Download PDF

Anca Dragan1, Ioanel Sinescu2,3

1 „Prof. Dr. C. C. Iliescu” Emergency Institute for Cardiovascular Diseases, Bucharest, Romania
2 Department of Urology, Fundeni Clinical Institute, Bucharest, Romania
3 „Carol Davila” University of Medicine and Pharmacy, Bucharest, Romania

Abstract: Objectives – Radical nephrectomy with cavo-atrial thrombectomy is a combination of abdominal and cardiac surgery in a patient with oncological problems. This work aims to identify which of the most commonly used risk scores in cardiac (EuroSCORE, Logistic EuroSCORE, EuroSCORE II) and non-cardiac (Lee Revised Cardiac Risk Index, National Sur-gical Quality Improvement Program-NSQIP) surgery better predicted the in-hospital mortality. Methods – The retrospec-tive study included patients diagnosed with kidney tumour with venous extension at the supra-diaphragmatic inferior vena cava level and/or at the atrial level, undergoing complex surgery, radical nephrectomy with extraction of the venous throm-bus using extracorporeal circulation, performed at Prof C.C. Iliescu Cardiovascular Diseases Emergency Institute between 2004 and 2018 by mixed teams composed by urologists, cardiac surgeons, anaesthetists, and perfusionists. The risk scores (EuroSCORE, Logistic EuroSCORE, EuroSCORE II, Lee Revised Cardiac Risk Index, NSQIP) were calculated according to the clinical records. Comparison between scores was performed by calculating the area under the ROC curve (AUC). The retrospective study received the approval of the ethical committee of the Institute. Results – There were included in the study 30 consecutive patients. There were recorded 6 (20%) in-hospital. The additive EuroSCORE and logistic EuroSCORE showed a high predictive power in the case of in-hospital death (AUC = 0.816 and AUC = 0.830, respectively). EuroSCORE II, NSQIP-mortality and Lee Revised Cardiac Risk Index failed to predict in-hospital mortality. Conclusions – Additive EuroS-CORE and logistic EuroSCORE were the risk scores that better predicted in-hospital mortality in our retrospective study. Keywords: complex surgery, extracorporeal circulation, mortality, risk scores.

INTRODUCTION
Among the solid neoplasms in adult people, renal car-cinomas represent 5% of neoplasms in men and 3% of neoplasms in women1. Albeit it seems a rather small percentage, it should be noted that they only become symptomatic at an advanced stage of the disease, often being unnoticed, so that at the time of diagnosis they are often in the metastasis stage2. In addition, a feature of renal carcinomas is to tumour extension at the level of the renal vein, up to the level of the inferior vena cava and even at the atrial level. The only therapeutic option in these cases is the surgical removal of the kid-ney tumour and its venous extension, with or without extracorporeal circulation. However, patients with venous extensions have a better prognosis than those diagnosed with metastases3. 60% of them survive for 5 years after the surgery4. Risk stratification in these patients is very diffi cult to perform, as we are dealing with a combination between abdominal surgery and cardiac surgery in a patient with oncological problems. This work aims to identify which of the most com-monly used scores in cardiac and non-cardiac surgery better predicted in-hospital mortality in case of these complex surgeries.
We used the most commonly used risk scores in cardiac surgery: EuroSCORE5, Logistic EuroSCORE5, EuroSCORE II5. From the risk scores used in non-cardiac surgery we chose Lee Revised Cardiac Risk Index6 and National Surgical Quality Improvement Program-NSQIP7.

PATIENTS AND METHODS
The retrospective study included all consecutive pa-tients diagnosed with kidney tumour with venous extension at the supra-diaphragmatic inferior vena cava and/or at the atrial level, undergoing complex surgery, radical nephrectomy with extraction of the venous thrombus under extracorporeal circulation. Surgical interventions were performed at Prof C.C. Iliescu Emergency Institute for Cardiovascular Di-seases between 2004 and 2018 by a mixed medical team composed by urologists, cardiac surgeons, ana-esthetists, and specialists in PCI. Because our study is a retrospective one, we obtained the approval of the ethical committee of the institute. The medical data was collected from the medical records.
EuroSCORE5, Logistic EuroSCORE5, EuroSCORE II5, Lee Revised Cardiac Risk Index6 and NSQIP7 have been calculated based on the medical records. For the NSQIP we used the prediction of mortality. In this study, in-hospital mortality refers to intraoperative and postoperative death in the same hospitalization.

STATISTICAL ANALYSIS
The statistical analysis was conducted by using the IBM SPSS Statistics 20 Program. Nominal (qualitative), or-dinal and quantitative variables were used. The quan-titative variables were presented as average, standard deviation, minimum and maximum values. After tes-ting the normal distribution of data by using the Sha-piro Wilk test, the T test or the Mann Whitney test were independently applied, as appropriate. Kendall’s tau test was used to identify and quantify the statistical correlation between the variables. The ROC curves were generated as well for each risk scores versus the endpoint in- hospital death, with the calculation of the area under the curve, testing also its statistical signi-ficance. We identified a cut-off of them, targeting as much sensitivity as possible. The area under the ROC curve was the main element of risk score comparison within the risk scores for the endpoint. The threshold of statistical significance is 95% (p<0.05).

Figure 1. Distribution of in-hospital nonsurvivors and survivors for each EuroSCORE value.

RESULTS
There were identified 30 consecutive patients that were included in our study. Their preoperative cha-racteristics and intraoperative data are presented in Table 1 and Table 2. There were 2 (6.67%) intraope-rative deaths: one due to intraoperative bleeding and one due to major pulmonary embolism. There were other 4 (13.33%) postoperative deaths: one patient with intestinal infarction, two patients with multiple organ dysfunction and one patient with major bleeding.
Patients included in the study presented an additive EuroSCORE of 3.87 +/- 1.83 (range 2-7). The additi-ve EuroSCORE was higher in nonsurvivors (Table 3). Figure 1 shows that in the case of the patients with higher EuroSCORE in-hospital mortality predomi-nantly occurred. Table 3 presents the results of sta-tistic tests. Generation of the ROC curve (Figure 2) of additive EuroSCORE for the in-hospital mortality endpoint identifies an area of 0.816 under the cur-ve, with statistical significance (p=0.018). The cut-off of EuroSCORE of 4.5 corresponds to a sensibility of 83.33% and a specificity of 70.8%.
In non-survivors patients, Logistic EuroSCORE was higher (5.38+/-2.17) than in surviving patients (2.68+/-1.41). Generation of the ROC curve of Logistic EuroSCORE II (Figure 4)versus the endpoint: in-hospital mortality reveals an area of 0.830 under the curve with statistical significance p=0.014. The cut-off of 4.54 corresponds to a sensitivity of 83.33% and a specificity of 91.7%.
EuroSCORE II was higher in non-survivors patients (2.12+/- 1.04) than in surviving ones (1.25+/-0.8), but this is not statistically significant (Mann Whitney Test). Table 3 presents the results of statistic tests. Generation of the ROC curve of EuroSCORE II (Figure 4) versus the endpoint: in-hospital death reveals a 0.736 area under the curve with no statistical significance (p=0.078).
In the case of nonsurvivors patients, NSQIP-mor-tality (7.08+/-3.01) had higher values than in survivors (5.58+/-0.85). Table 3 presents the results of statistic tests. Generation of the ROC curve of NSQIP-mor-tality values versus the endpoint (Figure 5): the occur-rence of in-hospital death shows an area of 0.740 un-der the curve, but with no statistical signifi cance (p = 0.074).
The Lee Revised Cardiac Risk Index (range 1-3) had a ROC curve (Figure 6) of the score versus the en-dpoint: in-hospital mortality shows a 0.587 area under the curve, with no statistical significance (p = 0.527).

Figure 2. ROC of additive EuroSCORE versus in-hospital mortality.

Figure 3. ROC of Logistic EuroSCORE versus in-hospital mortality.

Figure 4. ROC of EuroSCORE II versus in-hospital mortality

Figure 5. ROC of NSQIP-mortality versus in-hospital mortality versus in-hospital mortality.

Figure 6. ROC Lee Revised Cardiac Risk Index versus in-hospital morta-lity.

DISCUSSIONS
This study shows that only EuroSCORE and Logistic EuroSCORE predicted in-hospital mortality. Logistic EuroSCORE, with the area under the ROC curve of 0.830, with statistical significance, best predicted the occurrence of in-hospital death. The additive EuroS-CORE ranks second with a AUC of 0.816, also with statistical signifi cance. The association between the Logistic EuroSCORE and additive EuroSCORE and in-hospital mortality is of medium intensity.
EuroSCORE II had no significant statistical correlati-on to the occurrence of in-hospital deaths. Even if the area under the ROC curve had a fairly good value, it is not significant. The correlation Kendall tau statistical method sustains this idea. Thus, EuroSCORE II could not accurately predict the occurrence of in-hospital death. Albeit NSQIP-mortality had different distribu-tions in the two patients groups (Mann Whitney test), it was not correlated with the occurrence of in-hos-pital deaths (Kendall’s tau test). In addition, its ROC curve shows an area under the curve of 0.740 which is not statistically signifi cant. Lee Revised Cardiac Risk Index showed the lowest AUC of the ROC curve of the values versus the endpoint, with no statistical sig-nifi cance. In addition, it was not correlated with the occurrence of this event.
Based on these results, we recommend to esti-mate the risk of in-hospital mortality in the case of radical nephrectomy with cavo-atrial thrombectomy using Logistic EuroSCORE and especially additive Eu-roSCORE, a risk score easy to perform without com-putational tools, despite the medical literature which favours EuroSCORE II. A value of the additive score of 5 must draw attention and taken in consideration, because its cut-off in predicting in-hospital mortality was 4.5 and the additive score is a natural number.
Additive EuroSCORE, has been proven to be a pre-dictive tool for both mortality and morbidity after car-diac surgery since 19998,9. Logistic EuroSCORE also starts from additive EuroSCORE, which is transfor-med according to the logistic regression equation. In 2003, Michel P et al. concluded that the logistic model is a better risk predictor, especially for high-risk pati-ents, although it is more difficult to be performed in practice, requiring computational tools10. EuroSCORE II, first introduced in October 2011, seems to redu-ce risk overestimation by the first variant of the sco-re11. The review conducted in 2012 by Siregar et al. analyses the performance of the additive EuroSCORE and Logistic EuroSCORE. Its conclusions were that, albeit the additive EuroSCORE overestimates morta-lity, however, in the case of high-risk patients their mortality is underestimated by this risk score12. Niv Ad et al. make a comparison in 2016 between the Eu-roSCORE additive and EuroSCORE II, demonstrating that the latter provides a better prediction of morta-lity than the first variant of the score, which overesti-mates this risk. This is why recently EuroSCORE II is proposed as a risk score in complex cardiac surgery13. Our study does not support this idea.
NSQIP is a decision-making support based on mul-ti-institutional experience used to estimate the risk of most surgical interventions within the 30-day posto-perative period14. Although urology has been among the surgical specialities introduced in this system ever since it was founded, some authors demonstrated the fact that there are some inadequacies in this field. In 2014, Frazier et al. drew attention to poor data re-garding urology15. Some studies sustain the idea that NSQIP poorly predicted and discriminated among pa-tients in urologic surgery, underestimating the occur-rence of postoperative complications16-18. In 2018, Blair et al. made a comparison between the predicted results and those observed 30 days after surgery in patients with partial nephrectomy for renal carcinoma, concluding that NSQIP underestimates postoperative complications19 in line with our results.
It is known that NSQIP is based on 20 preopera-tive predictors, also taking into account the type of surgery7. The Surgical Operation Code was 50230. We selected a higher surgical risk in NSQIP becau-se of the cardiac time which was not included in any surgical operation code. In our study, NSQIP did not predict the endpoint. Thus, we do not recommend NSQIP-mortality as a tool in risk assessment of in-hospital mortality in the surgery described, although it is recommended by the guidelines of cardiovascular assessment in non-cardiac surgery20 and the guidelines of the American Urology Association . These recommend NSQIP as a good starting point in assessing preopera-tive risk in urology, but it is not recommended to be used alone, but together with other unspecified clini-cal information21.
The Lee Revised Cardiac Risk Index makes a pre-diction of major cardiac events in adults undergoing non-cardiac surgery6. It takes into account elements of patient history, but also the magnitude of the surgery conducted. The score classifies the patient into a risk class, depending on which the risk of death, myocardi-al infarction, or cardiac arrest is assessed 30 days after surgery. However, in our study Lee Revised Cardiac Risk Index poorly predicted in-hospital death. Thus, we do not recommend to use this risk score in the surgery described.
In the international medical literature, radical ne-phrectomy with thrombectomy has been shown to be associated with major perioperative morbidity (up to 70%) and mortality (3-16%)22. The review of Gau-dani et al. in 2016 finds surgical risk to be substantial in these cases, with in-hospital mortality up to 40%23. In our study there were recorded 6 (20%) in-hospital deaths. This is the reason why it is important to have a risk score to predict in-hospital mortality. Our study showed that additive EuroSCORE, and especially its logistic version predicts the endpoint, while EuroS-CORE II, NSQIP and Lee Revised Cardiac Risk Index failed.
To our knowledge, none of this risk scores have been used before to predict mortality in the surgery described in this study, both in international and Ro-manian medical literature.
This study has his limits: it is a retrospective one, performed in a single centre, with a small number of patients due to the low incidence of the diagnostic. The results have to be tested in a large prospective multicentre study.

CONCLUSION
In conclusion, additive and logistic EuroSCORE were the only scores that predicted in-hospital mortality, while EuroSCORE II, NSQIP and Lee Revised Cardiac Risk Index failed to predict this endpoint in this study.

Conflict of interest: none declared.

References
1. Siegel RL, Miller KD, Jemal A.Cancer statistics.2016. Cancer J Clin 2016; 66(1):7-30
2. Escudier B, Porta C, Schmidinger M, Rioux-Leclercq N, Bex A, Khoo V, Grunwald V, Gillessen S Horwich A. Renal cell carcinoma: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Annals of Oncology 2019;30:706-720
3. Kwon T, Lee JL, You D, et al. Impact of surgery on the prognosis of metastatic renal cell carcinoma with IVC thrombus received TKI therapy. J Surg Oncol 2014;110(2):145-150
4. Blutte ML, Leibovich BC, Lohse CM, Cheville JC, Zincke H. The Mayo Clinic experience with surgical management, complications and outcomes for patients with renal cell carcinoma and venous tu-mor thrombus. BJU int 2004;94:33-41
5. http://www.euroscore.org/calc.html
6. Revised Cardiac Risk Index for Pre-Operative Risk. 2018. MD+CALC website. https://www.mdcalc.com/revised-cardiac-risk-index-pre-operative-risk.
7. http://www.riskcalculator.facs.org/RiskCalculator
8. Nashed SA, Roques F, Michel P, Gauducheau E, Lemeshow S, Sal-amon R. European system for cardiac operative risk evaluation (Eu-roSCORE). Eur J Cardioyhorac Surg 1999;16(1):9-13.
9. Guimarães Andrade IN, de Moraes Neto FR, Guimarães Andrade T. Use of Euroscore as a predictor of morbidity after cardiac surgery. Rev Bras Cir Cardiovasc. 2014;29(2):292.
10. Michel P, Roques F, Nashef SA, EuroSCORE Project Group. Logistic or additive EuroSCORE for high-risk patients? Eur J Cardiothorac Surg 2003;23(5):684-7
11. Noyez L, Kievit PC, van Swieten HA, de Boer MJ. Cardiac operative risk evaluation: The EuroSCORE II, does it make a real difference? Neth Heart J. 2012;20(12):494-498
12. Siregar S, Groenwold RH, de Heer F, Bots ML, van der Graaf Y, van Herwerden LA. Performance of the original EuroSCORE. Eur J Car-diothoracic Surg 2012; 41(4): 746-54
13. Ad N, Holmes SD, Patel J, Pritchard G, Shuman DJ, Halpin L. Com-parison of EuroSCORE II, Original EuroSCORE, and The Society of Thoracic Surgeons Risk Score in Cardiac Surgery Patients. Ann Tho-rac Surg 2016; 102 (2): 573-9
14. Bilimoria K, Liu Y, Cohen M. Development and evaluation of the universal ACS NSQIP surgical Risk Calculator: a decision aide and informed consent tool for patients and surgeons. Journal of Ameri-can College of Surgeons 2013; 217(5): 833-842
15. Frazier II HA. NSQIP and urology outcomes. Can JUrol 2014; 21(6): 7547
16. Tomaszewski J, Handorf E, Kutikov A, et al. Evaluation of ACS NSQ-IP surgical risk calculator in patients undergoing radical cystectomy. Journal of Urology 2014; 191:e25-e26.
17. Winoker J, Paulucci D, et al. Journal of Urology 2017; 198 (4): 809
18. Walker JB, Gogoj A et al. Accuracy of the NSQIP risk calculator for predicting complications following adrenalectomy. IntUrol Nephrol 2019;51:1291
19. Blai BM, Lehman EB, Jafri SM, Kaag MG, Raman JD. Predicted versus observed 30-day perioperative outcomes using the ACS NSQIP sur-gical risk calculator in patients undergoing partial nephrectomy for renal cell carcinoma. Int Urol Nephrol 2018; 50 (7): 1249-1256
20. Kristensen SD, Knuuti J et al. 2014 ESC/ASA Guidelines on non-car-diac surgery: cardiovascular assessment and management. European Heart Journal 2014; 35, 2383-2431
21. Stoffel J, Montgomery J, Suskind A, Tucci C, Vanni A. Optimizing outcomes in urologic surgery: preoperative care for the patient undergoing urologic surgery or procedure. American Urological As-sociation website. https://www.auanet.org/guidelines/optimizing-outcomes-in-urologic- surgery-preoperative-care-for-the-patient-undergoing-urologic-surgery-or-procedure
22. Martinez-Salamanca J, et al. Lessons learned from the International Renal Cell Carcinoma-Venous Thrombus Consortium (IRCC-VTC), Curr Urol Rep 2014;15:404.
23. Gaudino M, Lau C, Cammertoni F, Vargiu V, Gambardella I, Mas-setti MGirardi L. Surgical Treatment of renal cell carcinoma with cavoatrial involvement: a systematic review of the literature. Ann Throc Surg. 2016;101(3):1213-21.