Pre-operative nomogram to predict risk of peri-operative mortality following liver resections for malignancy

Utilizing risk-stratified pathways to personalize post-hepatectomy discharge planning: A contemporary analysis of 1,354 patients

BACKGROUND

While risk-stratified post-hepatectomy pathways (RSPHPs) reduce length-of-stay, can they stratify hepatectomy patients by risk of early postoperative events.

METHODS

90-day outcomes from consecutive hepatectomies were analyzed (1/1/2017-12/31/2021). Pre/post-pathway analysis was performed for pathways: minimally invasive surgery ("MIS"); non-anatomic resection/left hepatectomy ("low-intermediate risk"); right/extended hepatectomy ("high-risk"); "Combination" operations. Time-to-event (TTE) analyses for readmission and interventional radiology procedures (IRPs) was performed.

RESULTS

1354 patients were included: MIS/n= ​119 (9 ​%); low-intermediate risk/n= ​443 (33 ​%); high-risk/n= ​328 (24 ​%); Combination/n= ​464 (34 ​%). There was no difference in readmission (pre: 13 ​% vs. post:11.5 ​%, p ​= ​0.398). There were fewer readmissions in post-pathway patients amongst MIS, low-intermediate risk, and Combination patients (all p ​> ​0.1). 114 (8.4 ​%) patients required IRPs. Time-to-readmission and time-to-IR-procedure plots demonstrated lower plateaus and flatter slopes for MIS/low-intermediate-risk pathways post-pathway implementation (p ​< ​0.001).

CONCLUSION

RSPHPs can reliably stratify patients by risks of readmission or need for an IR procedure by predicting the most frequent period for these events.

Predictive nomograms for postoperative 90-day morbidity and mortality in patients undergoing liver resection for various hepatobiliary diseases

BACKGROUND

Postoperative complications affect the long-term survival and quality of life in patients undergoing liver resection. No model has yet been validated to predict 90-day severe morbidity and mortality.

METHODS

The prospective recruitment of patients undergoing liver resection for various indications was performed. Preoperative clinical and laboratory data, including liver stiffness, indocyanine green retention, and intraoperative parameters, were analyzed to develop predictive nomograms for postoperative severe morbidity and mortality. Calibration plots were used to perform external validation.

RESULTS

The most common indications in 418 liver resections performed were colorectal metastases (N = 149 [35.6%]), hepatocellular carcinoma (N = 106 [25.4%]), and benign liver tumors (N = 60 [14.3%]). Major liver resections were performed in 164 (39.2%) patients. Severe morbidity and mortality were observed in 87 (20.8%) and 9 (2.2%) of patients, respectively, during the 90-day postoperative period. Post-hepatectomy liver failure was observed in 19 (4.5%) patients, resulting in the death of 4. The independent predictors of 90-day severe morbidity were age (odds ratio:1.02, P = .06), liver stiffness (odds ratio: 1.23, P = .04], number of resected segments (odds ratio: 1.28, P = .004), and operative time (odds ratio: 1.01, P = .01). Independent predictors of 90-day mortality were diabetes mellitus (odds ratio: 6.6, P = .04), tumor size >50 mm (odds ratio:4.8, P = .08), liver stiffness ≥22 kPa (odds ratio:7.0, P = .04), and operative time ≥6 hours (odds ratio: 6.1, P = .05). Nomograms were developed using these independent predictors and validated by testing the Goodness of fit in calibration plots (P = .64 for severe morbidity; P = .8 for mortality).

CONCLUSION

Proposed nomograms would enable a personalized approach to identifying patients at risk of complications and adapting surgical treatment according to their clinical profile and the center's expertise.

Preoperative nomogram to predict posthepatectomy liver failure

BACKGROUND AND OBJECTIVES

Posthepatectomy liver failure (PHLF) is associated with significant morbidity and mortality. However, it is often difficult to predict the risk of PHLF in an individual patient. We aimed to develop a preoperative nomogram to predict PHLF and allow better risk stratification before surgery.

METHODS

Data for patients undergoing a partial or major hepatectomy were extracted from the hepatectomy-specific NSQIP database for years 2014-2016. Data set from 2017 was used for validation. Patients with Grade B/C liver failure were compared with patients with no liver failure.

RESULTS

A total of 10 808 patients from 2014-2016 data set were included. Of these, 316 patients (2.9%) developed Grade B/C PHLF. In the multivariable model consisting of preoperative variables, the following were predictive of Grade B/C PHLF (all p < 0.05): male gender, biliary stent, neoadjuvant therapy, viral hepatitis B or C, concurrent resections, biliary reconstruction, low sodium, and low albumin (model c statistic-0.78). This model was used to construct a nomogram. In the 2017 validation cohort of 4367 patients the nomogram again demonstrated good c-statistic (0.78).

CONCLUSIONS

Our nomogram provides patient-specific probabilities for PHLF, and is easy to use. This is a valuable tool that can be utilized for preoperative patient counseling and selection.

Predicting adverse events in patients undergoing hepatectomy-validation of preoperative nomogram and risk score

BACKGROUND/PURPOSE

Much research exists on preoperative measures of postoperative mortality in the surgical treatment of liver malignancies, but little on morbidity, a more common outcome. This study aims (i) to validate the published calculations as acceptable measures of postoperative mortality and (ii) to assess the value of these published measures in predicting postoperative morbidity.

METHODS

Data were collected from a prospectively managed dataset of 1059 hepatectomies performed in Louisville, Kentucky from December 1990 to April 2014. Preoperative data were used to assign scores for each of two published measures and the scores were sorted into clinically relevant groups with corresponding ordinal scores, according to the previously published literature (Dhir nomogram and Simons risk score).

RESULTS

After selection, 851 hepatectomies were analyzed. Both the Dhir nomogram (p = 0.0004) and Simons risk score (p = 0.0017) were acceptable predictors of postoperative mortality. In the analysis of morbidity, Dhir scores were a poor predictor of morbidity. The Simons ordinal risk score was predictive of complications (p = 0.0029), the number of complications (p = 0.0028), complication grade (p = 0.0033), and hepatic-specific complications (p = 0.0003).

CONCLUSION

The Simons ordinal risk score can be useful in assessing postoperative morbidity among hepatectomy patients.

Mortality in hepatectomy: Model for End-Stage Liver Disease as a predictor of death using the National Surgical Quality Improvement Program database

BACKGROUND

The predictive value of the Model for End-stage Liver Disease (MELD) for mortality after hepatectomy is unclear. This study aimed to evaluate whether MELD score predicts death after hepatectomy and to identify the most useful score type for predicting mortality. We hypothesized that an increase in this score is correlated with 30-day mortality in patients undergoing hepatic resection.

METHODS

The American College of Surgeons National Surgical Quality Improvement Program database was queried for hepatectomy. Original MELD, United Network of Organ Sharing-modified MELD (uMELD), integrated MELD (i-MELD), and sodium-corrected MELD (MELD-Na) scores were calculated. Mortality was analyzed by multivariate logistic regression. MELD types were compared using receiver operating characteristic (ROC) curves.

RESULTS

From 2005 to 2011, 11,933 hepatic resections were performed, including 7,519 partial, 2,104 right, and 1,210 left resections, and 1,100 trisectionectomies. The mean duration of stay was 8.4 ± 22.0 days, and there were 275 deaths (2.4%). The 30-day mortality rates were 1.8%, 6.9%, 15.4%, and 25% according to uMELD strata of 0-9, 10-19, 20-29, and ≥ 30, respectively. Multivariate analysis revealed that increasing MELD stratum was independently associated with higher mortality (P < .001) for all MELD types. The uMELD had the largest effect size (odds ratio [OR], 1.16; 95% CI, 1.10-1.20), whereas i-MELD had the narrowest CI (OR, 1.13; 95% CI, 1.10-1.17) and largest area under the ROC curve.

CONCLUSION

The postoperative 30-day mortality after hepatectomy increases with increasing MELD score across all MELD types. There is a 16% increase in the odds of mortality for each point increase in uMELD.

Safety of hepatectomy for elderly patients with hepatocellular carcinoma

The number of elderly patients with hepatocellular carcinoma (HCC) has been increasing. Characteristics of elderly HCC patients are a higher proportion of females, a lower rate of positive hepatitis B surface antigen, and a higher rate of positive hepatitis C antibodies. Careful patient selection is vital for performing hepatectomy safely in elderly HCC patients. Treatment strategy should be decided by not only considering tumor stage and hepatic functional reserve, but also physiological status, including comorbid disease. Various assessment tools have been applied to predict the risk of hepatectomy. The reported mortality and morbidity rates after hepatectomy in elderly HCC patients ranged from 0% to 42.9% and from 9% to 51%, respectively. Overall survival rate after hepatectomy in elderly HCC patients at 5 years ranged from 26% to 75.9%. Both short-term and long-term results after hepatectomy for strictly selected elderly HCC patients are almost the same as those for younger patients. However, considering physiological characteristics and the high prevalence of comorbid disease in elderly patients, it is important to assess patients more meticulously and to select them strictly if scheduled to undergo major hepatectomy.

Risk assessment in patients undergoing liver resection

BACKGROUND

Liver resection is still a risky procedure with high morbidity and mortality. It is significant to predict the morbidity and mortality with some models after liver resection.

DATA SOURCES

The MEDLINE/PubMed, Web of Science, Google Scholar, and Cochrane Library databases were searched using the terms "hepatectomy" and "risk assessment" for relevant studies before August 2012. Papers published in English were included.

RESULTS

Thirty-four original papers were included finally. Some models, such as MELD, APACHE II, E-PASS, or POSSUM, widely used in other populations, are useful to predict the morbidity and mortality after liver resection. Some special models for liver resection are used to predict outcomes after liver resection, such as mortality, liver dysfunction, transfusion, or acute renal failure. However, there is no good scoring system to predict or classify surgical complications because of shortage of internal or external validation.

CONCLUSION

It is important to validate the models for the major complications after liver resection with further internal or external databases.

Early post-operative prediction of morbidity and mortality after a major liver resection for colorectal metastases

BACKGROUND

An early prediction of poor outcomes is essential in the management of patients after a liver resection. The aim of this study was to evaluate the role of selected biochemical parameters on post-operative day 1 (POD 1) in the prediction of morbidity and mortality after a liver resection for colorectal metastases.

METHOD

This retrospective study was based on 236 major liver resections for colorectal metastases performed between 2006 and 2011. Results of biochemical tests of blood samples obtained on POD 1 were assessed as predictors of primary outcome measures (hepatic and overall morbidity, 90-day mortality) using multiple regression and receiver-operating characteristics (ROC).

RESULTS

Hepatic morbidity, overall morbidity and 90-day mortality rates were 18.6%, 28.0% and 4.7%, respectively. On the basis of multiple regression analysis and comparisons of the prediction models, serum bilirubin was selected for the prediction of hepatic (>2.05 mg/dl, sensitivity 69.2%, specificity 71.2%) and overall (>2.05 mg/dl, sensitivity 61.1% and specificity 71.2%) morbidity, and aspartate aminotransferase (AST) was selected for the prediction of 90-day mortality (>798 U/l, sensitivity 62.5% and specificity 90.4%).

DISCUSSION

Biochemical analyses of blood on POD1 enables stratification of patients into low- and high-risk groups for negative outcomes, with serum bilirubin associated with overall and hepatic morbidity and AST associated with mortality.

Incidence and risks of liver resection: an all-inclusive French nationwide study

OBJECTIVE

To evaluate at a national level the incidence of liver resection, postoperative mortality, and variables that predict this outcome.

BACKGROUND

Data on indications of and results of liver resection are mainly derived from high-volume centers. Nationwide data are lacking.

METHODS

French health care databases were screened to identify all patients who had undergone elective hepatectomy between 2007 and 2010. The patients' age, address, associated conditions, indication and extent of hepatic (or extrahepatic) surgery and the hospital type, location, and hepatectomy caseload were retrieved. Logistic regression was used to measure the influence of these parameters on in-hospital and 90-day mortality rates. The model, created using patients operated on in 2007 and 2009, was tested in those operated on in 2008 and 2010.

RESULTS

Overall, 28,708 hepatectomies were performed. The annual incidence (13.2 per 10 adult inhabitants) varied between regions, but the extremal quotient was limited to 2.2 because 15% of the operations took place outside the patients' home region. Hospitals performed a median of 4 resections per year but 53% of all resections were performed in institutions with a volume of more than 50 per year. Treatment for primary tumors and major resections correlated with hepatectomy caseload. In-hospital and 90-day mortality were 3.4% and 5.8%, respectively. The area under the receiver operating characteristic curve of the prognostic model was 0.78/0.77 in the training and validation sample.

CONCLUSIONS

There were significant disparities in practice. In-hospital mortality underestimated true, postoperative mortality by more than 50%. The model created may be useful for more efficient regionalization of care and patient counseling.

A preoperative nomogram to predict the risk of perioperative mortality following gastric resections for malignancy

INTRODUCTION

Surgery remains one of the major treatment options available to patients with gastric cancer. The aim of this study was to develop a preoperative nomogram based on the presence of comorbidities to predict the risk of perioperative mortality following gastric resections for malignancy.

METHODS

The Nationwide Inpatient Sample (NIS) database was used to create a nomogram using SAS software. The training set (years 1993, 1996-97, 1999-2000, 2002, 2004-05) was used to develop the model which was further validated using the validation set (years 1994-95, 1998, 2001, and 2003).

RESULTS

A total of 14,235 and 9,404 patients were included in the training and validation sets, respectively, with overall actual observed perioperative mortality rates of 5.9 % and 6.6 %, respectively. The decile-based calibration plots for the training and validation sets revealed a good agreement between the observed and nomogram-predicted probabilities. The accuracy of the nomogram was further reinforced by a concordance index of 0.75 (95 % confidence interval 0.73 to 0.77) which was calculated using the validation set.

CONCLUSION

This preoperative nomogram may accurately predict the risk of perioperative mortality following gastric resections for malignancy and may be used as an adjunctive clinical tool in the preoperative counseling of these patients.

External validation of a pre-operative nomogram predicting peri-operative mortality risk after liver resections for malignancy

AIM

A pre-operative nomogram using a population-based database to predict peri-operative mortality risk after liver resections for malignancy has recently been developed. The aim of the present study was to perform an external validation of the nomogram using data from a high volume institution.

METHODS

The National Inpatient Sample (NIS) database (2000-2004) was used initially to construct the nomogram. The dataset for external validation was obtained from a high volume centre specializing in hepatobiliary surgery. Validation was performed using calibration plots and concordance index.

RESULTS

A total of 794 patients who underwent liver resection from the years 2000-2010 at the external institute were included in the validation set with an observed mortality rate of 1.6%. The mean total points for this sample of patients was 124.9 [standard error (SE) 1.8, range 0-383] which translates to a nomogram predicted mortality rate of 1.5%, similar to the actual observed overall mortality rate. The nomogram concordance index was 0.65 [95% confidence interval (CI) 0.46-0.82] and calibration plots stratified by quartiles revealed good agreement between the predicted and observed mortality rates.

CONCLUSIONS

The present study provides an external validation of the pre-operative nomogram to predict the risk of peri-operative mortality after liver resection for malignancy.