Background and rationale

The standard of care for gynecologic malignancies, such as ovarian, endometrial and cervical cancers, typically involves multimodal approaches, including surgery and systemic therapy. While the extent and distribution of the disease are primary factors in determining the treatment sequence, patient performance status also plays a critical role in decision-making. Decisions regarding surgery and chemotherapy, such as candidacy, sequencing, and regimen selection, significantly influence toxicity, morbidity, and survival outcomes [1,2]. Therefore, individualized assessments of patient prognosis and treatment-associated risks at several points along a patient’s treatment trajectory are crucial. These tailored evaluations form the foundation for optimizing oncologic outcomes and quality of life, and minimizing complications.

Older patients often experience adverse outcomes compared to the younger and healthier patient populations typically included in oncology clinical trials that establish the standard of care [3–7]. They experience a greater risk of complications, long-term side effects impairing quality of life, and shorter survival [8–10]. This disparity is also evident in Gynecologic Oncology, where for example the morbidity and mortality rates of uterine cancer among older patients continue to increase despite novel treatment options [11]. There remains considerable variability in how older adults with cancer tolerate and recover from surgery and systemic treatments [3–7]. As life expectancy increases and chronic disease management improves, chronological age and comorbidities alone fail to fully capture the risk of adverse outcomes or predict treatment responses in this population. Moreover, studies have shown that older patients with cancer often prioritize maintaining function and quality of life over extending survival [12].

Current risk-assessment tools often rely on predictors such as age and comorbidities or focus solely on short-term postoperative complications. They frequently overlook critical patient-centered outcomes such as quality of life, functional recovery, and long-term disability. These limitations underscore the pressing need for more comprehensive strategies and tools to assess risk, enabling personalized decision-making and treatment plans that better address the complexity and heterogeneity of older adults with cancer. Frailty could explain the inter-individual differences in health, function, and recovery, offering prognostic value beyond traditional clinical predictors such as age, comorbidities and cancer stage [13–17].

Age-related frailty in the perioperative and oncological setting

Frailty describes a state of vulnerability, often age- and disease-related, resulting from diminished physiological reserves across multiple systems. It can be assessed using various methods, including questionnaires, physical tests, comprehensive clinical evaluations, or imaging [18–20]. Although no single frailty measure is considered the gold standard, the most widely studied and utilized definition is the Frailty Phenotype, developed by Fried et al. in the Cardiovascular Health Study and the Women’s Health and Aging Studies [18,21,22]. This physical domain-focused model defines frailty as meeting at least three of five criteria: weak muscle strength, slow gait speed, unintentional weight loss, exhaustion, and low physical activity. Over time, additional frailty frameworks have been proposed, incorporating multiple domains beyond the physical to capture a broader spectrum of vulnerability [18–20].

In large cohort and population-based studies of noncardiac surgeries, preoperative frailty has been strongly associated with worse outcomes, including increased risks of short-term morbidity, falls, disability, higher resource utilization, long-term institutional placement, and 1-year mortality [23–27]. However, these studies largely exclude or minimally represent patients undergoing surgery for gynecologic malignancies, leaving a significant knowledge gap.

The prevalence of frailty in oncology may be as high as 50% [24]. In patients living with cancer, emerging evidence suggests that frailty may predict adverse oncologic outcomes [25,28,29], treatment tolerance, and all-cause mortality, independently of chronological age [8,17,24,30,31]. Among gynecologic oncology populations, studies have suggested that frailty is associated with adverse cancer-specific outcomes and treatment-related morbidity, such as reduced tolerance to chemotherapy, increased risk of treatment toxicity and perioperative complications, and decreased overall survival. However, the existing studies are small, retrospective, or focused on specific malignancies [32–39].

Current perioperative risk assessment in gynecologic oncology

Currently, there are no gold-standard tools for perioperative risk stratification tailored specifically to gynecologic oncology patients. While older adults undergoing noncardiac surgery are at risk for various complications, much of the research has focused on cardiovascular events and their preoperative risk stratification. Cardiovascular complications are particularly common in older patients and represent a leading cause of postoperative mortality [40,41].

Guidelines from organizations such as the Canadian Cardiovascular Society and the European Society of Cardiology [42,43] provide clear, straightforward algorithms for perioperative cardiovascular risk assessment. These typically begin with evaluating patient medical history and calculating a risk score, most commonly using the Revised Cardiac Risk Index (RCRI). The RCRI includes six factors, each contributing one point: history of ischemic heart disease, cerebrovascular disease, congestive heart failure, preoperative insulin use, preoperative creatinine levels above 177 µmol/L, and high-risk surgery [44].

For patients with an RCRI score of 1 or higher, or for those aged 65 and older with an RCRI score of 0, the Canadian guidelines recommend postoperative troponin monitoring for the first three days after surgery [42]. This recommendation is based on evidence that myocardial injury after noncardiac surgery (MINS), defined as postoperative troponin elevation due to myocardial ischemia, is a strong independent predictor of 30-day mortality, regardless of symptoms or electrocardiogram (ECG) changes [45]. MINS accounts for the highest attributable risk (34%) of 30-day mortality among major postoperative complications [45] and is strongly associated with 1-year mortality [46,47]. Therefore, while MINS is categorized within cardiovascular risk, its occurrence is a significant predictor of overall postoperative outcomes.

Due to its strong prognostic value and the simplicity of algorithm implementation, cardiovascular risk stratification is widely adopted as the primary or sole formal perioperative risk assessment, including in gynecologic oncology. However, the landmark studies underpinning these recommendations included limited representation of gynecologic oncology patients and did not account for cancer-specific risk factors, such as cancer type and stage [44–46].

In contrast, geriatric multidimensional assessments, which evaluate a broader range of risks and vulnerabilities, are increasingly recommended in international oncology and perioperative guidelines for older adults before treatment initiation [47–50]. Despite these recommendations, such assessments are rarely integrated into routine practice, particularly in gynecologic oncology, where a focused cardiovascular approach remains dominant. This gap highlights the need for more comprehensive and tailored approaches to risk stratification in this patient population.

Feasibility and acceptability of frailty assessment and role of biomarkers

Resource and time constraints are known barriers to routine adoption of more comprehensive perioperative risk evaluation, including frailty assessments [51–54]. For older or unwell gynecologic oncology patients, the additional time, physical tests, and detailed questionnaires required for such assessments could feel burdensome. This underscores the need to prioritize not only the predictive accuracy of frailty assessments but also their feasibility and acceptability in routine clinical practice. Research aimed at identifying and refining practical, patient-centered methods for frailty measurement is therefore essential.

Frailty biomarkers may offer a promising solution and could augment or even replace clinical assessments, enabling more streamlined and scalable evaluations. One such potential biomarker is mitochondrial DNA-copy number (mtDNA), measurable in peripheral blood leukocytes using a cost-effective and scalable assay. mtDNA-CN declines with age, and low mtDNA-CN is a marker of mitochondrial dysfunction, energy reserve depletion, oxidative stress, and systemic inflammation [55]. Mitochondrial dysfunction has been correlated with frailty and its adverse outcomes [55–58]; lower mtDNA-CN is associated with specific components of frailty assessment including with slower gait speed and weaker grip strength [59]. Another emerging biomarker is DNA methylation, which has been used to develop epigenetic frailty risk scores. DNA methylation clocks are the widely accepted instruments to measure biological aging.[60] DNA methylation, as opposed to the DNA sequence, which is overall preserved throughout the lifespan, is indeed the epigenetic process that can capture changes in response to environmental and lifestyle exposures and thus yield greater amount of potentially prognostic information than chronological age. These scores have shown correlations with frailty both at baseline and during long-term follow-up, highlighting their potential as prognostic tools [61]. Such innovations could reduce the burden on patients and clinicians alike, facilitating the broader adoption of frailty assessments in gynecologic oncology and beyond.

The need and role of the FARGO study

There is a clear need for robust prospective research to evaluate the role of frailty assessments through clinical assessment and biomarkers in gynecologic oncology patients who are surgical candidates. Understanding the added value of frailty assessments, either as a complement to or replacement for current cardiovascular-focused approaches, is critical.

In addition, research is needed to explore the feasibility and acceptability aspects of the adoption of frailty assessments in this population. To address these gaps, we designed the Frailty Assessment for Risk prediction in Gynecologic Oncology patients undergoing surgery and chemotherapy (FARGO) study, a multicentre prospective cohort study of patients aged 55 and older.

Study objectives

In patients aged 55 or older with confirmed or suspected gynecologic cancer undergoing laparotomy surgery with or without chemotherapy:

Primary objective

To evaluate the performance of preoperative frailty assessment based on the Frailty Phenotype compared to a perioperative cardiovascular risk assessment (combining the RCRI, age, and occurrence of MINS), in predicting the composite outcome of all-cause death or new disability at six months after surgery.

Secondary objectives

1. To compare the predictive performance of different preoperative/perioperative frailty assessments for all-cause death or new disability at 6 months after surgery. The assessments will include:
   1. a. different frailty tools, such as the whole Frailty Phenotype, the single components of the Frailty Phenotype, and the Clinical Frailty Scale; and
   2. b. a dynamic perioperative frailty assessment (based on a frailty assessment repeated 28 days after surgery, in addition to a preoperative frailty assessment).
2. To explore the predictive performance of frailty assessments upon chemotherapy-related outcomes, including completion, total dose and number of cycles received, treatment toxicity and side effects, decisional regret, and impact on function.
3. To explore the value of frailty assessment when added to a perioperative cardiovascular risk assessment and age, with or without other clinical predictors, in predicting all-cause death or new disability at 6 months after surgery.

To explore the predictive performance of a preoperative frailty assessment and of a preoperative cardiovascular risk assessment based on RCRI, age and MINS upon.

1. 4. postoperative outcomes at 28 days after surgery, including all-cause death or new disability, major vascular events, infection and sepsis, bleeding, new clinically relevant atrial fibrillation/flutter, acute congestive heart failure, length of stay, unplanned admission to intensive care unit, and delirium.
2. 5. all-cause death or new disability at 1 year after surgery.
3. 6. long-term postoperative outcomes, including major vascular events at 6 months and 1 year after surgery, infection and sepsis at 6 months and 1 year after surgery, and all-cause death at 6 months and 1 year after surgery.
4. 7. oncologic outcomes of progression-free survival and overall cancer-specific survival up to 1 year after surgery.
5. 8. To measure feasibility and acceptability of frailty and disability assessments to patients and healthcare providers.
6. 9. To establish a biobank to explore potential blood biomarkers of frailty.

Study design

This is a multicentre prospective cohort study of patients aged ≥55 years-old with confirmed or suspected gynecologic cancer who are candidates for laparotomy with or without chemotherapy.

Eligibility criteria

Study flow and study patient groups

Fig 1 presents the timeline of enrolment, assessments and visits. Fig 2 presents the FARGO patient flow diagram. We expect that some of the eligible patients (group A) will undergo surgery without neoadjuvant chemotherapy. Group A will thus undergo only one preoperative study visit (baseline) within 45 days before their surgery. Patients undergoing neoadjuvant chemotherapy (group B) will have a first baseline study assessment within 45 days before their first chemotherapy treatment. In this group, a second preoperative study visit will be repeated after neoadjuvant chemotherapy is considered completed, at least 18 days post-chemotherapy cycle and within 45 days before surgery. After completion of neoadjuvant chemotherapy, a small proportion of patients will no longer be deemed eligible for surgery by the treating physician (group C). Group C will be included in the evaluation of chemotherapy-related outcomes and will also be asked to complete a follow-up visit 6 months from the date of their registration. Group D, not represented in Fig 2, will consist of those who will eventually not undergo any surgery or neoadjuvant chemotherapy, after being deemed initially eligible and having completed the baseline assessment. We expect group D to consist of an even smaller proportion of recruited patients. For Group D, a follow-up visit at 6 months from the baseline visit will be completed. Group A and B will complete the study assessments and follow-ups. Some patients from group A and group B will undergo adjuvant chemotherapy following surgery. This will not affect their study timeline.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Study timeline: Enrolment, assessments, visits and data collection.

https://doi.org/10.1371/journal.pone.0325651.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Study Flow Diagram.

https://doi.org/10.1371/journal.pone.0325651.g002

Clinical follow-up at each participating centre is based on international oncology guidelines such as those of the National Comprehensive Cancer Network for follow-up for each primary disease site.

Frailty assessments

The primary frailty measurement will be based on the Frailty Phenotype. Research personnel (including research assistants) will assess the following 5 criteria: weak muscle strength (decreased grip strength measured with a dynamometer), slow gait speed (measured on the 15-ft or 4-m walking test), unintentional weight loss (> 10 lbs in the last 12 months), exhaustion (self-reported), and low physical activity (low weekly energy expenditure, calculated using a modified version of the Adult Compendium of Physical Activities [66,67]). For the primary purpose of the study, we will measure the patient’s current level of physical activity; with an exploratory intent, we will gather further information on the usual activity level (before the diagnosis of cancer was made, or before the cancer likely started affecting their performance and lifestyle). Participants receive 1 point for each criterion met, for a total score of: 0–1, not frail (robust); 2–3, intermediate frail (pre-frail); 4–5, frail.

Frailty will also be assessed using the Clinical Frailty Scale (CFS) [22]. The CFS is a global frailty scale according to which the patient is assigned a category from 1 (“Very fit”) to 9 (“Terminally ill”), as a summative clinical judgment; frailty is present if a category≥4 is assigned.

Research personnel are trained to assess the CFS before administering the Frailty Phenotype so that they judgement on the CFS is not affected by the results of the Frailty Phenotype assessment. Both frailty assessments are performed at baseline and at 28 days after surgery. Patients undergoing neoadjuvant chemotherapy will have two preoperative frailty assessments, at enrolment before neoadjuvant chemotherapy, and after the completion of neoadjuvant chemotherapy.

Study outcomes

The predictive performance of perioperative frailty and cardiovascular risk assessments will be assessed upon the following clinical outcomes.

Patient recruitment and informed consent

At the participating sites at academic hospitals in Canada, research personnel will screen the list of patients aged 55 years or older, diagnosed with a suspected or confirmed gynecologic malignancy and requiring a) a surgical intervention performed by laparotomy or b) NACT, with plan for interval surgery via laparotomy, who are scheduled for any preoperative or pre-chemotherapy visit at the participating centres. Clinics involved in the preoperative or pre-chemotherapy assessment of the potential study population as per local practice (such as gynecologic oncology clinics, internal medicine clinics, anesthesiology clinics, etc.) will be engaged in the patient screening process. If a patient meets the criteria, study personnel will approach the patient to obtain their informed consent. Based upon the clinic schedule and access to the patients, a combination of remote and written consent may be used based upon the patient preferences. If the patient chooses remote consent, the signed informed consent form will be returned to the site prior to initiation of study procedures. These informed consent options may be provided to reduce burden on the patient population. Patient recruitment commenced on June 12^th, 2023 and is expected to be completed by December 31, 2025. The list of study sites can be obtained by contacting the corresponding author.

Data collection and follow-up

Fig 1 presents the timeline of the study data collection.

Sample size

The FARGO study is sized for the primary objective. We expect a 20% rate of the primary outcome of death or new disability at 6 months after surgery, based on previous literature on patients undergoing major abdominal surgery [81,82]. We performed preliminary analyses of the VISION [83] cohort. Based on these analyses, we expect that, in our study population, the ability of the baseline RCRI and age combined with the occurrence of MINS to predict 6-month death or new disability, measured as Area Under the receiver operating characteristic Curve (AUC), will not be greater than 0.600 (range 0.550–0.600, Appendix I). A sample size of 250 patients will allow at least 80% power to detect a difference of at least 0.130 in the AUC for a model including only the FP-frailty assessment, compared with the AUC for the model including preoperative RCRI and age combined with MINS (i.e., AUC for FP-frailty model of at least 0.680) [23]. The table in the Appendix III displays the sample size calculation based on these assumptions for a range of scenarios, based on the comparison of two Receiving Operating Characteristic (ROC) curves, 2-sided test, alpha = 0.05, and assumed correlation between the 2 predictive models = 0.5.

We will include 280 patients to conservatively account for ≤ 10% of participants who will be lost before or during surgery, and in whom troponin measurement to assess for MINS will not be possible. In fact, we expect that after recruitment, significantly fewer patients will not undergo or survive surgery. Based on our eligibility criteria, we expect that approximately 50% of participants will undergo chemotherapy, before and/or after surgery, and will be included in our exploratory secondary analyses on chemotherapy outcomes.

Statistical analyses

The primary analysis will compare the performance of preoperative frailty assessment based on the FP and perioperative cardiovascular risk assessment (combining the preoperative RCRI, age and occurrence of MINS), in predicting the composite of all-cause death or new disability at 6 months after surgery. We will compare the predictive performance of FP and of RCRI, age and MINS with the primary outcome by comparing the AUC calculated using logistic regression models. We will perform one logistic regression model including frailty as predictor, and one logistic regression model including the baseline RCRI, age and MINS as predictors. For each model we will then calculate the c-statistics (corresponding to the AUC) and compare them using a non-parametric approach [84]. We will include in the primary analysis only study participants who have at least one postoperative troponin measurement. We anticipate that this will exclude from the primary analysis a very small proportion of patients initially enrolled in the study (i.e., ≤ 2%). We will then perform a sensitivity analysis, accounting for missing troponin data and reasons why troponin was not measured (early death or other reasons). We will secondarily include the surgery indication, based on our inclusion criteria, in both models, as covariate and as effect modifier.

As part of our secondary objectives, we will also evaluate the utility of frailty in predicting 6-month disability or death, when added to the RCRI, age and MINS as predictors by calculating the net absolute reclassification improvement [85].

To compare the predictive performance of different frailty assessments, we will use a similar approach as for the primary outcome: we will compare the c-statistics of logistic regression models, each including one type of frailty assessment. To examine the added value of a dynamic perioperative frailty assessment on the primary outcome, we will use a logistic regression model and add preoperative frailty and change in frailty at 28 days after surgery, either as absolute change or as percent change as covariates.

Logistic regression or time-to-event models will be performed for the association of frailty and perioperative cardiovascular risk assessment with other outcomes, including chemotherapy-related outcomes, as appropriate.

Effects sizes will be presented together with 95% confidence intervals. The threshold for statistical significance will be set as p < 0.05.

The physicians’ and patients’ acceptability scores will be compared between frailty instruments using Wilcoxon signed rank tests. The mean and standard deviation (in seconds) required to complete each frailty instrument will be compared using Student t tests. The numbers of missing values will be compared using 2x2 tests.

Ethical considerations

This study will be conducted in compliance with the protocol, principles laid down in the Declaration of Helsinki, and all applicable laws and regulations. Before study initiation, each site must have written and dated approval/favorable opinion from the Institutional Review Board/Independent Ethics Committee for the protocol and consent form. The FARGO study was granted Ethics approval by the Ontario Cancer Research Ethics Board (OCREB), Clinical Trials Ontario Project identification number 3714. Amendments to the protocol have received Ethics approval.

Current status

To date 177 participants have been recruited across three Canadian sites. We expect recruitment to be completed by the fall of 2025.

Appendix I. Clinical outcomes definitions

The following list provides the definitions for the clinical outcomes for which the predictive ability of perioperative frailty and cardiovascular risk assessments will be assessed in the study.

Appendix II – Measuring acceptability of frailty assessments

Physicians’ acceptability.

Study personnel will invite a convenience sample of consenting physicians involved in the perioperative clinical assessment of the study population (e.g., the Gynecologic Oncology group, the Perioperative Medicine Service, etc.), with no experience with frailty tools, to participate in the administration of the FP and CFS. The study personnel will subsequently administer the acceptability questionnaire (OADRI) to the physicians. The questionnaire is a modified version of the OADRI where only items deemed relevant are retained. Physicians are asked to rate each item, for each tool, along a 7-point Likert scale including the following options: strongly disagree, moderately disagree, slightly disagree, no opinion/don’t know, slightly agree, moderately agree, strongly agree.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

https://doi.org/10.1371/journal.pone.0325651.t001

Patients’ acceptability.

Patient’s acceptability of frailty assessments will be evaluated administering a questionnaire in a purposeful sample of the main study participants. The first 10 patients at each site will be asked about the acceptability of the Frailty Phenotype Assessment and WHODAS questionnaire shortly following their completion at the baseline study visit. The next 10 patients (different from those asked at baseline) will be asked about the acceptability of the Frailty Phenotype and WHODAS questionnaire shortly following their completion at the 28-day follow-up visit. In total, 20 patients at each site will be asked about the acceptability of the FP and WHODAS, 10 at each indicated visit.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

https://doi.org/10.1371/journal.pone.0325651.t002

Appendix III. Sample size calculations

The sample size calculation is based on the comparison of two receiver operator characteristic curves (ROC), 2-sided test, with alpha = 0.05. The assumed AUC for the RCRI, age, and MINS model is 0.55 (AUC1). The assumed AUC of the frailty assessment model is between 0.65–0.75 (AUC2). The assumed AUC difference is approximately 0.13. We expect an incidence of the primary outcome (all-cause death or disability at 6 months after surgery) to be 20%.

A sample size of 250 patients will give us at least 80% of power to detect a difference of at least 0.130 in the AUC for a model including solely the FP-frailty assessment, compared with the AUC for the model including preoperative RCRI and age combined with MINS.

The assumed correlation between the two predictive models is 0.50 for the positive group (those experiencing the outcome) and 0.50 for the negative group (those not experiencing the outcome).

We will include 280 patients to conservatively account for ≤10% who will be lost before or during surgery, or in whom troponin measurement will not be possible.

Software: PASS 13 (NCSS, LLC. Kaysville, Utah, USA. www.ncss.com)

Sample size calculation.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

https://doi.org/10.1371/journal.pone.0325651.t003