Minimal residual disease in solid tumors: Clinical applications and future directions.

INTRODUCTION
Minimal residual disease or molecular residual disease (MRD) refers to the cancer cells or their fragments that remain in the body after treatment completion. These cells or fragments are not detectable by the traditional imaging or protein methods but can be indicators of persistence, recurrence or metastasis of the cancer.
1
Unlike histology and stage that predict the risk of recurrence, MRD could directly identify the presence of residual cancer.
2

Circulating tumor DNA (ctDNA) has emerged as a dynamic biomarker of MRD that enables real‐time tracking of the cancer as opposed to a static biopsy, and monitoring treatment response as well as potentially identifying drug resistance.
3
CtDNA is a component of cell‐free DNA that originates from tumor tissue in the patient's plasma.
4
It can be detected by methods such as polymerase chain reaction (PCR), and next‐generation sequencing.
5
Current ctDNA genotype assays have not reached optimal clinical sensitivity, especially in patients with early‐stage disease or in immediate post‐treatment settings, due to lower tumor shedding compared to patients with advanced‐stage disease.
4
Thus, there is a need for more sensitive assays for detecting tumor ctDNA.
There are currently two main types of ctDNA testing: tumor‐informed and tumor‐agnostic. Tumor‐informed testing (e.g., Signatera assay
6
) employs a personalized approach that uses the patient’s cancer tissue (following biopsy or surgery) to create a panel of mutations specific to the individual tumor.
6
It is reported to be highly sensitive and specific,
6
but it can be costly, time‐consuming and requires a prior invasive procedure to obtain the tissue sample. In contrast, tumor‐agnostic testing (like Guardant360 assay) is done via pre‐made panels, which do not require tissue sampling, have lower sensitivity and specificity, but are more likely to be cost‐effective with a rapid tunnaround.
7
Additionally, they have demonstrated utility for MRD detection in multiple cancers, such as NSCLC and colorectal cancer, supporting broader pan‐tumor application.
7
Test selection should be based on the clinical context.
8
Table 1 presents a list of some of the ctDNA assays used in the adjuvant setting.
One of the main clinical applications of ctDNA MRD is its potential use as a prognostic and predictive marker for patients with cancer to further estimate a patient’s outcome and help assess the patient’s likeliness to benefit from treatment, respectively.
9
Although MRD is currently used as a prognostic and predictive marker to guide treatment decisions, and as a surrogate end point for clinical trials in hematologic malignancies (acute lymphoblastic leukemia, acute myeloid leukemia, multiple myeloma, etc.),
10
,
11
it is not widely integrated into standard‐of‐care clinical practices in solid tumors. One of the major controversies surrounding MRDs use is that it could allow the detection of tumors in advance of clinical symptoms (longer lead time), without evidence showing this can lead to better outcomes for the patients.
12
Furthermore, it can lead to potential harm including added costs, increased diagnostic procedures, patient anxiety and although rare, potential exposure to toxic unnecessary therapies in case of false‐positive findings.
12
Several trials are currently examining the clinical utility of MRD testing including ASPIRA
13
and ALTAIR
14
that are included in Tables 2, 3, and 4. Hence, the main benefits of MRDs thus far have been in approximating recurrence rate and likelihood of cure that can be communicated to the patients.
15

In this comprehensive review article, we explore the recent literature on the topic and provide practical updates on the clinical applications, limitations and future directions of MRD in solid tumors.

ctDNA and MRD in gastrointestinal cancers
One of the largest studies exploring the clinical utility of ctDNA MRD in gastrointestinal (GI) tumors, particularly in colorectal cancer (CRC) is the CIRCULATE‐JAPAN multinational initiative, which includes three trials: the GALAXY, VEGA, and the ALTAIR trials.
16
The GALAXY trial consisted of monitoring MRD status in patients with stage II–IV CRC after surgery with ctDNA. An interim analysis of this arm showed that ctDNA positivity during a 60‐day postoperative window was associated with inferior disease‐free survival (DFS; Hazard Ratio (HR), 11.99; p < .0001) and overall survival (OS; HR, 9.68; p < .0001) compared to ctDNA‐negative patients.
17
Notably, patients with positive ctDNA results who received adjuvant chemotherapy had sustained ctDNA clearance that correlated with statistically significant improvement in DFS and OS compared to patients with positive ctDNA results who did not receive adjuvant chemotherapy, whereas no significant benefit to adjuvant chemotherapy was observed in patients with ctDNA‐negative results,
17
supporting the role of ctDNA as a stratification tool for adjuvant treatment. The VEGA arm is currently assessing the noninferiority of observation alone compared to standard adjuvant chemotherapy in ctDNA‐negative patients, whereas the ALTAIR arm is assessing the superiority of adjuvant chemotherapy compared to no treatment in ctDNA‐positive patients.
16
,
18
No results have been published yet for the later arms. The CIRCULATE‐US (NCT05174169) is an ongoing trial exploring ctDNA‐guided escalation and de‐escalation of adjuvant chemotherapy in stage II–III colon cancer, comparing surveillance versus immediate therapy in ctDNA‐negative patients and standard versus intensified regimens in ctDNA‐positive patients.
19
Notably, optimizing ctDNA positivity thresholds for sensitivity and specificity proved more effective for guiding adjuvant chemotherapy and de‐escalation decisions than a simple binary cutoff.
20
The COBRA trial (NCT04068103),
21
a multicenter, randomized phase 2/3 study, was discontinued early, after showing that ctDNA‐positive patients post‐resection did not derive benefit from adjuvant chemotherapy,
22
highlighting that the routine use of MRD testing to guide treatment decisions remains controversial and underscoring the need for further research to define the appropriate clinical context for its implementation. It is important to note that many of the mentioned trials were conducted outside the United States and therefore clinicians and researchers should interpret the results within their specific clinical and health care setting. Accordingly, Table 5 highlights ongoing observational and interventional studies within the United States.
Other ongoing trials include the DYNAMIC‐I trial (ACTRN12615000381583), which explores how ctDNA status (positive or negative) can guide adjuvant chemotherapy in stage II colon cancer patients (i.e., by not giving adjuvant chemotherapy to patients with ctDNA negative results). Initial results indicated a noninferior 2‐year recurrence‐free survival (RFS) among patients with ctDNA‐guided management compared to standard risk‐based (non‐ctDNA guided) management.
23
Long‐term follow‐up demonstrated a noninferior 5‐year RFS and OS with ctDNA‐guided versus standard management (88% and 93.8% vs. 87% and 93.3%, respectively).
24
Interestingly, a post hoc analysis of ctDNA clearance predicted a 5‐year recurrence‐free probability of 97% versus 0% for ctDNA persistence (p < .001).
24
A follow‐up study, DYNAMIC‐III, explored the impact of ctDNA‐guided de‐escalation or escalation of adjuvant chemotherapy in patients with resected stage III colon cancer. This study was presented at the American Society of Clinical Oncology (ASCO) 2025 meeting, suggesting that the recurrence risk increased with increasing postoperative ctDNA levels; further analysis of post‐adjuvant chemotherapy ctDNA status are currently ongoing.
25

Several other recent trials have reported on the role ctDNA as a prognostic and predictive marker in CRC, including the COSMOS‐CRC‐01 trial including patients with stage I–III CRC. The study concluded that the presence of postoperative ctDNA was significantly associated with recurrence risk, with a median lead time of 5.3 months before radiographic detection.
26
Similarly, in a large Danish cohort of 851 patients with stage II–III CRC treated with curative intent, a tumor‐informed test showed that both postoperative and serial ctDNA detection were strongly prognostic for cancer recurrence rates (HR, 11.3 and 30.7, respectively; p < .001).
20
This study also highlighted the challenges in detecting postoperative ctDNA in metastatic CRC based on different locations of metastasis. For example, ctDNA positivity was lower in lung metastases and peritoneal metastases compared to metastasis to lymph nodes, liver, other sites and multiple sites,
20
which argues the need for higher sensitivity of MRD tests when used in disease monitoring.
27
Another report from The TRACC part B trial in the United Kingdom demonstrated that postoperative ctDNA detection was the strongest independent predictor of recurrence in multivariable analysis.
28
Discordant results were attributed to sampling frequency, low‐shedding tumors, or immune‐mediated disease control in ctDNA‐negative individuals whose cancer recurred and second primaries or misclassification of clonal hematopoiesis variants in ctDNA‐positive variants that remained disease free. Part C of this trial is actively evaluating ctDNA‐guided ACT treatment de‐escalation.
28

Currently, the use of ctDNA is a standard clinical practice for CRC in the adjuvant setting. Although there is no consensus on the optimal time and frequency to test for ctDNA in CRC, real‐world reports indicated that ctDNA is commonly tested for in the immediate postoperative setting and every 3 to 6 months afterward.
29

Similar to CRC, MRD has been explored in rectal cancer and other GI tumors. In locally advanced rectal cancer (LARC), where the primary cause of treatment failure is generally due to distant metastasis,
30
early identification of high‐risk individuals is prioritized. A multicenter study of LARC patients from China found that ctDNA levels correlated with size and stage of the tumor, and that ctDNA clearance following neoadjuvant chemoradiotherapy (CRT) predicted improved survival outcomes.
31
Additionally, a report from the DYNAMIC‐rectal study recently showed that the recurrence rate in treated ctDNA‐positive patients was lower when compared to untreated historical controls.
32

Other studies in different GI tumors
33
,
34
have reinforced the role of MRD in both outcome prognosis and guiding treatment decisions. In metastatic GI tumors, ctDNA was reported to predict treatment response by detecting mutations like BRAF V600E post‐chemotherapy
35
and tracking tumor evolution through RAS status.
36
,
37
Similarly in advanced gastrointestinal stromal tumor (GIST), ctDNA next‐generation sequencing (NGS) identified imatinib‐resistant mutations missed by tissue testing, enabling personalized second‐line therapy and patients with KIT exons 13/14 benefited from sunitinib, whereas those with exons 17/18 responded better to ripretinib.
38
Ongoing GI trials examining the role of CT DNA in the adjuvant setting are listed in Table 2.

ctDNA and MRD in lung cancer
In the landmark Checkmate 816 phase 3 trial, increased event‐free survival (EFS) and pathological complete responses (pCR) were reported following neoadjuvant nivolumab plus chemotherapy compared to chemotherapy alone for resectable lung cancer.
39
A notable exploratory analysis revealed that EFS and pCR were correlated with ctDNA clearance in both treatment groups, highlighting the prognostic potential of ctDNA in lung cancer.
39
This was followed by a recent report from the Checkmate 77T trial in which ctDNA clearance following treatment was identified as an independent predictor of treatment response.
40
Similar results were also seen in the phase 3 AEGEAN trial where overall DFS rates at 12 months were worse in MRD‐positive (14.3%; 95% CI, 2.4–36.3) versus MRD‐negative patients (89.3%; 95% CI, 82.6–93.5).
41
Other studies have also noted the prognostic
42
,
43
and predictive
44
,
45
,
46
potential of ctDNA in lung cancers. In addition, Tan et al.
47
used a TI‐assay to risk‐stratify patients with early‐stage non–small cell lung cancer (NSCLC).
Other studies evaluated the role of ctDNA at baseline and suggested that positive ctDNA after surgical resection was strongly associated with a high risk of relapse. The authors also remarked that detection in early‐stage NSCLC is difficult, with a limited sensitivity and variable performance across different NSCLC subtypes.
47
Another study by Herbst et al.
48
in a post hoc analysis of the ADAURA trial noted that at randomization, ctDNA detection rates correlated with disease advancement: ctDNA was detected in 0% of stage IB disease, 39% of stage II disease, and 61% of stage III disease. As well, MRD detection preceded imaging‐based DFS events by a median of 4.7 months.
Currently, there are no Food and Drug Administration (FDA)‐approved MRD tests for lung cancer. However, ongoing and future trials are expected to help establish a standardized usage and interpretation of the ctDNA testing in lung cancer. The MERMAID trials 1 and 2 will assess the benefits of adjuvant therapy in patients with resected, stage II–III MRD‐positive NSCLC,
49
and the efficacy and safety of durvalumab versus placebo in patients with resected, stage II‐III NSCLC who become MRD‐positive after curative‐intent therapy,
50
respectively. Ongoing lung cancer trials examining the role of CT DNA in the adjuvant setting are listed in Table 3.

ctDNA and MRD in breast tumors
The role of ctDNA is currently established in advanced breast cancer and originated with ctDNA testing for actionable mutations in metastatic cancer.
51
Most notably, the FDA approved elacestrant for ER‐positive, HER2‐negative, ESR1‐mutated advanced or metastatic breast cancer with a companion ctDNA assay Guardant360 CDx,
52
following results from the EMERALD trial (NCT03778931) that demonstrated significantly improved progression‐free survival (PFS) with elacestrant for people with ctDNA detected ESR1 mutations.
53
A subsequent meta‐analysis of metastatic breast cancer studies identified significant associations of ctDNA detection of TP53 and ESR1 alterations with worse survival, further highlighting the prognostic value of ctDNA.
54
In the phase 3 double‐blinded randomized SERENA‐6 trial, patients with hormone receptor–positive, HER2‐negative advanced breast cancer who developed an emergent ESR1 mutation detected by ctDNA experienced a 56% reduction in risk of progression or death when switched to camizestrant (a selective estrogen receptor degrader) plus a CDK4/6 inhibitor compared with continuing an aromatase inhibitor regimen (HR, 0.44; 95% CI, 0.31–0.60; p < .00001).
55
Median PFS was prolonged to 16.0 months in the camizestrant arm versus 9.2 months, with consistent benefit across all subgroups.
55

The role of ctDNA in metastatic breast cancer is currently being assessed in early‐stage breast cancer. A pilot study, including 30 patients with primary breast cancer who were monitored with ctDNA for up to 12 years following surgery and adjuvant chemotherapy, suggested that ctDNA could be detected ahead of clinical or radiologic relapse with a lead interval of up to 38 months.
56
More recently, a patient‐informed ctDNA assay (CloneSight) was reported to detect ctDNA positivity in 50% of recurred patients with hormone receptor‐positive early breast cancer up to 68 months in advance of recurrence, while it remained negative in 93% of patients who did not recur.
57
A study using a NGS panel was able to detect distant recurrence with a lead time of 3.4–18.5 months in 79% of patients with early‐stage breast cancer, whereas ctDNA was negative in patients who did not recur.
58
Another study reported a clinical sensitivity of 76.9%, specificity of 100%, and a median lead time of 11.7 months from clinical relapse for patients with early‐stage breast cancer using the ctDNA‐based MRD personalized cancer monitoring (PCM) assay.
59

The c‐TRAK‐TN trial employed a tumor‐informed Signatera assay to assess ctDNA‐based MRD detection in triple‐negative breast cancer (TNBC) and evaluated the addition of pembrolizumab in patients who were ctDNA‐positive. However, the trial failed to achieve its goal because 72% of patients already had metastatic diseases on ctDNA detection.
60
A recent systematic review and meta‐analysis showed that post‐neoadjuvant (before or after surgery) ctDNA positivity is associated with higher recurrence rates and worsened overall survival.
61
However, many of these patients receive several adjuvant therapies that can affect long‐term outcomes, therefore, it remains unclear if therapeutic intervention for patients with ctDNA positive testing can improve outcomes.
61

In summary, despite the interesting results on the prognostic ability of ctDNA in breast cancer, no clear association between its monitoring and improved patient outcomes has been established yet, nor has strong evidence supported its role in guiding treatment escalation or de‐escalation.
62
,
63
Ongoing and future trials should help further guide the clinical application of ctDNA in early‐stage breast cancer. Promising results from the PRE‐PHENIX trial, a multicenter observational study that explores the prevalence of MRD measured by Guardant Reveal in patients with HER2‐positive metastatic breast cancer on long‐term first‐line trastuzumab‐pertuzumab maintenance, were presented at the 2025 ASCO meeting. Guardant Reveal identified that 10 of the 11 patients (91%) with disease progression had MRD, whereas no MRD was found among the 32 patients with Complete Response.
64
A prospective study (PHENIX) to investigate trastuzumab‐pertuzumab maintenance interruption by ctDNA monitoring is planned.
64
Table 3 presents a list of ongoing examining ctDNA utility in the adjuvant setting.

ctDNA and MRD In other solid tumors
In glioblastoma multiforme, efforts are currently being made to detect MRD in cerebrospinal fluid (CSF) postoperatively.
65
A recent trial examined the feasibility of using dynamic tumor in situ fluid ctDNA (TISF‐DNA) to monitor MRD and evaluate postsurgical treatment response.
66
Patients with positive TISF‐ctDNA (n = 62.2%) showed a significantly higher risk of recurrence (HR, 2.512; 95% CI, 1.264–4.993, p = .0054). Importantly, TISF‐ctDNA positivity preceded imaging signs of recurrence by a median of 71 days.
66
In two other studies, TISF‐ctDNA MRD status was strongly correlated with clinical outcomes,
67
,
68
highlighting the potential to use TISF‐ctDNA as a more accurate monitoring tool for tumor recurrence and treatment efficacy in glioblastoma. However, the challenge that presents with CSF MRD is the feasibility and invasiveness of obtaining CSF for repeated analysis compared to standard imaging monitoring.
69

In ovarian cancer, an ancillary analysis of the CHIVA phase 2 GINECO trial analyzed the ctDNA of patients with advanced epithelial ovarian cancer at diagnosis and after each neoadjuvant chemotherapy (NACT).
62
The results suggested that a significant decrease in ctDNA positivity after one NACT cycle was independently associated with improved PFS, OS, and higher eligibility for interval cytoreductive surgery.
70
In locally advanced cervical cancer, a recent analysis of a large global patient population from the CALLA trial presented at the 2025 ASCO meeting showed that high baseline ctDNA was associated with increased risk of progression or death. Interestingly, lower levels of ctDNA following treatment with durvalumab plus chemoradiotherapy (CRT) compared to CRT alone correlated with improved survival, particularly among patients with PD‐L1 tumor‐associated positivity ≥20%.
71
This analysis supports the potential utility of using ultrasensitive ctDNA analysis to guide future treatment decisions in cervical cancer.
In genitourinary cancer, the prognostic role of ctDNA in advanced prostate cancer was recently evaluated. In metastatic castration‐resistant prostate cancer, ctDNA measured at baseline and during treatment independently predicted poorer outcomes with enzalutamide, providing robust prognostic value than traditional biomarkers.
72
Those results were further are corroborated by Knuston et al.
73
Interestingly, a similar outcome was observed in muscle‐invasive bladder cancer, where the detection of ctDNA post–radical cystectomy was strongly associated with shorter DFS, comparing to patients with undetectable ctDNA with better outcomes and did not benefit from adjuvant therapy. This suggests a potential therapeutic role in bladder cancer, beyond prognostic implications.
74
Several additional studies have further demonstrated the prognostic and predictive potential of MRD detection in genitourinary cancers.
75
,
76
Importantly, urinary tumor DNA profiling, which identifies mutations associated with urothelial carcinoma, alternative to ctDNA (blood), has also been successful in predicting recurrence, and enhancing personalized treatment strategies.
77
,
78
Although plasma ctDNA allows minimally invasive serial monitoring, it may miss site‐specific tumors that urine or CSF ctDNA could detect; thus, analyte choice should be guided by clinical context.
In addition, recent studies have emerged recently with similar results on the predictive and prognostic value of ctDNA‐MRD in melanoma,
79
endometrial cancer,
80
and soft tissue sarcomas.
81

Limitations of ctDNA MRD In solid tumors
ctDNA‐based MRD testing in solid tumors has demonstrated significant prognostic and predictive value, but several limitations must be addressed before it can be widely adopted in clinical practice. This methodology is particularly important in the setting of tumor heterogeneity. Single‐site biopsies may miss the full genomic profile of a tumor (spatial heterogeneity) and fail to capture subsequent clonal evolution, especially in advanced or previously treated disease.
82
,
83
The addition of liquid biopsy assays has been revolutionary in addressing this issue by providing a noninvasive, dynamic method of studying whole tumor genomics, albeit with limitations
82
,
83
,
84
,
85
Some of these limitations were discussed above. A notable limitation is that ctDNA assays typically targeting a limited set of known mutations, might miss novel subclones that drive relapse, even in tumor‐informed assays, potentially leading to false‐negative results.
86
The novel technologies that are currently in development, such as methylation pattern‐based sequencing and novel ultra‐sensitive mutation detection methods could further refine and improve these assays.
28
,
48
,
87
,
88

Another limitation is the variable tumor shedding, or release of tumor fragments including ctDNA into the bloodstream,
89
based on stage, location, and timing of the assay. Early‐stage tumors, brain tumors, and certain solid tumors with metastasis to the lungs or peritoneum tend to have lower ctDNA shedding, which can yield false‐negative results.
20
,
90
,
91
,
92
Conversely, perioperative complications may elevate MRD levels, and subsequently confound assessment.
93
Furthermore, the lack of standardization across assays, such as variations in sample collection, processing, sequencing depth, and mutation panels, limits results generalizability for clinical use.
94
In addition, the variability in currently available assays makes it difficult to compare results from different studies and consequently to have standardized guidelines for interpretation of MRD assays and subsequent management.
95
,
96
,
97

Another potential limitation of the frequent use of ctDNA is the high cost of ctDNA testing and could be one of the major obstacles to its adoption into standard of care usage. Kramer et al.
98
examined the cost‐effectiveness of ctDNA‐guided treatment by simulating 1000 stage II colon cancer patients in the Netherland. They measured costs, life years, quality‐adjusted life years (QALYs), recurrences, and cancer‐related deaths. Although the study concluded that the most clinically beneficial strategy is combining MRD testing with standard of care: mismatch repair status (MMR) and pathological tumor stage (PT), to determine the need for adjuvant chemotherapy, it noted that this is not currently cost‐effective under the Dutch willingness‐to‐pay (WTP) threshold (€50,000 per QALY) as it resulted in an incremental cost‐effectiveness ratio (ICER) of €67,413 per QALY.
98
Notably, the study mentioned that this combination strategy could be cost‐effective if the costs of ctDNA testing were lower than €1500 per patient, or if ctDNA‐positive patients respond substantially better to ACT than ctDNA‐negative patients, or if the ctDNA test performance improves substantially.
98

In contrast, a budget impact analysis conducted in the United States incorporating age‐specific incidence of colon cancer, use of adjuvant chemotherapy, costs associated with ctDNA testing, drug acquisition, administration, surveillance, and adverse events, concluded that ctDNA testing led to cost savings for both commercial and Medicare Advantage payers.
99
The cost savings increased with higher test adoption rates and were mainly driven by reduced adjuvant chemotherapy use in ctDNA‐guided treatment compared to standard clinical evaluation.
99
It is important to know that cost‐effectiveness may vary substantially between health systems and require further evaluation.

CONCLUSION
In conclusion, ctDNA‐based MRD detection is reshaping cancer care by identifying high‐risk patients, guiding personalized treatment, and monitoring treatment response. Current evidence supports the clinical use of ctDNA most strongly in colorectal cancer, with consistent prognostic and emerging predictive utility, whereas in lung, breast, and other solid tumors ctDNA primarily serves as a prognostic biomarker, with treatment‐guiding applications remaining investigational. Ongoing trials in the United States (Table 5) and worldwide (Tables 2, 3, 4) seek to explore ctDNA guided interventions and whether these can be translated to better patient outcomes Although standardization, sensitivity, and cost remain a challenge, this field is rapidly evolving with ongoing and future clinical trials expected to further establish its value and guide its safe implementation in everyday clinical settings. As research continues to unfold, the insights gained from ctDNA will undoubtedly shape the future of personalized cancer treatment.

FUTURE RESEARCH DIRECTION
Future research efforts should aim to enhance the sensitivity and specificity of ctDNA assays to improve their clinical utility. Advances in sequencing technologies, including ultra‐deep sequencing and whole genome sequencing, enable more reliable detection in early‐stage and low‐shedding tumors.
88
,
100
,
101
Additionally, the development of comprehensive genomic panels that include noncanonical variants and epigenetic markers, such as DNA methylation, could potentially increase detection sensitivity and broaden the applicability of ctDNA assays across diverse tumor types.
28
,
87
Machine learning algorithms may improve predictive accuracy and enable more precise risk stratification, thereby supporting the advancement of personalized treatments.
102
,
103

Another focus of future research is optimizing ctDNA monitoring schedules
Longitudinal sampling, where ctDNA is measured at multiple time points during treatment or follow‐up, has demonstrated greater consistency and clinical relevance than single time‐point measurements.
59

,

104

97
Accordingly, standardizing sampling schedules would be critical to ensure accurate detection and effective disease monitoring.
47
,
48
Admittedly, these trials are difficult and costly to conduct, due to three major challenges: the high level of variability in ctDNA detection, the extensive costs associated with trial requirements, and the complex logistics and ethical considerations involved.
105
Finally, future research should clearly determine whether earlier detection of recurrence or metastasis and subsequent management changes using MRD would actually improve patient outcomes and survival.
62

AUTHOR CONTRIBUTIONS

Theresa Abdo: Conceptualization; investigation; methodology; project administration; resources; writing—original draft; writing—review and editing. Sacha Yaghi: Writing—original draft. Ahmad Alhalabi: Writing—original draft; writing—review and editing. Mohammad Aloran: Writing—original draft. You Li: Writing—review and editing. María Herrán: Writing—review and editing. Rami Tfayli: Writing—review and editing. Thomas A. Samuel: Supervision; writing—review and editing. Zeina Nahleh: Supervision; writing—review and editing.

CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest.