Liquid Biopsies Based on Cell-Free DNA Integrity as a Biomarker for Cancer Diagnosis: A Meta-Analysis

Liquid biopsies have been identified as a viable source of cancer biomarkers. We aim to evaluate the diagnostic accuracy of cell-free DNA integrity (cfDI) in liquid biopsies for cancer. A comprehensive literature search was conducted through PubMed, Embase, Web of Science, and Cochrane Library up to June 2024. Seventy-two study units from forty-six studies, comprising 4286 cancer patients, were identified and evaluated. The Quality Assessment for Studies of Diagnostic Accuracy-2 (QUADAS-2) was used to assess study quality. Meta-regression analysis was employed to investigate the underlying factors contributing to heterogeneity, alongside an evaluation of publication bias. The bivariate random-effect model was utilized to compute the primary diagnostic outcomes and their corresponding 95% confidence intervals (CIs). The pooled sensitivity, specificity, and positive and negative likelihood ratios of cfDI in cancer diagnosis were 0.70 and 0.77, 3.26 and 0.34, respectively. The overall area under the curve was 0.84, with a diagnostic odds ratio of 10.63. This meta-analysis suggested that the cfDI index has a promising potential as a non-invasive and accurate diagnostic tool for cancer. Study registration: The study was registered at PROSPERO (reference No. CRD42021276290).


Introduction
Cancer keeps its position as one of the predominant causes of death and decreases in life expectancy worldwide [1,2].According to data from the World Health Organization's GLOBOCAN network, an estimated 20 million new cases and almost 9.7 million cancer deaths occurred in 2022 [3], and these numbers are not expected to stop growing.
In the era of precision oncology [4,5], molecular biomarkers have been proposed as helpful tools, as they cover a broad range of biochemical entities [6,7], reporting relevant information to improve the diagnosis and decision-making.Although tissue biopsy remains the "gold standard" [8,9], liquid biopsies represent an important approach because they offer the opportunity to detect, analyse, and track cancer in real-time in different body fluids, thus representing a minimally invasive tool [10,11].Furthermore, they allow us to determine the molecular landscape, including the heterogeneity of the tumor, through different biomarkers during the clinical management of the disease [12,13].
A wide variety of biomarkers exist that can be studied on the different biofluids in the body, each one with different applications and approaches, reporting valuable information not only about the primary tumor and/or metastasis but also the microenvironment, the prognosis of the disease and the response to therapy.These biomarkers include: circulating tumor DNA (ctDNA), that forms a part of the broader category of cell-free DNA (cfDNA), which provides insights into genetic mutations and tumor burden; circulating tumor cells (CTCs), which can indicate the presence of metastasis; tumor-derived extracellular vesicles (tdEVs), which play a role in cell communication and can carry tumor-specific proteins and RNA; circulating tumor RNA (ctRNA), which can reveal gene expression profiles and potential therapeutic targets; and tumor-educated platelets (TEPs), which are involved in the interaction between the tumor and the host's blood coagulation system and can provide information on tumor growth and metastasis [14].
Among the aforementioned biomarkers, one of the most studied is cfDNA.CfDNA is a mixture of fragmented DNA molecules released from the cells of the body under different physiological and pathological conditions [15][16][17] into different body fluids such as blood, saliva, or urine [18][19][20].Cancer cells have high anti-apoptotic activity [21], and they release different, arbitrary-size, incompletely digested genomic DNA fragments, resulting from different processes apart from the aforesaid apoptosis, such as necrosis, autophagy, or mitotic catastrophe, into different body fluids [22,23].On the contrary, healthy cells mostly suffer apoptosis, with an average size of cfDNA ranging between 150-200 bp in plasma [24].This phenomenon is due to non-random nuclease activity in the inter-nucleosomal linker regions [25], which shows the well-known 'ladder' pattern when it is composed of several nucleosomal units [15,[26][27][28].The cell-free DNA integrity (cfDI) index is calculated based on the amplifications by quantitative real-time PCR (qPCR) of a genomic repetitive sequence targeted at the same locus [23,29].The fragmentation pattern of cfDNA can be characterized using a delta-delta formula based on a threshold cycle (Cp), e (−∆∆Cp×ln (2)) , or based on a ratio between longer fragments, which are more related to the process of necrosis, to shorter fragments, which are considered to represent the overall amount of cfDNA that stems from the totality of cells [29,30].
The aim of this meta-analysis was to systematically assess and synthesize the results of previously published clinical studies in this field and appraise the overall diagnostic accuracy of the cfDI index across various biofluids for discriminating cancer.We seek to provide a comprehensive assessment of the utility of cfDI as a minimally invasive biomarker for cancer detection, thereby contributing valuable insights that can potentially enhance the precision and effectiveness of cancer diagnostics through liquid biopsy approaches.

Protocol and Registration
The present meta-analysis was conducted in conformity with the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) guidelines [31], and the cor-responding protocol was registered on PROSPERO International Register of Systematic Reviews (reference No. CRD42021276290).

Search Strategy and Study Selection
We performed a systematic literature search of eligible articles published until 10 June 2024, which was carried out without any language restrictions or specified start dates in several electronic databases, including PubMed, EMBASE, Web of Science, and Cochrane Library.Additionally, free-text searches were performed.
The search strategy was grounded on the following combinations of free-style keywords and Medical Subjects Headings (MeSH): (cell-free DNA OR cfDNA) AND (cancer OR carcinoma OR tumor) AND (integrity index OR DNA integrity OR circulating DNA integrity).All studies were screened through perusal of the titles and abstracts, and manuscripts meeting the inclusion criteria were obtained for comprehensive text evaluation.In addition, we manually scrutinized the reference lists of all original and review articles to identify additional pertinent studies that may have been overlooked in the primary electronic database search.
The literature search for the studies was carried out independently by two investigators (AMRC and ORG), and any disparities were resolved through consensus.Disagreements in study identification during the selection process were reconciled through consensus in conjunction with a third investigator (MMSC).
The studies identified through the search strategy and additional sources were compiled and organized using the RefWorks software (version 2.1.0.1) (https://www.refworks.com/content/path_learn/faqs.asp,accessed on 14 June 2024).The associated tools were utilized to eliminate any duplicate items, and the remaining references were exported to an Excel sheet that contained basic information for screening.

Study Selection
Study inclusion criteria were: (1) liquid biopsies studies based on cfDI as a diagnostic biomarker; (2) studies that enrolled more than 15 patients; (3) inclusion of cancer patients and non-cancer controls (healthy and benign diseases); (4) sufficient data for generating a two-by-two (2 × 2) contingency table containing true-positive (TP), false-positive (FP), truenegative (TN), and false-negative (FN) data; (5) studies that only calculate the cfDI index using the direct cfDNA concentration ratio between a longer fragment to short fragments and/or the delta-delta formula: e (−∆∆Cp×ln (2)) .The exclusion criteria consisted of the following: (1) reviews, letters, personal opinions, book chapters, case reports, conference abstracts, and meetings; (2) duplicate publications; (3) those experimental in vitro and in vivo studies that reported correlations between liquid biopsies based on cfDI and cancer.
We defined "study unit" (or database) as the analysis of a relationship between cfDI and cancer diagnosis.As different lengths, repetitive DNA sequences chosen as an amplicon, and cfDI calculation methods might be reported in the same manuscript; a single publication may contain multiple study units.

Ethical Statement
All analyses were based on previously published studies and did not involve human beings or experimental subjects.As a result, ethical approval was not necessary.

Data Extraction and Quality Assessment
Data extraction from the eligible full-text articles was carried out independently by two authors (AMRC and ORG) using a structured Microsoft Excel spreadsheet (Microsoft Corp. Redmond, WA, USA).The following information was extracted from each study: author, publication year, journal, country, type of biofluid, type of cancer, number of cases and controls, type of targeted DNA sequence, including length of the amplicons, quantification method, type of cfDI calculation method, and statistical analysis outcomes.If the necessary data were inadequate or required verification, endeavours were made to communicate with the authors to obtain the missing information and minimize reporting bias.

Assessment of Risk Bias
As recommended by the Healthcare Research and Quality Agency, the Quality Assessment of Diagnostic Accuracy Studies-2 checklist (QUADAS-2) was applied by two independent researchers (AMRC and ORG) [32].Disagreements were resolved by a third reviewer (MMSC) or consensus-based discussion.QUADAS-2 is a tool designed to assess the quality of primary diagnostic accuracy studies, and it consists of 4 principal domains: (1) patient selection, (2) index test, (3) reference standard, and (4) flow of patients and timing of the index test.Each domain was rated for risk of bias and applicability concerns and classified as "low", "high", or "unclear".A score of one was assigned to each item if the response to the risk of bias and applicability questions was "low."Subsequently, the articles were categorized based on their score into high (6-7 points), moderate (4-5 points), and low (0-3 points) quality groups.

Statistical Analysis
Data analysis was executed using MetaDiSc (version 1.4), R (version 3.4.4),and Stata (version 14.0).For the diagnostic meta-analysis, we determined sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), and diagnostic odds ratio (DOR), each accompanied by their 95% confidence intervals (CIs).These calculations were performed using either a bivariate random effects model or a fixed effects model, depending on the context.To evaluate the effectiveness of cfDI in cancer detection, we generated the summary receiver operating characteristic (sROC) curve and computed the area under the curve (AUC).A thorough heterogeneity analysis was conducted to uncover factors impacting accuracy metrics and the choice of statistical model [33].We utilized Spearman's correlation and ROC plane plots to investigate heterogeneity caused by the threshold effect, whereas non-threshold heterogeneity was assessed through Cochran's Q test and I-squared (I2) statistics.Significant heterogeneity was identified by I2 values exceeding 50% and/or p-values under 0.05 for Cochran's Q test.In cases where significant heterogeneity was detected, the DerSimonian and Laird random effects model was applied; otherwise, the Mantel-Haenszel fixed effects model was used.Meta-regression and subgroup analyses were employed to pinpoint potential sources of non-threshold heterogeneity.Furthermore, Fagan's nomogram was utilized to gauge the predictive value of the cfDI index in diagnosing cancer.Publication bias was evaluated using Deeks' funnel plot asymmetry test, with statistical significance established at p < 0.05.

Study Selection
As shown in Figure 1, a PRISMA flowchart was used to depict the literature identification and selection process.A total of 1270 eligible published articles were yielded by the electronic database searching, and another three records were identified through a free search and reference checking.After removing duplicates, 776 records were left for the initial title and abstract screening, and 619 were excluded after a review of the text.Finally, 135 articles that are closely related to the cfDI index as a diagnostic tool for cancer were selected for full-text screening and further assessment of their eligibility.A review of the full-text articles led to the exclusion of 89 and the inclusion of 46 articles as part of our present meta-analysis [29,30,.

Characteristics of Included Studies
The main characteristics of the included studies are presented in Table 1.The selection process yielded a total of 46 studies, which, in turn, encompassed 72 study units.These papers focus on different types of cancer, from the most prevalent, such as breast, lung, and prostate cancer, to others with a lower incidence, such as ampullary carcinoma cancer, in which cfDI can be used as a diagnostic tool.In total, 4286 cases (sample sizes ranging from 16 to 148) and 3270 controls (ranging from 10 to 132) were included.All articles were published between 2003 and 2024 and were conducted in the following geographical regions: China ( 14), Egypt (12), Germany (7), EEUU (4), India (2), Italy (1), Poland (1), Korea (1), Australia (1), Canada (1), Spain (1), and Lithuania (1).In all cases, cfDI was detected by qPCR.Regarding the fluids used, plasma, serum, and other biofluids (pleural effusions, saliva, and urine) were the biofluids chosen for the cfDNA extraction.Regarding the methods used to calculate the cfDI index, a simple ratio of the concentrations of longer DNA fragments to shorter fragments of a targeted DNA sequence was used

Characteristics of Included Studies
The main characteristics of the included studies are presented in Table 1.The selection process yielded a total of 46 studies, which, in turn, encompassed 72 study units.These papers focus on different types of cancer, from the most prevalent, such as breast, lung, and prostate cancer, to others with a lower incidence, such as ampullary carcinoma cancer, in which cfDI can be used as a diagnostic tool.In total, 4286 cases (sample sizes ranging from 16 to 148) and 3270 controls (ranging from 10 to 132) were included.All articles were published between 2003 and 2024 and were conducted in the following geographical regions: China (14), Egypt (12), Germany (7), EEUU (4), India (2), Italy (1), Poland (1), Korea (1), Australia (1), Canada (1), Spain (1), and Lithuania (1).In all cases, cfDI was detected by qPCR.Regarding the fluids used, plasma, serum, and other biofluids (pleural effusions, saliva, and urine) were the biofluids chosen for the cfDNA extraction.Regarding the methods used to calculate the cfDI index, a simple ratio of the concentrations of longer DNA fragments to shorter fragments of a targeted DNA sequence was used in 60 study units.On the other side, the drawn-up formula based on Cp value differences described by Wang [29] was used in only 12 study units.Arthrobacter luteus (ALU) was the most used targeted repetitive sequence for calculating the ratio between lengths, chosen in 65.28% of the study units (47/72).β-actin (ACTB) was in second place (13/72), and the other proposed repetitive elements included LINE-1 (Long Interspersed Nuclear Elements) (5/72), GAPDH (Glyceraldehyde-3-Phosphate Dehydrogenase) (4/72), ERV (Endogenous Retrovirus) gene (2/52), and APP gene (Amyloid Precursor Protein) (1/72) that were chosen with less frequency.The length of the DNA sequences used for ratio calculation had a wide variability between studies, the most popular being the one that divides 247 to 115 base pairs of ALU repetitive sequences (35/72).

Quality Assessment of the Included Studies
The studies included in this meta-analysis showed an average QUADAS-2 score of 5.76.In terms of the risk of bias, numerous studies did not specify patient selection (random or consecutive) and did not provide a detailed description of the inclusion/exclusion criteria.Regarding the index test, most studies report threshold values for classifying the disease through the cfDI index, which could overestimate its predictive value.Additionally, regarding flow and timing, various studies were not clear about the appropriate intervals between the index test and reference test.The reference standard was specified clearly in the majority of studies.In terms of applicability, all domains showed a low risk of bias.An overview of the quality characteristics of included studies with QUADAS-2 template is shown in Supplementary Table S1, Supplemental Digital Content.

Diagnostic Accuracy of Cell-Free DNA Integrity Index
SE and 95% CIs for sensitivity and specificity estimates of a total of 72 study units were included in this meta-analysis.The forest plots (Figure 2) illustrate that the combined diagnostic sensitivity and specificity of the cfDI index were 0.70 (95% CI: 0.68-0.71)and 0.77 (95% CI: 0.76-0.79),respectively.This corresponds to a pooled positive likelihood ratio (PLR) of 3.26 (95% CI: 2.73-3.89)and a negative likelihood (NLR) of 0.34 (95% CI: 0.28-0.41)(Figure 3).The summary DOR was 10.63 (95% CI: 8.21-13.77),and the AUC was 0.84 (95% CI: 0.81-0.86)with a standard error (SE) of 0.014, indicating the overall diagnostic performance of the cfDI index (Figure 4).The Q-value for the sROC curve was 0.77 (95% CI: 0.74-0.79)with an SE of 0.01.(A) (B)  Fagan's nomogram plot was generated for visually presenting diagnostic performance (see Figure S2, Supplemental Digital Content); given a pre-test probability of 43%, a positive measurement leads to a post-test cancer probability of 71%, whereas a negative measurement leads to a post-test probability of 20%.
As shown in Figures 2-4, the I2 of sensitivity, specificity, PLR, NLR, and DOR were 91.1% (p < 0.0001), 86.1% (p < 0.0001), 86.5% (p < 0.0001), 94.6% (p < 0.0001), and 74.1% (p < 0.0001), respectively.These results of the diagnostic meta-analysis demonstrated that a substantial heterogeneity exists between the included studies.Furthermore, a typical pattern of the "shoulder arm" plot shown in the ROC space, which contains the representation of accuracy, suggests the presence of a threshold effect (see Figure S3, Supplemental Digital Content).
(A) (B) Fagan's nomogram plot was generated for visually presenting diagnostic performance (see Figure S2, Supplemental Digital Content); given a pre-test probability of 43%, a positive measurement leads to a post-test cancer probability of 71%, whereas a negative measurement leads to a post-test probability of 20%.
As shown in Figures 2-4, the I2 of sensitivity, specificity, PLR, NLR, and DOR were 91.1% (p < 0.0001), 86.1% (p < 0.0001), 86.5% (p < 0.0001), 94.6% (p < 0.0001), and 74.1% (p < 0.0001), respectively.These results of the diagnostic meta-analysis demonstrated that a substantial heterogeneity exists between the included studies.Furthermore, a typical pattern of the "shoulder arm" plot shown in the ROC space, which contains the representation of accuracy, suggests the presence of a threshold effect (see Figure S3, Supplemental Digital Content).The Spearman rank correlation coefficient between the logit of sensitivity and the logit of 1-specificity was determined to be 0.159 (p = 0.183).This suggests that, in addition to the threshold effect, there may be other factors contributing to variations in accuracy estimates among individual studies.
In addition to the threshold effect test, a meta-regression analysis was carried out to explore the possible sources of heterogeneity using the following covariates as predictor variables: the biofluids tested (serum, plasma, and others), the targeted gen used to the integrity index formula (ALU vs. others), the type of controls (healthy, benign, or noncancer controls), and the cohort size (n ≥ 100 vs. n < 100).None of these variables reported any kind of effect on the diagnostic accuracy of cancer (see Table S1, Supplemental Digital Content).Subsequently, subgroup analysis was performed to further explore the potential diagnostic value of cfDI in liquid biopsies.No statistically significant differences were observed in the subgroup analysis, suggesting that none of the subgroups report a potential resource heterogeneity.The results of the subgroup analysis are shown in Table 2.
(A) (B) The Spearman rank correlation coefficient between the logit of sensitivity and the logit of 1-specificity was determined to be 0.159 (p = 0.183).This suggests that, in addition to the threshold effect, there may be other factors contributing to variations in accuracy estimates among individual studies.
In addition to the threshold effect test, a meta-regression analysis was carried out to explore the possible sources of heterogeneity using the following covariates as predictor variables: the biofluids tested (serum, plasma, and others), the targeted gen used to the integrity index formula (ALU vs. others), the type of controls (healthy, benign, or noncancer controls), and the cohort size (n ≥ 100 vs. n < 100).None of these variables reported any kind of effect on the diagnostic accuracy of cancer (see Table S1, Supplemental Digital Content).Subsequently, subgroup analysis was performed to further explore the potential diagnostic value of cfDI in liquid biopsies.No statistically significant differences were observed in the subgroup analysis, suggesting that none of the subgroups report a potential resource heterogeneity.The results of the subgroup analysis are shown in Table 2. .

Publication Bias
The assessment of potential publication bias in each of the selected studies was conducted using Deek's funnel plot asymmetry test, which yielded an overall slope coefficient p-value greater than 0.05.The observed data pattern exhibits symmetry, indicating no evidence of publication bias in the pooled analysis.Therefore, our meta-analysis suggests the absence of significant publication bias (refer to Figure S4, Supplemental Digital Content).

Discussion
Although the 5-year relative survival rate has improved in the past decades for most cancer types [78], it is still necessary to keep developing early diagnostic strategies in order to increase the quality of life the chances of successful treatment for patients [79], and reduce the healthcare costs.In the last few years, liquid biopsies have emerged as an attractive source of molecular biomarkers for cancer diagnosis.Specifically, cfDNA analysis has shown potential as a tool for diagnosis and cancer management [80][81][82].It was described in blood for the first time in 1948 [83], but it was not until 1977 that it started to be studied for its possible link with cancer regarding cfDNA quantification [84].However, the cfDNA concentration is not cancer-specific, as elevated levels of cfDNA have been detected as well in other physiological and pathological conditions [85][86][87][88].Apart from the cfDNA concentration, cfDI has been proposed as a diagnostic cancer biomarker.Several studies have observed that the cfDNA fragmentation pattern in cancer patients is different from that in non-cancer individuals [89][90][91].A great number of fragments longer than 180 bp have been detected in cancer, which leads to the possibility of a necrotic cell death origin [25,92].Based on this theory, the cfDI index was used in numerous studies as a potential cancer biomarker [49,55,62].Initially, a formula based on the threshold (Cp) value for each qPCR reaction was proposed [29].Subsequently, a method was developed to measure the integrity of circulating DNA using ALU repeats, which are the most abundant repetitive elements in the genome, with a simple ratio between the longer and the shorter DNA sequences [34].Although these two methods are different from each other, they report similar measurements [45,47], and both set the stage for the following researchers who keep using repeated sequences of the human genome to measure the integrity index.
To the best of our knowledge, this article is the first meta-analysis to evaluate the diagnostic accuracy of cfDI in different liquid biopsies in cancer patients.A total of 46 articles (72 study units) involving 4286 cases and 3270 controls have been analyzed.According to QUADAS-2 quality evaluation, most of the included studies showed moderate quality.To incorporate cfDI as an early, cost-effective cancer diagnostic proof by itself in clinical settings, the specificity and sensitivity of the test very close to 100% should be achieved.In the present study, the overall diagnostic accuracy of the cfDI for discriminating between cancer and non-cancer patients reported pooled sensitivity and specificity values of 0.70 and 0.77, respectively.These results indicate a moderate sensitivity and a moderatehigh diagnostic specificity.Furthermore, no evidence of publication bias among eligible studies was observed, indicating that the results of the present meta-analysis are reliable.
The summary DOR of 10.63 (95% CI: 8.21-13.77)reflects the diagnostic capacity of cfDI for cancer without being influenced by its prevalence among the population.This value indicates that the odds of having a high cfDI level for cancer patients are more than 15 times higher than the odds for the controls.For its part, the AUC under sROC for cfDI was 0.84, showing its diagnostic capacity.The pooled PLR value of 3.26 means that a person with a high cfDI index has a little more than a 3-fold higher probability of having cancer than a control.On the contrary, the pooled NLR results indicated that a person with a low cfDI index had only a 34% probability of not having cancer.
Additionally, given a pre-test probability of 43% in Fagan's nomogram, correct cancer diagnosis increased to 71% after a positive test (high values of the cfDI index) and reduced to 34% after a negative test (low value of the cfDI index).Overall, all the present results suggest that cfDI is quite specific, which might be helpful in the assessment of cancer diagnosis, but it shows relative sensitivity as a cancer diagnostic biomarker.As cancer involves a complex cascade of molecular events, some studies have combined the cfDI index with other biomarkers, showing an improvement in diagnostic accuracy [36,46,55,58,60].However, this meta-analysis was not evaluated because there was not enough data to carry out an analysis.
Heterogeneity is a significant factor in any diagnostic accuracy meta-analysis; therefore, the possible reasons that could lead to inconsistencies across the studies should be evaluated.In the present work, a high heterogeneity was observed, so a bivariate random-effect model was applied.Despite the fact that the threshold effect is one of the primary causes of heterogeneity in diagnostic test accuracy studies [33], we could not demonstrate with the ROC plane or Spearman correlation the visual deviation denoted by accuracy estimates on forest plots.Furthermore, other possible heterogeneity sources were analyzed, but no significant differences were found among the subgroups.However, other reasons might explain this great heterogeneity in accuracy estimates, such as sociodemographic characteristics (e.g., gender, ethnicity, age, or lifestyle) or clinicopathological conditions (e.g., TNM staging, tumor anatomic location, histological type, or other concomitant diseases).Nonetheless, there was not a sufficient number of articles to assess these variables as a group by themselves.
Apart from the heterogeneity drawback already explained and the different biological effects that were not considered in most of the included studies, other limitations were presented in this meta-analysis.First, the majority of the studies used the DNA repetitive elements' length reported in blood to evaluate cfDI in other liquid biopsies.However, for example, DNase I enzymatic activity is more than 100-fold higher in urine than in the blood [93,94], which leads to smaller size DNA fragments (around 82 bp long) [95,96].In this sense, future investigations should be carried out to establish the proper ratio for each body effluent [97].Last, because of the different thresholds set in the included studies to define a positive test result, a lack of standardization is denoted, which is one of the reasons for the differences seen in the sensitivities and specificities of the test accuracy of the studies.In this line, it is necessary to standardize both the pre-analytical and analytical steps in data interpretation, including cut-off determination.

Conclusions and Future Perspectives
From a clinical perspective, cfDI shows promising value as a minimally invasive biomarker for cancer management.The present research provides updated evidence on the diagnostic potential of cfDI across a wide variety of malignancies, highlighting and reinforcing the growing interest in cfDI as a tumor biomarker.Additionally, this work details the design and methodological characteristics of each included study, emphasizing the significant variability among different research efforts that could influence the overall diagnostic performance of this biomarker.In this sense, subgroup analyses based on the type of biofluid, sample size, DNA repetitive sequence, and type of control were carried out to understand the heterogeneity among the studies better; however, no significant differences in terms of diagnostic accuracy were observed, which suggests the importance to evaluate the impact of other factors in the performance of this biomarker.Therefore, future perspectives must focus on developing robust studies using standardized protocols to establish the real clinical value of cfDI as a liquid biopsy diagnostic tool.For that, it is important to better understand the biological significance of cfDI due to the behavior of this biomarker varying in the different liquid biopsies and malignancies.In this line, the advances in molecular biology techniques, such as next-generation sequencing platforms, could contribute to providing more information about the size of cfDNA fragments and cfDI under different physiological and pathological conditions.Moreover, the cfDI index could be combined with other cancer-related biomarkers, such as conventional tumor markers or epigenetic alterations, improving the performance of liquid biopsies for cancer detection.
Overall, our investigation provides updated knowledge and valuable insights into various aspects of cfDI as a cancer biomarker, identifying the main limitations of the research in this field, which will help the medical and scientific community to address

Figure 1 .
Figure 1.PRISMA flow diagram of literature search and selection criteria.Abbreviations: PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-analysis [31].

Figure 1 .
Figure 1.PRISMA flow diagram of literature search and selection criteria.Abbreviations: PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-analysis [31].

Figure 4 .
Figure 4. Pooled performance of cfDI based on liquid biopsies in cancer diagnosis [29,30,34-77].(A) Forest plot displaying the pooled DOR of cfDI.(B) sROC curve illustrating the pooled estimates of sensitivity, specificity, and the AUC of cfDI.Q* is the point on the sROC curve where the sensitivity equals the specificity.Abbreviations: AUC = area under the curve; cfDI = cell-free DNA integrity; DOR = diagnostic odds ratio; sROC = summary receiver operator characteristic.

Figure 4 .
Figure 4. Pooled performance of cfDI based on liquid biopsies in cancer diagnosis [29,30,34-77].(A) Forest plot displaying the pooled DOR of cfDI.(B) sROC curve illustrating the pooled estimates of sensitivity, specificity, and the AUC of cfDI.Q* is the point on the sROC curve where the sensitivity equals the specificity.Abbreviations: AUC = area under the curve; cfDI = cell-free DNA integrity; DOR = diagnostic odds ratio; sROC = summary receiver operator characteristic.

Table 1 .
Summary of descriptive characteristics of included studies (n =46).