elevated in patients with certain medical conditions (eg, sepsis,10 trauma,10 surgery,11 immune-mediated hemolytic anemia,12 and gastric dilatation-volvulus13). Additionally, cfDNA quantification with analysis of cfDNA fragment size profiles has been shown to be a promising tool for cancer prognostication.14 Given that cfDNA quantification and fragment size profiling is an early prerequisite step in the NGS testing process, a method in which cfDNA quantification alone can be used to stratify the probability of cancer in a majority of eligible patients could be expected to increase accessibility to cancer screening in dogs by decreasing both cost and turnaround time.15,16

Figure 1
Figure 1

Simplified overview of the laboratory steps involved in next-generation sequencing (NGS; following sample arrival and accessioning at the laboratory) compared to the abbreviated workflow required for cell-free DNA (cfDNA) quantification.

Citation: Journal of the American Veterinary Medical Association 262, 5; 10.2460/javma.23.10.0564

The aim of the current study was to validate the performance of a novel integrated test for canine cancer screening using cfDNA quantification complemented with NGS analysis in select cases. The study specifically focuses on 7 predefined, clinically relevant cancer types that meet 1 or more of the following criteria: (1) may be difficult to detect by physical examination, (2) may be aggressive and call for prompt intervention, and (3) may be associated with improved outcomes when detected and treated early.1739 The predefined cancer types are as follows: lymphoma (intermediate to large cell), hemangiosarcoma, osteosarcoma, leukemia (acute lymphoid leukemia and chronic lymphoid leukemia), histiocytic sarcoma, primary lung tumors, and urothelial carcinoma (ie, transitional cell carcinoma). This integrated test represents a novel option for routine cancer screening in dogs, with shorter turnaround time and lower cost as compared to an approach that requires NGS testing of all patients.

Methods

Validation of a blood-based liquid biopsy test integrating cfDNA quantification and NGS analysis (herein referred to as “the integrated test”) was performed using retrospective data from 1,947 dogs (1,212 presumably cancer-free dogs and 735 cancer-diagnosed dogs) divided between a training set and a testing set (Figure 2). The study population comprised dogs with reportable results (ie, excluding test failures) enrolled in the CANcer Detection in Dogs (CANDiD) study6 that had clinical data available as of April 2023 (n = 1,504), as well as dogs with samples submitted for commercial NGS-based testing and for which information was available regarding cancer status as of April 2023 (443).40

Figure 2
Figure 2

Disposition of samples and results of the integrated test combining cfDNA quantification and NGS analysis for the 1,947 dogs in the study population. Samples were assigned a Cancer Probability Index (High, Moderate, or Low Probability of Cancer) based on a proprietary cfDNA machine learning algorithm, with High and Low Probability considered final result classifications. Samples classified as Moderate Probability were subjected to additional analysis in which results from NGS were used: patients with Cancer Signal Detected results from NGS were assigned a final result classification of High Probability of Cancer, and patients with Cancer Signal Not Detected results from NGS were assigned a final result classification of Low Probability of Cancer. Test sensitivity was calculated on the basis of the final result classification for all presumably cancer-free dogs and cancer-diagnosed dogs with 1 or more of the 7 predefined cancers in the testing set (High/Low Probability of Cancer, indicated by shaded boxes in this figure). The calculation of test sensitivity excluded dogs diagnosed with a cancer type other than the 7 predefined cancers.

Citation: Journal of the American Veterinary Medical Association 262, 5; 10.2460/javma.23.10.0564

Patients from the CANDiD study were enrolled under protocols that received IACUC or site-specific ethics approval, according to each site’s requirements, as previously described.6 Commercial patient data were retrospectively collected on the basis of medical record review, as previously described.40

Presumably cancer-free dogs had no suspicion of cancer based on wellness examination and/or veterinarian assessment as indicated on the case report forms for CANDiD patients, or during collection of clinical data for commercial patients. Cancer-diagnosed dogs were an all-comers cohort representing 49 diverse cancer types (Supplementary Table S1).

Cancer-diagnosed patients from the CANDiD study had complete staging, with diagnoses made by tissue sampling (where possible) or imaging (for cases in which tissue sampling was not possible due to anatomic location), as previously described.6 These patients represented the full spectrum of cancer stages (localized/regional to disseminated/metastatic). Cancer-diagnosed patients from commercial testing had definitive diagnoses (based on histopathology or cytology) or presumptive diagnoses (based on imaging, direct examination/visualization, or strong suspicion on cytology/histopathology) as determined by the managing veterinarian or by PetDx veterinarians after a review of medical records provided by the managing veterinarian, as previously described.40

Blood collection and analysis

For all dogs in the study, whole blood was collected from a peripheral vein (jugular, cephalic, or saphenous) using cfDNA-optimized blood collection tubes (cell-free DNA collection tube; F. Hoffmann-La Roche Ltd), with a minimum of 7 mL of blood in each tube. Blood samples were shipped to the PetDx laboratory at ambient temperature without any required processing or special sample handling (such as refrigeration, freezing, or centrifugation) at the collection site. Samples were collected without any restrictions related to the time of day or the time of the dog’s last feeding.

The following were considered exclusion criteria for enrollment in the CANDiD study and contraindications for commercial testing: pregnancy and/or physical trauma (including injury, surgery, or core needle biopsy for any clinical indication) in the 7 days prior to blood collection; fine-needle aspiration, routine blood collection, and cystocentesis were considered acceptable. The following known or suspected comorbidities were not considered exclusion criteria: benign skin and subcutaneous tumors (eg, lipomas, sebaceous adenomas, skin tags) and acute or chronic inflammatory, infectious, autoimmune, degenerative, or other conditions.

One or 2 tubes of blood were collected from each dog. Upon arrival at the laboratory, blood samples were subjected to a centrifugation protocol to obtain plasma, as previously described.6 For samples collected as part of the CANDiD study, plasma aliquots were stored at –80 °C until they were thawed for testing. Samples collected as part of commercial testing were analyzed in real time.

For all samples, cfDNA was extracted using a highly standardized, proprietary bead-based chemistry optimized to maximize cfDNA yield in canine subjects. Cell-free DNA was then quantified and profiled using TapeStation (Agilent Technologies Inc), and amplified DNA libraries were subjected to NGS on a NovaSeq 6000 (Illumina Inc) for detection of cancer-associated genomic alterations. Sequencing data were analyzed as previously described,6 and sequencing results were classified as Cancer Signal Detected or Cancer Signal Not Detected. All data and results were stored in cloud-based repositories until retrieved and compiled for analysis.

Retrospective data analysis for cfDNA quantification algorithm development

For the current study, all historical data were retrieved for analysis. The dataset was filtered to exclude samples with < 2 mL of plasma (n = 27), samples with Indeterminate6 NGS results (3), and samples from dogs enrolled with a presumptive diagnosis of cancer whose tumors were subsequently confirmed to be benign (11). Dogs were then randomly assigned to independent training and testing sets containing 60% (n = 1,144 dogs; 711 presumably cancer-free and 433 cancer-diagnosed) and 40% (n = 762; 473 presumably cancer-free and 289 cancer-diagnosed, including 122 with 1 or more of the 7 predefined cancer types) of subjects, respectively (Figure 2).

To enhance generalizability and mitigate overfitting, the model was optimized using all cancer-diagnosed dogs in the training set, without consideration of cancer type (ie, an “all-comers” cancer cohort). Sensitivity was then estimated using cancer-diagnosed dogs in the testing set who were diagnosed with 1 of the 7 predefined cancer types: lymphoma (intermediate to large cell), hemangiosarcoma, osteosarcoma, leukemia (acute lymphoid leukemia and chronic lymphoid leukemia), histiocytic sarcoma, primary lung tumors, and urothelial carcinoma.

Training set

TapeStation data were exported and analyzed using the BioanalyzeR package version 0.10.0 in R version 4.3.0 (The R Project for Statistical Computing). The cfDNA concentrations for fragments of various sizes (ie, regions of the trace) were calculated, as well as various ratios between such regions. These regions and ratios are hereafter referred to as features. The features were subsequently analyzed using a proprietary machine learning algorithm that stratified patients into cancer probability categories. Briefly, features were normalized in the training dataset to constrain all values between 0 and 1. A logistic regression model was selected for the predictive analysis. Before model fitting, a feature selection step was performed to reduce the number of features and avoid overfitting. A logistic model was then developed, with parameter optimization executed through grid search, complemented by cross-validation.

The probabilities determined by the model were used to assign results to 1 of 3 categories: Low Probability of Cancer, Moderate Probability of Cancer, or High Probability of Cancer, collectively referred to as the Cancer Probability Index. Low Probability of Cancer and High Probability of Cancer were considered final result classifications. For all dogs classified as Moderate Probability of Cancer, an additional analysis was performed in which the results of NGS-based testing were used to arrive at a final result classification of either High Probability of Cancer (for dogs with Cancer Signal Detected from NGS) or Low Probability of Cancer (for dogs with Cancer Signal Not Detected from NGS). In summary, an integrated testing approach was established that employed cfDNA quantification analysis for all samples, followed by NGS analysis for a subset of samples.

Testing set

The locked-down pipeline (consisting of the optimized machine learning algorithm for cfDNA quantification data, combined with NGS analysis when required) was subsequently applied to patients in the testing set to determine clinical performance characteristics of the integrated test. All data reviewers were blinded to the cancer status (including type of cancer) for all patients in the testing set until after data analysis was complete. The cancer statuses of these patients were then unblinded. Test performance was calculated on the basis of dogs in the testing set using each patient’s final result classification (High or Low Probability of Cancer) by the integrated test. Test sensitivity was defined as the number of dogs diagnosed with 1 (or more) of the 7 predefined cancers that were classified as High Probability of Cancer divided by the total number of dogs that had been diagnosed with 1 (or more) of the predefined cancers in the testing set. Test specificity was defined as the number of presumably cancer-free dogs that were classified as Low Probability of Cancer divided by the total number of presumably cancer-free dogs in the testing set.

Performance assessment

Overall performance and individual detection rates in the 7 predefined cancer types were compared between an exclusive NGS analysis (using the analysis pipeline in production as of September 2023) and the novel integrated approach (in which NGS is only used for a subset of samples following cfDNA quantification). Separately, detection rates by cancer type were determined using final result classifications from all dogs with each of the 7 predefined cancer types across both the training and testing sets.

Statistical analysis

Confidence intervals for sensitivity and specificity were computed using the Wilson method. Differences in detection rates between the integrated test and the exclusive NGS analysis were estimated using the McNemar test. P values were adjusted for multiple testing using the Benjamini-Hochberg procedure.

Results

Patients ranged in age from 1.0 to 17.3 years (median, 8.0 years); weights ranged from 1.8 to 106.6 kg (median, 27.1 kg); 52% of patients were male, and 48% were female. Purebred and mixed-breed dogs comprised the study population, with over 80 distinct breeds represented in the purebred population. As described above, the performance of the integrated approach was established from 595 dogs in the testing set comprising 473 presumably cancer-free dogs and 122 dogs with 1 or more of the 7 predefined cancer types (from a total of 289 all-comer cancer-diagnosed dogs in the testing set; Figure 2).

Cell-free DNA quantification analysis

In agreement with observations in humans,8 higher cfDNA concentrations were observed in cancer-diagnosed dogs (n = 722) compared to presumably cancer-free dogs (1,184; Mann-Whitney rank sum test, P value < .001; Figure 3). Furthermore, fragment size analysis of canine plasma cfDNA revealed a pattern of multiple nucleosomal peaks, similar to what has been previously observed in human cfDNA studies.41,42

Figure 3
Figure 3

A—Distribution of cfDNA concentrations in presumably cancer-free and cancer-diagnosed dogs across both the training and testing sets. B—Representative fragment size profiles of cfDNA in a presumably cancer-free patient and a cancer-diagnosed patient (TapeStation data).

Citation: Journal of the American Veterinary Medical Association 262, 5; 10.2460/javma.23.10.0564

Taking into account features from cfDNA quantification and fragment size analysis, the machine learning algorithm described above provided a final result for 89% (530/595) of patients in the testing set: 79 dogs were classified as High Probability of Cancer and 451 as Low Probability of Cancer. The remaining 11% (65/595) of patients were classified as Moderate Probability of Cancer; for these patients, results of NGS-based testing were subsequently integrated to arrive at a final result of High Probability of Cancer or Low Probability of Cancer (Figure 2).

Next-generation sequencing analysis

Among the 65 patients classified as Moderate Probability of Cancer following cfDNA quantification analysis, 14 patients had Cancer Signal Detected results from NGS (resulting in a final classification of High Probability of Cancer) and 51 patients had Cancer Signal Not Detected results from NGS (resulting in a final classification of Low Probability of Cancer; Figure 2).

Analysis of overall test performance using the integrated test

The final result classifications from cfDNA quantification and NGS analyses were combined to determine the overall sensitivity and specificity of the integrated test. Overall, the integrated test classified 87 of the 122 dogs with 1 or more of the 7 predefined cancer types in the testing set as High Probability of Cancer (74 from cfDNA quantification analysis plus 13 from NGS analysis), corresponding to a sensitivity of 71.3% (95% CI, 62.7% to 78.6%). The test classified 467 of the 473 presumably cancer-free dogs in the testing set as Low Probability of Cancer (423 from cfDNA quantification analysis plus 44 from NGS analysis), corresponding to a specificity of 98.7% (95% CI, 97.3% to 99.4%; Figure 2).

Detection rates by cancer type

The integrated test detection rates by cancer type (calculated using samples across both the training and testing sets) were as follows: lymphoma, 78.7% (122/155); hemangiosarcoma, 74.3% (26/35); osteosarcoma, 58.7% (37/63); leukemia, 77.8% (7/9); primary lung tumors, 44.4% (8/18); histiocytic sarcoma, 43.8% (7/16); and urothelial carcinoma, 20% (4/20). As NGS results were also available for all patients in the study, the detection rates of exclusive NGS analysis for these cancer types are provided for comparison (Table 1).

Table 1

Cancer type–specific detection rates of the integrated test compared to exclusive next-generation sequencing (NGS) analysis (calculated using samples across both the training and testing sets).

Cancer type Detection rate of the integrated test Detection rate of exclusive NGS analysis Adjusted P value
Lymphoma (intermediate to large cell) 78.7% (122/155) 84.5% (131/155) .31
Leukemia (ALL/CLL) 77.8% (7/9) 77.8% (7/9) 1.0
Hemangiosarcoma 74.3% (26/35) 80.0% (28/35) .80
Osteosarcoma 58.7% (37/63) 73% (46/63) .02*
Primary lung tumors 44.4% (8/18) 55.6% (10/18) .75
Histiocytic sarcoma 43.8% (7/16) 62.5% (10/16) .45
Urothelial carcinoma (transitional cell carcinoma) 20.0% (4/20) 20.0% (4/20) 1.0

Two patients had multiple concurrent primary cancers from the 7 predefined cancer types (1 dog with hemangiosarcoma plus urothelial carcinoma; 1 dog with lymphoma plus urothelial carcinoma) and were represented in each of the cancer type–specific detection rate calculations above.

ALL = Acute lymphoid leukemia. CLL = Chronic lymphoid leukemia.

*Significant P value at P < .05.

As noted above, the overall sensitivity of the integrated test for the detection of the 7 predefined cancer types was 71.3% in the testing set, with a specificity of 98.7%. In the same cohort of patients, the overall sensitivity by exclusive NGS analysis (using the analysis pipeline in production as of September 2023) was 78.7% (96/122; 95% CI, 70.6% to 85.0%; P = .15), with a specificity of 98.7% (467/473; 95% CI, 97.3% to 99.4%). Similarly, in the all-comers cohort of cancer-diagnosed patients in the testing set, the sensitivity by exclusive NGS analysis was 52.6% (which falls within the CI established in the CANDiD study6). This was significantly higher (P < .01) than the sensitivity of 44.6% by the integrated test in the all-comers cohort (Table 2).

Table 2

Overall sensitivity of the integrated test compared to overall sensitivity of exclusive NGS analysis for the 7 predefined cancer types and for all cancer-diagnosed dogs (calculated from samples in the testing set).

Cancer type Sensitivity of the integrated test Sensitivity by exclusive NGS analysis Adjusted P value
7 predefined cancer types 71.3% (87/122; 95% CI, 62.7% to 78.6%) 78.7% (96/122; 95% CI, 70.6% to 85.0%) .15
All-comers including the 7 predefined cancer types 44.6% (129/289; 95% CI, 39.0% to 50.4%) 52.6% (152/289; 95% CI, 46.9% to 58.3%) < .01*

*Significant P value at P < .05.

Discussion

Blood-based testing by cfDNA quantification analysis (with incorporation of NGS analysis in select cases) can be used to screen for 7 clinically relevant canine cancer types with an overall sensitivity of 71.3% and specificity of 98.7%. As some of these cancer types may occur in anatomic locations that are difficult to evaluate, this blood-based test can act as a complement to the physical examination to increase the number and types of cancer cases detectable at a dog’s wellness visit.

Prior evidence suggests that each of the seven predefined cancer types analyzed in this study have subtypes that may be associated with improved outcomes when detected and treated at an early stage (localized vs metastatic spread) and/or at an early substage (preclinical vs clinical presentation) in dogs. For example, lymphoma affecting intra-abdominal or intrathoracic sites may be difficult to detect by physical examination. Patients whose lymphoma is detected at an early stage19 or substage (ie, clinically healthy, categorized as substage “a”),17,18 have been shown to have improved outcomes (remission rates/durations and survival times) as compared to dogs with advanced stage or substage (ie, substage “b”) lymphoma. Another example is hemangiosarcoma, which commonly occurs at internal sites such as the spleen, an organ that may be difficult to assess on physical examination unless the tumor is large or is causing significant blood loss. When hemangiosarcoma is detected at an early disease stage (before it has spread) or early clinical stage (eg, prior to rupture and the resulting abdominal hemorrhage), improved outcomes have been reported.2226 There are also published studies that support the importance of early detection for the other cancers evaluated: leukemia,20,21,38,39 osteosarcoma,2729 primary lung tumors,3033 histiocytic sarcoma,34,35 and urothelial carcinoma.36,37 It should be noted that the levels of evidence supporting these claims are variable, and not every subtype of cancer that falls under these 7 general types has published data demonstrating that early detection and treatment result in improved outcomes.

As noted above, cancer histologies not included in the 7 predefined types in the current study may also result in a High Probability of Cancer from the integrated test, allowing for detection of additional cancer cases in clinical practice. Across all cancer types, the overall sensitivity was 44.6% for the integrated test and 52.6% for NGS-based liquid biopsy; the latter is similar to the 51.5% overall sensitivity—across > 50 cancer types—demonstrated in the clinical validation study for the NGS-based multi-cancer early detection test available for commercial use in humans.43 Of note, the integrated test does not identify the tissue or organ of origin, and detection rates are variable by cancer type.

Cancer screening by integrating cfDNA quantification with NGS analysis has other benefits as well. Blood-based cancer screening is “cancer-agnostic,” meaning that it is not aimed at detecting a single type of cancer; rather, it has the ability to detect cancer that occurs throughout the body, including cancers that may be difficult to detect on physical examination alone due to their anatomic location. This is especially important in the screening setting, where (by definition) there is no current suspicion of cancer based on clinical presentation.44

A prior study45 performed by the authors of the current study established recommended ages to initiate blood-based cancer screening based on breed or weight, with a general recommendation to start screening at age 7 for all dogs and as early as age 4 for dogs of certain breeds or sizes. Additional longitudinal and/or population-based studies would be helpful in confirming or further refining these recommendations.

The integrated test is aimed at making cancer screening more accessible to canine patients. No fasting is required for the test, and results from cfDNA quantification analysis are reported more quickly, with a typical turnaround time of 2 to 4 calendar days (as compared to the current turnaround time of 7 to 10 calendar days for an approach that uses NGS analysis for all samples). As nearly 90% of patients in this study received a final result classification from cfDNA quantification analysis alone, results are returned quickly for the vast majority of patients. Additionally, cfDNA quantification-based analysis (with integration of NGS results only in select cases) is a lower-cost approach to screening for cancer compared to an approach that requires NGS analysis for all samples. These factors are expected to make the integrated test more accessible to canine patients by making it easier and more feasible for clinics to incorporate blood-based cancer screening into their wellness or preventive care protocols.

Cancer screening by cfDNA concentration does face limitations. Prior studies of humans and dogs have shown that the concentration of cfDNA in blood can be impacted by conditions other than cancer; for instance, transient increases in cfDNA concentration may be seen in humans and/or dogs with trauma, sepsis, organ failure, autoimmune disease, systemic infection/inflammation, or pregnancy.1012,4649 Therefore, concentration-based testing should be avoided in dogs that have any of these conditions at the time of blood draw or within the previous 7 days; this extended time frame is based on the known clearance of cfDNA in plasma (ie, typically within 48 hours50), with additional allowance made for biologic and patient variability. In these cases, it is recommended to wait until the condition is well controlled or resolved before using the integrated test. If noninvasive cancer testing is desired while these conditions are present, NGS-based liquid biopsy (with superior performance and proven ability to detect a larger number of cancer types) is recommended.

It should be noted that the presumably cancer-free patients included in the training and testing sets of this study represented an all-comers cohort, meaning that dogs with a variety of acute and chronic comorbidities commonly encountered in routine practice were not excluded from analysis; this allowed the false-positive rate of the test to account for a variety of scenarios that may be encountered in a real-world cancer screening population of older dogs. It should also be noted that not all skin/subcutaneous tumors present in dogs in the study population were sampled, so the true incidence of cancer and specific tumor types may be underestimated in the population.

Detection rates by cancer type are variable for both tests (eg, ranging from 78.7% for intermediate to large cell lymphoma to 20% for urothelial carcinoma for the integrated test). This variability is likely due to a number of factors, including differential shedding of cfDNA from the tumor (referred to in this context as circulating tumor DNA or ctDNA) into circulation based on cancer type, shedding of ctDNA into blood versus other biological fluids, and tumor volume/extent of disease.7

Neither integrated testing nor NGS-based testing detects all cancers, and not all cancers are detectable from a blood sample. A dog with cancer may have an elevated cfDNA concentration but may not have genomic alterations detected in plasma. Alternatively, a dog with cancer may have cancer-associated genomic alterations identified by NGS despite a normal cfDNA concentration. Therefore, blood-based cancer screening tools are not meant to be used in isolation or to replace existing diagnostics; they are meant to be used in conjunction with clinical expertise as a supplement to the current diagnostic toolbox, taking into consideration each patient’s unique circumstances and presentation. Also, blood-based cancer screening results should never be used as the sole basis for making important decisions such as treatment or euthanasia.

Prospective lifetime screening studies are needed to better understand the duration of the preclinical phase of cancer in dogs, the performance of blood-based tests (eg, NGS-based liquid biopsy and integrated testing methods) for detecting preclinical cancer in canine patients, the incremental benefits of cumulative sensitivity with interval screening, and the ultimate clinical utility of earlier detection (including impact on clinical management decisions and actual survival benefit after accounting for lead-time bias) for various cancer types.6

Lastly, the integrated test described in this study provides results indicating the likelihood that cancer is present at the time the patient’s blood was drawn. Neither cfDNA quantification analysis nor NGS analysis currently provides information about a patient’s genetic risk for developing cancer in the future.

Blood-based cancer screening using integrated cfDNA quantification and NGS analysis has the ability to identify dogs with a high probability of having 7 common, clinically relevant cancer types. The test is able to provide a clear “high” or “low” probability result from cfDNA quantification analysis alone for almost 90% of patients, with shorter turnaround time and at lower cost when compared to NGS-based testing in all patients. In the small percentage of cases for which a clear answer is not attained from cfDNA quantification analysis, integration of NGS analysis for detection of cancer-associated genomic alterations in cfDNA can provide a final, actionable result. When used in conjunction with a thorough physical examination, this noninvasive test has the potential to increase cancer detection at a dog’s wellness visit. The availability of blood-based cancer screening tools that can be readily incorporated into preventive care protocols opens the door to the development of cancer screening guidelines for early cancer detection in dogs.

Supplementary Materials

Supplementary materials are posted online at the journal website: avmajournals.avma.org.

Acknowledgments

The authors thank all of the dogs involved in this study and the humans who love and care for them. The authors also thank the doctors and staff of the clinics that collected the samples and provided the clinical outcome data that made this study possible. The authors thank the PetDx Clinical Studies and Laboratory teams for their assistance with data generation for this study, and Ms. Lauren Wenstad for her assistance in developing the tables and figures in this manuscript.

Disclosures

The authors of this study are all current or former employees of PetDx and hold vested or unvested equity in PetDx. Additionally, the authors have filed patents related to the technology described in this manuscript. This does not alter the authors’ adherence to JAVMA policies. The authors declare no additional conflicts of interest.

No AI-assisted technologies were used in the generation of this manuscript.

Funding

This study received funding from PetDx. The funder had the following involvement: study design, data collection and analysis, decision to publish, and preparation of the manuscript.

References

  • 1.

    American Cancer Society guidelines for the early detection of cancer. American Cancer Society. Accessed December 1, 2023. https://www.cancer.org/healthy/find-cancer-early/american-cancer-society-guidelines-for-the-early-detection-of-cancer.html

    • Search Google Scholar
    • Export Citation
  • 2.

    Sabatino SA, Thompson TD, White MC, et al. Cancer screening test receipt - United States, 2018. MMWR Morb Mortal Wkly Rep. 2021;70(2):29-35. doi:10.15585/mmwr.mm7002a1

    • Search Google Scholar
    • Export Citation
  • 3.

    Cancer in pets. AVMA. Accessed April 24, 2023. https://www.avma.org/resources/pet-owners/petcare/cancer-pets

  • 4.

    Is my dog at risk for cancer? American Animal Hospital Association. Accessed April 24, 2023. https://www.aaha.org/your-pet/pet-owner-education/ask-aaha/canine-cancer/

    • Search Google Scholar
    • Export Citation
  • 5.

    Flory A, McLennan L, Peet B, et al. Cancer detection in clinical practice and using blood-based liquid biopsy: a retrospective audit of over 350 dogs. J Vet Intern Med. 2023;37(1):258-267. doi:10.1111/jvim.16616

    • Search Google Scholar
    • Export Citation
  • 6.

    Flory A, Kruglyak KM, Tynan JA, et al. Clinical validation of a next-generation sequencing-based multi-cancer early detection “liquid biopsy” blood test in over 1,000 dogs using an independent testing set: the CANcer Detection in Dogs (CANDiD) study. PLoS One. 2022;17(4):e0266623. doi:10.1371/journal.pone.0266623

    • Search Google Scholar
    • Export Citation
  • 7.

    Chibuk J, Flory A, Kruglyak KM, et al. Horizons in veterinary precision oncology: fundamentals of cancer genomics and applications of liquid biopsy for the detection, characterization, and management of cancer in dogs. Front Vet Sci. 2021;8:664718. doi:10.3389/fvets.2021.664718

    • Search Google Scholar
    • Export Citation
  • 8.

    Alborelli I, Generali D, Jermann P, et al. Cell-free DNA analysis in healthy individuals by next-generation sequencing: a proof of concept and technical validation study. Cell Death Dis. 2019;10(7):534. doi:10.1038/s41419-019-1770-3

    • Search Google Scholar
    • Export Citation
  • 9.

    Kim J, Bae H, Ahn S, et al. Cell-Free DNA as a Diagnostic and Prognostic Biomarker in Dogs With Tumors. Front Vet Sci. 2021;8:735682. doi:10.3389/fvets.2021.735682

    • Search Google Scholar
    • Export Citation
  • 10.

    Letendre JA, Goggs R. Measurement of plasma cell-free DNA concentrations in dogs with sepsis, trauma, and neoplasia. J Vet Emerg Crit Care (San Antonio). 2017;27(3):307-314. doi:10.1111/vec.12592

    • Search Google Scholar
    • Export Citation
  • 11.

    Wilson IJ, Burchell RK, Worth AJ, et al. Kinetics of plasma cell-free DNA and creatine kinase in a canine model of tissue injury. J Vet Intern Med. 2018;32(1):157-164. doi:10.1111/jvim.14901

    • Search Google Scholar
    • Export Citation
  • 12.

    Jeffery U, Ruterbories L, Hanel R, LeVine DN. Cell-Free DNA and DNase activity in dogs with immune-mediated hemolytic anemia. J Vet Intern Med. 2017;31(5):1441-1450. doi:10.1111/jvim.14808

    • Search Google Scholar
    • Export Citation
  • 13.

    Troia R, Giunti M, Calipa S, Goggs R. Cell-free DNA, high-mobility group box-1, and procalcitonin concentrations in dogs with gastric dilatation-volvulus syndrome. Front Vet Sci. 2018;5:67. doi:10.3389/fvets.2018.00067

    • Search Google Scholar
    • Export Citation
  • 14.

    Lapin M, Oltedal S, Tjensvoll K, et al. Fragment size and level of cell-free DNA provide prognostic information in patients with advanced pancreatic cancer. J Transl Med. 2018;16(1):300. doi:10.1186/s12967-018-1677-2

    • Search Google Scholar
    • Export Citation
  • 15.

    Pathak AK, Bhutani M, Kumar S, Mohan A, Guleria R. Circulating cell-free DNA in plasma/serum of lung cancer patients as a potential screening and prognostic tool. Clin Chem. 2006;52(10):1833-1842.

    • Search Google Scholar
    • Export Citation
  • 16.

    Chen E, Cario CL, Leong L, et al. Cell-free DNA concentration and fragment size as a biomarker for prostate cancer. Sci Rep. 2021;11(1):5040. doi:10.1038/s41598-021-84507-z

    • Search Google Scholar
    • Export Citation
  • 17.

    Jagielski D, Lechowski R, Hoffmann-Jagielska M, Winiarczyk S. A retrospective study of the incidence and prognostic factors of multicentric lymphoma in dogs (1998-2000). J Vet Med A Physiol Pathol Clin Med. 2002;49(8):419-424. doi:10.1046/j.1439-0442.2002.00458.x

    • Search Google Scholar
    • Export Citation
  • 18.

    Škor O, Bicanová L, Wolfesberger B, et al. Are B-symptoms more reliable prognostic indicators than substage in canine nodal diffuse large B-cell lymphoma. Vet Comp Oncol. 2021;19(1):201-208.

    • Search Google Scholar
    • Export Citation
  • 19.

    Valli VE, Kass PH, San Myint M, Scott F. Canine lymphomas: association of classification type, disease stage, tumor subtype, mitotic rate, and treatment with survival. Vet Pathol. 2013;50(5):738-748. doi:10.1177/0300985813478210

    • Search Google Scholar
    • Export Citation
  • 20.

    Rout ED, Labadie JD, Yoshimoto JA, Avery PR, Curran KM, Avery AC. Clinical outcome and prognostic factors in dogs with B-cell chronic lymphocytic leukemia: A retrospective study. J Vet Intern Med. 2021;35(4):1918-1928. doi:10.1111/jvim.16160

    • Search Google Scholar
    • Export Citation
  • 21.

    Workman HC, Vernau W. Chronic lymphocytic leukemia in dogs and cats: the veterinary perspective. Vet Clin North Am Small Anim Pract. 2003;33(6):1379-1399, viii. doi:10.1016/S0195-5616(03)00120-7

    • Search Google Scholar
    • Export Citation
  • 22.

    Treggiari E, Borrego JF, Gramer I, et al. Retrospective comparison of first-line adjuvant anthracycline vs metronomic-based chemotherapy protocols in the treatment of stage I and II canine splenic haemangiosarcoma. Vet Comp Oncol. 2020;18(1):43-51. doi:10.1111/vco.12548

    • Search Google Scholar
    • Export Citation
  • 23.

    Masyr AR, Rendahl AK, Winter AL, Borgatti A, Modiano JF. Retrospective evaluation of thrombocytopenia and tumor stage as prognostic indicators in dogs with splenic hemangiosarcoma. J Am Vet Med Assoc. 2021;258(6):630-637. doi:10.2460/javma.258.6.630

    • Search Google Scholar
    • Export Citation
  • 24.

    Faroni E, Sabattini S, Guerra D, et al. Timely adjuvant chemotherapy improves outcome in dogs with non-metastatic splenic hemangiosarcoma undergoing splenectomy. Vet Comp Oncol. 2023;21(1):123-130. doi:10.1111/vco.12875

    • Search Google Scholar
    • Export Citation
  • 25.

    Wendelburg KM, Price LL, Burgess KE, Lyons JA, Lew FH, Berg J. Survival time of dogs with splenic hemangiosarcoma treated by splenectomy with or without adjuvant chemotherapy: 208 cases (2001-2012). J Am Vet Med Assoc. 2015;247(4):393-403. doi:10.2460/javma.247.4.393

    • Search Google Scholar
    • Export Citation
  • 26.

    Prymak C, McKee LJ, Goldschmidt MH, Glickman LT. Epidemiologic, clinical, pathologic, and prognostic characteristics of splenic hemangiosarcoma and splenic hematoma in dogs: 217 cases (1985). J Am Vet Med Assoc. 1988;193(6):706-712.

    • Search Google Scholar
    • Export Citation
  • 27.

    Spodnick GJ, Berg J, Rand WM, et al. Prognosis for dogs with appendicular osteosarcoma treated by amputation alone: 162 cases (1978-1988). J Am Vet Med Assoc. 1992;200(7):995-999. doi:10.2460/javma.1992.200.07.995

    • Search Google Scholar
    • Export Citation
  • 28.

    Hillers KR, Dernell WS, Lafferty MH, Withrow SJ, Lana SE. Incidence and prognostic importance of lymph node metastases in dogs with appendicular osteosarcoma: 228 cases (1986-2003). J Am Vet Med Assoc. 2005;226(8):1364-1367. doi:10.2460/javma.2005.226.1364

    • Search Google Scholar
    • Export Citation
  • 29.

    Nolan MW, Green NA, DiVito EM, Lascelles BDX, Haney SM. Impact of radiation dose and pre-treatment pain levels on survival in dogs undergoing radiotherapy with or without chemotherapy for presumed extremity osteosarcoma. Vet Comp Oncol. 2020;18(4):538-547. doi:10.1111/vco.12576

    • Search Google Scholar
    • Export Citation
  • 30.

    McNiel EA, Ogilvie GK, Powers BE, Hutchison JM, Salman MD, Withrow SJ. Evaluation of prognostic factors for dogs with primary lung tumors: 67 cases (1985-1992). J Am Vet Med Assoc. 1997;211(11):1422-1427. doi:10.2460/javma.1997.211.11.1422

    • Search Google Scholar
    • Export Citation
  • 31.

    Ichimata M, Kagawa Y, Namiki K, et al. Prognosis of primary pulmonary adenocarcinoma after surgical resection in small-breed dogs: 52 cases (2005-2021). J Vet Intern Med. 2023;37(4):1466-1474. doi:10.1111/jvim.16739

    • Search Google Scholar
    • Export Citation
  • 32.

    Polton GA, Brearley MJ, Powell SM, Burton CA. Impact of primary tumour stage on survival in dogs with solitary lung tumours. J Small Anim Pract. 2008;49(2):66-71. doi:10.1111/j.1748-5827.2007.00403.x

    • Search Google Scholar
    • Export Citation
  • 33.

    Lee BM, Clarke D, Watson M, Laver T. Retrospective evaluation of a modified human lung cancer stage classification in dogs with surgically excised primary pulmonary carcinomas. Vet Comp Oncol. 2020;18(4):590-598. doi:10.1111/vco.12582

    • Search Google Scholar
    • Export Citation
  • 34.

    Klahn SL, Kitchell BE, Dervisis NG. Evaluation and comparison of outcomes in dogs with periarticular and nonperiarticular histiocytic sarcoma. J Am Vet Med Assoc. 2011;239(1):90-96. doi:10.2460/javma.239.1.90

    • Search Google Scholar
    • Export Citation
  • 35.

    Marlowe KW, Robat CS, Clarke DM, et al. Primary pulmonary histiocytic sarcoma in dogs: a retrospective analysis of 37 cases (2000-2015). Vet Comp Oncol. 2018;16(4):658-663. doi:10.1111/vco.12437

    • Search Google Scholar
    • Export Citation
  • 36.

    Iwasaki R, Shimosato Y, Yoshikawa R, et al. Survival analysis in dogs with urinary transitional cell carcinoma that underwent whole-body computed tomography at diagnosis. Vet Comp Oncol. 2019;17(3):385-393. doi:10.1111/vco.12483

    • Search Google Scholar
    • Export Citation
  • 37.

    Dhawan D, Ramos-Vara JA, Utturkar SM, et al. Identification of a naturally-occurring canine model for early detection and intervention research in high grade urothelial carcinoma. Front Oncol. 2022;12:1011969. doi:10.3389/fonc.2022.1011969

    • Search Google Scholar
    • Export Citation
  • 38.

    Bennett AL, Williams LE, Ferguson MW, et al. Canine acute leukaemia: 50 cases (1989-2014). Vet Comp Oncol. 2017;15(3):1101-1114. doi:10.1111/vco.12251

    • Search Google Scholar
    • Export Citation
  • 39.

    Matus RE, Leifer CE, MacEwen EG. Acute lymphoblastic leukemia in the dog: a review of 30 cases. J Am Vet Med Assoc. 1983;183(8):859-862.

    • Search Google Scholar
    • Export Citation
  • 40.

    O’Kell AL, Lytle KM, Cohen TA, et al. Clinical experience with next-generation sequencing-based liquid biopsy testing for cancer detection in dogs: a review of 1,500 consecutive clinical cases. J Am Vet Med Assoc. 2023;261(6):827-836. doi:10.2460/javma.22.11.0526

    • Search Google Scholar
    • Export Citation
  • 41.

    Thierry AR. Circulating DNA fragmentomics and cancer screening. Cell Genomics. 2023;3(1):100242. doi:10.1016/j.xgen.2022.100242

  • 42.

    Lo YMD, Han DSC, Jiang P, Chiu RWK. Epigenetics, fragmentomics, and topology of cell-free DNA in liquid biopsies. Science. 2021;372(6538):eaaw3616. doi:10.1126/science.aaw3616

    • Search Google Scholar
    • Export Citation
  • 43.

    Klein EA, Richards D, Cohn A, et al. Clinical validation of a targeted methylation-based multi-cancer early detection test using an independent validation set. Ann Oncol. 2021;32(9):1167-1177. doi:10.1016/j.annonc.2021.05.806

    • Search Google Scholar
    • Export Citation
  • 44.

    Cancer screening overview. National Cancer Institute. Accessed October 11, 2023. https://www.cancer.gov/about-cancer/screening/patient-screening-overview-pdq

    • Search Google Scholar
    • Export Citation
  • 45.

    Rafalko JM, Kruglyak KM, McCleary-Wheeler AL, et al. Age at cancer diagnosis by breed, weight, sex, and cancer type in a cohort of more than 3,000 dogs: determining the optimal age to initiate cancer screening in canine patients. PLoS One. 2023;18(2):e0280795. doi:10.1371/journal.pone.0280795

    • Search Google Scholar
    • Export Citation
  • 46.

    Burnett DL, Cave NJ, Gedye KR, Bridges JP. Investigation of cell-free DNA in canine plasma and its relation to disease. Vet Q. 2016;36(3):122-129. doi:10.1080/01652176.2016.1182230

    • Search Google Scholar
    • Export Citation
  • 47.

    Gögenur M, Burcharth J, Gögenur I. The role of total cell-free DNA in predicting outcomes among trauma patients in the intensive care unit: a systematic review. Crit Care. 2017;21(1):14. doi:10.1186/s13054-016-1578-9

    • Search Google Scholar
    • Export Citation
  • 48.

    Lo YMD, Corbetta N, Chamberlain PF, et al. Presence of fetal DNA in maternal plasma and serum. Lancet. 1997;350(9076):485-487. doi:10.1016/S0140-6736(97)02174-0.

    • Search Google Scholar
    • Export Citation
  • 49.

    van der Meer AJ, Kroeze A, Hoogendijk AJ, et al. Systemic inflammation induces release of cell-free DNA from hematopoietic and parenchymal cells in mice and humans. Blood Adv. 2019;3(5):724-728. doi:10.1182/bloodadvances.2018018895

    • Search Google Scholar
    • Export Citation
  • 50.

    Yu SC, Lee SW, Jiang P, et al. High-resolution profiling of fetal DNA clearance from maternal plasma by massively parallel sequencing. Clin Chem. 2013;59(8):1228-1237. doi:10.1373/clinchem.2013.203679

    • Search Google Scholar
    • Export Citation

Supplementary Materials

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 12451 12451 11079
PDF Downloads 1543 1543 336
Advertisement