The influence of frailty on perioperative outcomes in patients undergoing surgical resection of liver metastases: a nationwide readmissions database study

Introduction

Liver metastases are neoplasms that have spread from cancer elsewhere in the body [1], arising most frequently from colorectal, pancreatic, and breast cancers. In fact, 50% of patients with colorectal cancer are diagnosed with liver metastases [2]. The liver is the most common organ affected by metastasis, because of its large blood supply [1,3]. As the incidence of colon cancer continues to rise, it is increasingly important to categorize its association with liver metastases [4]. The median 1-year survival rate of patients with liver metastases (15.1%) is significantly lower than the 1-year survival rate for patients diagnosed with non-hepatic metastases (24.0%). In a study that reviewed 2.4 million patients diagnosed with any type of cancer, 5.14% presented with liver metastases at the time of initial diagnosis. The most frequent primary cancer sources in this study were the pancreas (35.6%) and colon-rectum (26.9%) [5]. Despite their advanced cancer stage, most of these patients were asymptomatic, with only some reporting constitutional symptoms [1]. In fact, liver metastases are more common than primary liver cancer in the US, with 5-year survival rates ranging around 25% for those not receiving early surgical intervention [6]. The most common treatment of liver metastases remains surgical resection [7], although these cases are frequently inoperable because of the heavy metastatic burden. Even with resection, the prognosis of this disease remains extremely poor, with recurrence of disease in two thirds of patients [7,8].

Frailty is an important factor affecting patients’ health outcomes, as it reflects the patients’ overall physiological reserve [9]. Frailty is defined as an age-associated condition that reduces the patient’s physiologic ability to handle stressors, both chronic and acute [10]. Regarding patients undergoing elective surgery, frailty has been reported to be a more accurate predictor of outcomes than an array of other patient demographics, including age [11]. In fact, frailty has proved to be an independent, more accurate preoperative predictive risk factor, even after adjustment for socioeconomic status, depression, and disability [12].

Because frailty is an important risk factor affecting health outcomes, we aimed to investigate its role in predicting outcomes in patients with liver metastases who underwent surgical resection. We hypothesized that frail patients would have a higher rate of postoperative complications and longer hospital stays. Finally, using statistical modeling and predictive analytics, we investigated the relationship between patient’s frailty status and perioperative outcomes. The goals of this study were to help surgeons better identify which patients are metastasectomy candidates, to improve perioperative management for frail patients, and to assist physicians in communicating more accurate prognoses to patients diagnosed with secondary metastatic disease of the liver.

Materials and methods

Patient sample

Between 2016 and 2017, we identified a total of 28,781 inpatient admissions with ICD-10 codes for liver resection procedures (ICD-10: 0FT0xZZ, 0FBxxZZ). Within this cohort, appropriate coding was utilized to identify 10,799 (37.5%) patients who underwent a liver resection procedure for liver metastases. Frail patients were identified using the Johns Hopkins Adjusted Clinical Groups (JHACG) frailty-defining diagnosis indicator, which uses 10 categories of ICD-10 codes (malnutrition, dementia, vision impairment, decubitus ulcer, urine control, weight loss, fecal control, social support, difficulty walking, and history of a fall) to predict a patient’s frailty status [13]. A patient is deemed categorically frail if at least one of these comorbidities has been discovered. Frailty is measured over 5 phenotypic characteristics, including accidental weight loss, tiredness, poor energy expenditure, limited grip strength and/or sluggish walking pace. This measure, which takes into account the decline in a number of physiological systems, was created to help medical professionals identify people more susceptible to suffering negative health effects [13]. Several studies have confirmed the clinical validity of the JHACG frailty-defining diagnosis indicator [13-16].

Based on the above, the cohort was then subdivided into frail (n=766) and propensity score matched non-frail (n=749) patients. Nearest-neighbor propensity score matching for age, sex, Elixhauser Comorbidity Index (ECI), insurance type, median income by ZIP code, and NRD discharge weighting was performed using the R “MatchIt” algorithm [17]. In this technique, parametric models are chosen based on the minimum “distance” parameter, determined through logistic regression models that minimize the propensity score with no replacement. MatchIt improves parametric statistical models and reduces model dependence by preprocessing data with semi-parametric and non-parametric matching methods. Model balance, defined as the similarity of empirical covariate distributions between the 2 groups undergoing propensity matching, is analyzed and the model with the best balance is selected to ensure the best model fit (Fig. 1). Complications queried for analysis in this study included postoperative infections, acute posthemorrhagic anemia, ileus, wound dehiscence, mortality, readmission rates, urinary tract infection (UTI), pulmonary embolism (PE), deep vein thrombosis (DVT), inpatient length of stay (LOS), costs, and discharge disposition. Nonroutine discharges were defined as discharges to places other than home (e.g., skilled nursing facility, home health care, short-term care facility, etc.)

Figure 1 Distribution of propensity scores following matching. Frail patients are shown as Matched Treatment Units and propensity matched non-frail patients are shown as the Matched Control Units. The Unmatched Control Units represent non-frail patients who were not chosen by the propensity matching algorithm. The comparable distribution of patients in both matched treatment and control units implies achievement of excellent propensity score matching

Results

Demographics

The average age of the frail cohort was 61.5±14.2 years and 49.0% were female. The average age of the non-frail cohort was found to be 62.7±13.4 years and 48.6% were female. Because the 2 cohorts were propensity score matched, the age, sex, ECI, insurance type and median income quartile by ZIP code did not differ statistically between the 2 cohorts. No significant differences in hospital size (P=0.63) or teaching status (P=0.66) were found between the 2 cohorts. However, significant differences were found between frail and non-frail patients when comparing discharge disposition (P<0.001) (Table 1).

Table 1 Demographics of frail and non-frail patients

Predictive models and ROC analysis

Two sets of logistic regression models were developed: the first used age alone as the primary predictor, and the second used patient frailty status and age as the primary predictors. These models were used to assess the predictive capabilities of age and frailty status for nonroutine discharge, DVT and UTI. ROCs were plotted for both the logistic regression models for each outcome (Fig. 2-4). As the figures show, the logistic regression models using frailty and age as primary predictors outperformed the model using age alone. In addition, the AUC of the ROC incorporating frailty was found to be significantly higher when compared to age alone for nonroutine discharge (P=0.017), DVT (P=0.040), and UTI (P=0.040).

Figure 2 Receiver operating characteristic (ROC) plot for prediction of nonroutine discharge status. The black ROC represents the logistic model using age alone as the primary predictor, the blue ROC represents the logistic model using frailty alone as the primary predictor, and the red ROC represents the logistic model using frailty status and age as the primary predictors. A noticeable increase in predictive power occurs when frailty is jointly considered for prediction of discharge status

AUC, area under the curve

Figure 3 Receiver operating characteristic (ROC) plot for prediction of postoperative deep vein thrombosis (DVT). The black ROC represents the logistic model using age alone as the primary predictor, the blue ROC represents the logistic model using frailty alone as the primary predictor, and the red ROC represents the logistic model using frailty status and age as the primary predictors. A noticeable increase in predictive power occurs when frailty is jointly considered for prediction of DVT

AUC, area under the curve

Figure 4 Receiver operating characteristic (ROC) plot for prediction of postoperative urinary tract infection (UTI). The black ROC represents the logistic model using age alone as the primary predictor, the blue ROC represents the logistic model using frailty alone as the primary predictor, and the red ROC represents the logistic model using frailty status and age as the primary predictors. A noticeable increase in predictive power occurs when frailty is jointly considered for prediction of UTI

AUC, area under the curve

Postoperative complication rates

During the primary admission, the costs associated with the inpatient stay were significantly higher for frail patients compared to non-frail patients (Frail: $63,164.99±$102,884.00 vs. Non-Frail: $37,103.88±$31,366.27; P<0.001). Similarly, frail patients had a significantly greater inpatient LOS compared to non-frail patients, despite propensity score matching for demographics (Frail: 16.5.0±22.7 days vs. Non-Frail: 8.3±7.2 days; P<0.001).

Complication rates were found to be significantly higher in frail patients compared to non-frail patients. More specifically, frail patients had higher rates of postoperative infection (P<0.001), acute posthemorrhagic anemia (P<0.001), UTI (P=0.0053), DVT (P=0.0093), wound dehiscence (P=0.0026), mortality (P=0.0076), and readmission (P=0.0057) (Table 2). However, rates of pneumonia (P=0.25), PE (P>0.99), and ileus (P=0.58) were not significantly different between the 2 cohorts.

Table 2 Complications in frail and non-frail patients

Discussion

In this retrospective study of patients treated surgically for secondary metastatic disease of the liver in 2016 and 2017, we investigated the influence of frailty on perioperative complications. Using propensity score matching techniques, we analyzed the association between frailty and complications of interest, while controlling for demographic confounders. Further modeling allowed for the creation of several ROCs for nonroutine discharge, DVT and UTI, which demonstrated that the addition of frailty to age alone within predictive models improved the AUC significantly. This study contributes to the body of work dedicated to improving the postoperative management of patients who undergo surgical intervention for metastasis to the liver, highlighting specific complications that may predominate in frail populations.

Over the last several decades, frailty has become a topic of particular interest in hepatobiliary surgery, and has been shown to be highly correlated with rates of postoperative morbidity and mortality [18-20]. A 2018 review by Laube et al concluded that frailty may affect 17-43% of patients with advanced liver disease: frail patients who undergo hepatectomy have a higher incidence of postoperative complications, with a longer LOS and greater short- and long-term mortality [18,21-24]. In addition, 2 recent 2020 and 2021 studies by Yamada et al demonstrated that elderly frail patients undergoing surgery for hepatocellular carcinoma (HCC) had significantly worse overall and disease-free survival compared to non-frail patients [19,20]. These findings suggest that, even when controlling for age, elderly patients who meet clinical criteria for frailty continue to have worse perioperative outcomes. In other words, the decreased physiological reserve that defines frailty is poorly captured by age alone; thus, considering age together with patient frailty status may provide a superior predictor of perioperative morbidity.

Furthermore, while frailty has been well studied within the field of hepatobiliary surgery, data outlining the influence of frailty in patients with metastatic disease to the liver are still limited. A 2021 study by Tokuda et al used multivariate regression analysis to assess the role of frailty in 29 frail and 58 non-frail patients with primary colorectal cancer (CRC) metastatic to the liver [25]. Their study found that overall and disease-specific survival rates were significantly worse in frail patients, while 21 of 58 patients with disease recurrence were frail patients, representing 72.4% of the frail cohort [25]. Similar findings were also demonstrated in a 2021 study by Dauch et al, who used the modified frailty index and multivariable regression analysis to evaluate the influence of frailty on postoperative outcomes in patients with primary CRC metastatic to the liver [26]. In their study, they found that frail patients had significantly higher rates of minor/major complications, readmissions, unfavorable discharges and mortality, and a longer LOS [26]. While both of these studies contribute important information to the existing literature, our study expands upon these studies in several important ways. First, we used propensity score matching techniques, which have been shown to be more robust in estimating causal effects using observational data compared to multivariate and multivariable analyses [27]. Additionally, our study included all patients with secondary metastatic disease to the liver captured in the NRD, allowing our models to be applicable to all patients with liver metastasis, regardless of the primary origin of the cancer.

Because frailty is undeniably associated with worse outcomes, several studies have investigated interventions that may reduce frailty burden and improve frail patients’ outcomes. Frailty is associated with a reduced physiologic reserve and resistance to stressors, resulting in vulnerability to adverse outcomes [18,28]. Intuitively, approaches that increase physiological reserve could act as a means of combating frailty. In a 2021 propensity score-matched study by Tsuchihashi et al, patients were started on an exercise regimen the day after HCC resection, and received interventions in the form of physical therapy 5 days per week, ranging through stretching, resistance, balance and aerobic exercises [29]. They found that the patients who completed an in-hospital exercise regimen improved their frailty status and had lower rates of postoperative complications [29]. This study suggests that specific in-hospital interventions may prevent the development of frailty-associated illnesses following surgical intervention for liver cancer. However, they are limited by patient compliance and additional studies are necessary to demonstrate the same findings in patients with secondary metastatic disease of the liver. In addition, a broad body of literature has evaluated the influence of nutrition on frailty. Specifically, adequate energy intake, especially protein intake, has been shown to reduce rates of frailty in large population studies and systematic reviews [30,31]. Thus, a combination of adequate preoperative nutrition and postoperative physical therapy may reduce rates of patient frailty, leading to lower perioperative complication rates in patients like those in our cohort.

This study has several limitations, including those inherent in retrospective cohort analyses. Namely, the quality of analysis is dependent on the depth and accuracy of patient encounters documented in the NRD, and Berkson’s bias is present when working with inpatient databases. Furthermore, this study is limited by its retrospective nature, focusing on a narrow range of time (2016 and 2017 only). However, the choice of dates was due to the implementation of mandatory ICD-10 coding in late 2015, which allowed for more detailed codes to be drawn for analysis. Lastly, the NRD allows for retrospective readmission analysis within one calendar year (January to December). Therefore, additional readmissions not occurring within the same calendar year are not captured and cannot be analyzed using the NRD.

Our study suggests that patient frailty status strongly correlates with rates of medical complications, costs, LOS and discharge disposition in patients with secondary metastatic disease of the liver following surgical intervention. Frailty also improved the prediction of nonroutine patient discharges, DVT and UTI, compared to patient age alone, when incorporated into logistic models. Overall, frailty represents a robust predictor of patient outcomes and a better understanding of frailty may aid surgeons’ decision-making following surgical intervention for liver metastases. Further research, including multicenter analyses with a large number of participants, is necessary to fully understand the influence of frailty on outcomes in patients with secondary metastatic disease of the liver.