Abstract
Background: Peri-operative variables associated with prolonged Intensive Care Unit (ICU) admission following cytoreductive surgery for ovarian cancer were investigated. Patients and Methods: A retrospective review was carried out of patients admitted to the ICU following cytoreductive surgery for ovarian cancer in a single tertiary referral centre from 2015-2019. Patients were categorized according to length of ICU stay (<48 h and ≥48 h), and peri-operative variables were compared across the two groups. Results: A total of 56 patients were admitted to the ICU post-operatively, 37 for <48 h and 19 for ≥48 h (range=3-11 days). Greater duration of procedure and estimated blood loss, bowel resection, higher post-operative lactate level, lower post-operative albumin level and requirement for post-operative blood products were associated with prolonged ICU stay. Increased intraoperative fluid requirement was an independent predictor of extended ICU stay. Conclusion: Utilizing identified intra-operative risk factors to perform individualized risk assessments might improve planning of ICU resources. Optimizing intraoperative fluid management may improve short-term patient outcomes.
Ovarian cancer is the sixth most common female malignancy in the United Kingdom with a peak incidence in those aged 75-79 (1). Greater than 60% of patients present with advanced disease (1). Complete cytoreduction and platinum sensitivity are the two most powerful predictors of prognosis (2-4). Optimal surgical treatment of ovarian cancer involves complete resection of all visible macroscopic disease and this may require splenectomy, bowel or urinary tract resection and extensive peritonectomy, including diaphragmatic stripping or segmental resection (2-4). A significant proportion of patients have ascites at the time of surgery (2) and are often malnourished with resultant hypoproteinaemia, which leads to significant intraoperative fluid shifts (5).
Following lengthy and complex cytoreductive surgery in these high-risk patients, approximately 21-36% require post-operative admission to an Intensive Care Unit (ICU) and frail or co-morbid patients are more likely to require ICU support following ovarian cancer surgery (6, 7). The duration and complexity of ICU care is determined by individual patient requirements, including the extent and duration of surgery, ‘third-spacing’ of fluids and co-morbidity; the COVID-19 pandemic has resulted in a reduction in availability of intensive-care beds for patients with ovarian cancer (8). Currently identified predictors of prolonged length of ICU stay in gynaecological cancer include older age, increased medical co-morbidity and pre-operative hypo-albuminaemia (9). The length of ICU stay is a surrogate for both nosocomial infections and a prolonged hospital stay, whilst unplanned ICU admission is associated with poorer survival outcomes in patients with ovarian cancer (10, 11).
Identifying patients who are likely to require prolonged ICU support might facilitate shared decision-making with patients, improve interdisciplinary planning, and enable more effective resource management. The primary objective of this study was to identify modifiable peri-operative variables related to prolonged (greater than 48 h) ICU stay in patients undergoing surgery for ovarian, tubal and primary peritoneal cancer (hereafter referred to as ovarian cancer), in particular, modifiable treatment factors for which peri-operative interventions might be targeted to reduce morbidity.
Patients and Methods
A single-centre retrospective review of all patients admitted to ICU following surgery for newly diagnosed ovarian cancer was performed at the University Hospitals of Leicester, a tertiary cancer referral centre, between January 2015 and December 2019. Approval for the study was given by the University Hospitals of Leicester Clinical Audit Team (Audit number 10643-ITU). The inclusion criterion was undergoing cytoreductive surgery for ovarian cancer with post-operative admission to an ICU. Patients were identified using the computerised ICU and Theatre Operating Department databases, which recorded all inpatient admissions to ICU during the study period. Prospectively recorded data from the ICU database, Theatre Operating Department database, Pathology database and medical records were collected into a pre-defined proforma. The proforma captured pre-operative details including patient age, body mass index (BMI) and ethnicity, as well as pre-operative CA125, biochemical and haematological parameters, timings of any prior neo-adjuvant chemotherapy and whether the surgery was a primary or interval debulking procedure. Patient complexity was further characterised using the Charlson Co-morbidity Index (CCI) (12, 13). Intra-operative details including the duration of the surgical procedure, total volume of intra-operative fluids administered, estimated blood loss, transfusion status and number of blood products transfused, presence of ascites, need for bowel resection and any complications were recorded. The modified Surgical Complexity Score was delineated from the operative records (14). The type of admission to ICU (planned or unplanned) and the duration of stay as well as the level of physiological support required were recorded alongside the need for post-operative transfusion, initial post-operative biochemical and haematological parameters obtained within 24 h of surgery and any post-operative complications. Patients were categorised according to the length of ICU stay (<48 h and ≥48 h); the cut-off of 48 h was chosen as this represented the median length of ICU stay for the study population and 48 h has previously been used to define prolonged ICU stay (15).
Statistical analysis was performed using SPSS Statistics (Version 26.0; IBM Armonk, NY, USA). Group comparisons for continuous variables were made using Student’s t-test for parametric data or Mann–Whitney U-test for non-parametric data. For categorical variables, group comparisons were made using Pearson’s chi-squared or Fisher’s exact test, as appropriate. Backwards stepwise logistic regression modelling was performed to evaluate for independent predictors of prolonged stay (≥48 h). Associations are reported giving the mean value and range for parametric continuous variables or median value and range for non-parametric continuous variables and frequency for categorical data. Two-sided p-values are displayed for each variable; a p-value of less than 0.05 was considered to be statistically significant. Subgroup analysis was performed using identical statistical analysis for advanced stage III or IV ovarian cancer. Pearson’s correlation was performed to assess each variable against total length of ICU stay.
Results
Patient demographics. During the study period, 247 laparotomies were performed for newly diagnosed ovarian cancer or as a staging laparotomy for suspicious ovarian masses. Fifty-six patients (22.7%) were admitted to the ICU following surgical treatment for ovarian cancer during the study period (2015-2019); 37 (66.1%) of these patients were discharged within 48 h of admission whilst 19 (33.9%) required a stay for over 48 hours. The median age was 63 years, the median BMI was 26.5 kg/m2 and 85.7% and 14.3% of patients were of White or South Asian ethnicity, respectively. There were no significant differences in age, ethnicity, BMI, neoadjuvant chemotherapy status, mean Surgical Complexity Score or mean CCI between the two subgroups. The distribution of cases by histological subtype and stage was similar between groups, as illustrated in Table I, with the majority of cases (n=44, 80.4%) diagnosed with advanced-stage III or IV disease. Of the 12 cases diagnosed with stage I-II disease, nine cases were planned ICU admissions, three of whom had a bowel resection, with three unplanned admissions.
Patient demographics, pathological and surgical characteristics according to duration of ICU stay.
Variables predicting prolonged ICU stay. Significant intraoperative variables associated with prolonged ICU stay included greater duration of operation (mean 242 versus 313 min, p=0.020), greater estimated blood loss (600 versus 1,000 ml, p=0.014), requirement for intra-operative blood transfusion (p=0.046), bowel resection (p=0.046) and greater total volume of intraoperative fluids transfused (4000 versus 6000 ml, p=0.001) (Table II). Post-operative variables associated with prolonged ICU stay included higher immediate post-operative lactate (1.45 versus 1.85 mmol/l, p=0.018), lower post-operative albumin (30 versus 24 g/dl, p=0.018) and requirement for post-operative blood transfusion (p=0.001). The type of ICU admission and other biochemical or haematological parameters were not significantly associated with extended ICU stay and are detailed further in Table II. Logistic regression backward stepwise analysis demonstrated that only higher volumes of intra-operative fluid transfusion were independently associated with prolonged ICU stay (p=0.004). There was a significant correlation between the length of post-operative ICU stay and total post-operative length of stay (r=0.703, p<0.001).
Perioperative variables according to duration of ICU stay.
Subgroup analysis of patients with stage III or IV advanced ovarian cancer identified the following variables to be significantly associated with ICU stay ≥48 h: increased volume of intra-operative fluids transfused (4,000 versus 5,500 ml, p=0.002), higher estimated blood loss (750 versus 1,250 ml, p=0.035), lower post-operative albumin (30 versus 23 g/dl, p=0.009), higher total number of packed red cell units transfused post-operatively (0 versus 1.5, p=0.045) and higher CCI (9 versus 10, p=0.025). Backwards stepwise logistic regression analysis identified lower post-operative albumin (p=0.012) and higher CCI (p=0.023) as independent predictors of prolonged ICU stay in patients with stage III or IV disease.
Discussion
An increasing proportion of patients undergoing surgical management for ovarian cancer require post-operative ICU care. Identifying peri-operative variables which are associated with prolonged ICU stay would assist gynaecology-oncology surgeons, anaesthetists, and intensivists in predicting resource allocation, and improve patient counselling regarding their post-operative journey. In this study, we identified utilisation of higher volumes of intra-operative fluids as an independent predictor of prolonged ICU stay. Other factors associated with prolonged ICU stay include higher estimated blood loss, requirement for post-operative blood transfusion, extended duration of surgery, bowel resection, higher post-operative lactate and lower post-operative albumin levels.
Fluid management challenges in cytoreductive ovarian cancer surgery are multifactorial in nature. Surgical factors include haemorrhage, evaporative losses from prolonged surgery, sequestration of fluid into tissues and fluid shifts associated with extensive peritonectomy or ascites. Anaesthetic-related factors include hypotension secondary to drug- and dose-dependent vasodilatation, and myocardial depression or sympathetic blockade during neuraxial anaesthesia (16, 17). Laparotomy reduces intra-abdominal pressure, which further increases intraoperative intraperitoneal third-space fluid losses (18). These factors are compounded by dilutional coagulopathy induced by large volume replacement (18-20). Therefore, intraoperative fluid management requires a delicate balance between choice of anaesthetic agent, fluid replacement and vasoconstrictors to optimise end-organ perfusion whilst avoiding fluid overload and excessive vasoconstriction.
Intra-operative fluid management is an important modifiable factor which can alter perioperative patient outcomes. This study revealed a median transfused volume of 6,000 ml as being associated with prolonged ICU stay and confirms the previous finding by Diaz Montes et al. that higher volumes of crystalloid transfusion (>5,000 ml) in ovarian cancer surgery are associated with prolonged ICU stay (15). Current Enhanced Recovery After Surgery (ERAS) guidelines in gynaecological malignancy recommend goal-directed fluid management, facilitated by minimally invasive haemodynamic monitoring to reduce intraoperative fluid requirements (21). Goal-directed haemodynamic management has also been shown to increase early return to feeding, shorten bowel recovery and reduce the length of hospital stay (22, 23).
Fresh frozen plasma (FFP) has high concentrations of albumin and pro-coagulant factors which help to preserve oncotic pressure and avoid dilutional coagulopathy associated with high-volume replacement (24). In patients with ovarian cancer with large-volume malignant ascites, FFP use has been shown to increase stroke volume and reduce inotrope requirements, promoting haemodynamic stability (24). In patients with peritoneal carcinomatosis, FFP transfusion and restrictive fluid management reduced requirements for transfusion of other blood products (24). Routine FFP use is not widely established in current ERAS protocols, however, prophylactic FFP transfusion alongside tranexamic acid is now given at induction in our centre to reduce circulatory losses and need for blood product as part of our ERAS protocol in women with significant malignant ascites, the outcomes of which will be investigated in future to determine the impact of this change (21, 25).
Our study revealed a higher median blood loss and higher blood product requirement in those staying in ICU for ≥48 h. A multicentre randomised controlled trial demonstrated reduced overall blood loss and peri-operative transfusion requirements following the pre-operative administration of a single bolus of tranexamic acid in ovarian cancer surgery (26). A UK prospective cohort study identified that upfront tranexamic acid and early cryoprecipitate transfusion reduced red cell requirements compared with FFP alone during cytoreductive surgery with hyperthermic intraperitoneal chemotherapy for peritoneal malignancy and this might represent an alternate strategy (27). These measures are easily accessible, cost-effective interventions and implementation of such strategies could have wide ranging benefits to patients and service providers.
Frailty is increasingly recognised as an important predictor of adverse outcomes following ovarian cancer surgery (28, 29). Yao et al. reported 48.9% of patients with a frailty index ≥0.15 were admitted to ICU post-operatively following ovarian cancer surgery compared to 20.5% of non-frail patients, with frailty acting as an independent predictor of ≥30-day ICU stays (6). Prolonged ICU stay is frequently associated with significant muscle wastage secondary to a reduction in functional use, with approximately 10% of body muscle lost within 3 days of immobility in elderly patients (30), and patients often require high levels of physiotherapy input post-discharge. Sarcopenia, or skeletal muscle wastage has been shown to worsen cancer survival outcomes (31), therefore, avoiding a prolonged ICU stay should be a priority for clinicians treating frail or co-morbid patients with ovarian cancer.
The CCI is an independent predictor of survival and perioperative outcomes in malignancy, including post-debulking surgery for ovarian cancer (13, 32, 33). Importantly, patients with ovarian cancer with a high age-adjusted CCI are reported to receive fewer adjuvant chemotherapy cycles, or be less likely to receive adjuvant chemotherapy at all, with 12.1% of patients with a high CCI score not receiving adjuvant chemotherapy following neoadjuvant chemotherapy and interval debulking surgery in a single-centre study by Phillips et al. (34). Our study demonstrated a higher CCI score to be associated with prolonged ICU stay in patients with advanced-stage disease, which has significant implications for post-operative recovery and therefore consideration of ‘fitness’ to undergo adjuvant chemotherapy.
Pre-operative optimisation of co-morbidities in frail or co-morbid patients may help to reduce prolonged ICU stays.
Limitations of this study include the risk of selection bias due to the retrospective nature of data collection and the overall small sample size from a single centre. The findings of this study warrant multicentre evaluation of modifiable factors and post-operative outcomes.
In conclusion, knowledge of the factors associated with adverse post-operative outcomes and prolonged ICU stay following ovarian cancer surgery is important when designing services and patient pathways to reduce the risk of adverse peri-operative events and prioritise long-term recovery and fitness for adjuvant chemotherapy. Identifying modifiable factors such as intra-operative fluid management should improve uptake of targeted interventions such as goal-directed haemodynamic therapy, FFP use and enhanced recovery principles into everyday clinical practice.
Footnotes
Authors’ Contributions
Study concepts and design: AC and RB; data collection: SS, JH, MC and AC; literature research: AC and RB; statistical analysis: AC and FR; article preparation: AC, RB and EM; article editing: AC, SS, JH, MC, FR, QD, TB, SC, EM and RB.
Conflicts of Interest
The Authors declare no conflicts of interest.
- Received April 22, 2021.
- Revision received May 8, 2021.
- Accepted May 18, 2021.
- Copyright © 2021 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.
