A simple-to-use nomogram to predict long term survival of patients undergoing coronary artery bypass grafting (CABG) using bilateral internal thoracic artery grafting technique


Authors: Tomer Ziv-Baran aff001;  Rephael Mohr aff002;  Dmitry Pevni aff002;  Yanai Ben-Gal aff002
Authors place of work: Department of Epidemiology and Preventive Medicine, School of Public Health, Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel aff001;  Department of Cardio-Thoracic Surgery, Tel-Aviv Sourasky Medical Center, and Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel aff002
Published in the journal: PLoS ONE 14(10)
Category: Research Article
doi: 10.1371/journal.pone.0224310

Summary

Background

Several risk scores have been created to predict long term mortality after coronary artery bypass grafting (CABG). Several studies demonstrated a reduction in long-term mortality following bilateral internal thoracic arteries (BITA) compared to single internal thoracic artery. However, these prediction models usually referred to long term survival as survival of up to 5 years. Moreover, none of these models were built specifically for operation incorporating BITA grafting.

Methods

A historical cohort study of all patients who underwent isolated BITA grafting between 1996 and 2011 at Tel-Aviv Sourasky medical center, a tertiary referral university affiliated medical center with a 24-bed cardio-thoracic surgery department. Study population (N = 2,935) was randomly divided into 2 groups: learning group which was used to build the prediction model and validation group. Cox regression was used to predict death using pre-procedural risk factors (demographic data, patient comorbidities, cardiac characteristics and patient's status). The accuracy (discrimination and calibration) of the prediction model was evaluated.

Methods and findings

The learning (1,468 patients) and validation (1,467 patients) groups had similar preoperative characteristics and similar survival. Older age, diabetes mellitus, chronic obstructive lung disease, congestive heart failure, chronic renal failure, old MI, ejection fraction ≤30%, pre-operative use of intra-aortic balloon, and peripheral vascular disease, were significant predictors of mortality and were used to build the prediction model. The area under the ROC curves for 5, 10, and 15-year survival ranged between 0.742 and 0.762 for the learning group and between 0.766 and 0.770 for the validation group. The prediction model showed good calibration performance in both groups. A nomogram was built in order to introduce a simple-to-use tool for prediction of 5, 10, and 15-year survival.

Conclusions

A simple-to-use validated model can be used for a prediction of 5, 10, and 15-year mortality after CABG using the BITA grafting technique.

Keywords:

Coronary artery bypass grafting – Chronic obstructive pulmonary disease – Learning curves – Myocardial infarction – Surgical and invasive medical procedures – Vascular surgery – Peripheral vascular disease

Introduction

Coronary artery bypass grafting (CABG) procedures are the most common open heart procedure performed in most heart surgery centres. The standard of care using the left internal thoracic artery (LITA) as a single arterial graft was outlined in the mid-eighties by Loop et al. who showed better long–term patency of a LITA graft to left anterior descending (LAD) compared to that of a saphenous vein graft (SVG) connected to the LAD [1]. Several studies demonstrated a reduction in long-term mortality following bilateral internal thoracic arteries (BITA) compared to single internal thoracic artery (SITA) [26].

Several prediction models for long term survival after CABG have been proposed over the last years [715]. However, these models usually referred to long term survival as survival of up to 5 years [811, 13, 15], while only a few observed 7 or 8-year survival [7, 14, 12]. To the best of our knowledge, no study has published a model predicting survival beyond this period. Moreover, none of these models were built specifically for BITA grafting.

The main objective of the present study was to identify predictors for long term survival after BITA grafting and to build a simple validated prediction tool.

Methods

Study design

Historical cohort study.

Setting

A 24-bed cardio-thoracic surgery department located at the Tel Aviv Sourasky Medical Center, a university affiliated 1500-bed medical center. The medical center is located in the center of Israel and serves a population of nearly 1 million people. The Tel Aviv Sourasky Medical Center ethics committee approved the study and waived the requirement for informed consent (number 0827-18-TLV).

Participants

All patients who underwent CABG using two ITAs between January 1996 and December 2011 were included in the study. Patients who underwent concomitant procedures were excluded from the study.

In our department the use of BITA grafting has been the standard of care since 1996 [16], and is employed for all patients unless the patient has single vessel disease, increased risk of sternal wound infection (mostly female patients with combination of diabetes and obesity), or the consensus among senior staff is that the patient is too sick to undergo BITA grafting.

Variables

Data on patients' characteristics were recorded at the time of admission. Demographic data (age and sex); comorbidities: diabetes mellitus (DM), chronic obstructive lung disease (COPD), chronic renal failure (CRF), peripheral and cerebral-vascular disease (PVD, CVD), and any neurologic dysfunction (ND); cardiac characteristics: congestive heart failure (CHF), old and acute myocardial infarction (MI), unstable angina, severely reduced ejection fraction (EF), previous percutaneous transluminal coronary angioplasty (PCI), 3 or more diseased vessels, and left main disease (LM); patient's status: need of pre-operative intra-aortic balloon (IABP), emergency operation, and critical pre-operative status; and surgical characteristics: repeated operation, 3 or more bypass grafts, sequential grafts, the use of saphenous vein grafts, gastro-epiploic, radial grafts, grafts to the right coronary system and the use of off-pump technique, were collected. Old and acute myocardial infarction (MI) were considered as MI more than 7 days before surgery, and MI within the week before the surgery, respectively. Severely reduced ejection fraction was considered as ejection fraction equal or less than 30%. Critical pre-operative status was defined as the status of a patient who needed ventilation or IABP support or inotropic support, or any combination thereof before surgery. Emergency surgery was a surgery performed within the first week after cardiac catheterization due to the patient's condition. Mortality was defined as all-cause mortality occurring up to 15 years after surgery.

Data sources

All data was collected from patient files and hospital databases by residents or senior physicians. Late mortality was collected from the Israeli Internal Affairs Office National Registry Database.

Bias

All residents and senior physicians were instructed by one of the researchers (Mohr R) on the data collection in order to minimize collection bias and all data were recorded into an Excel file.

Sample size

Sample size was calculated according to log-rank test using significance level of 0.5% and power of 90%. The hazard ratio was set to 1.5 and survival at the end of follow-up was considered as 60%. According to these parameters, 1,046 patients were needed to identify predictors for mortality and to build the prediction model.

Quantitative variables

Age was divided into 5-year categories in order to simplify the use of the prediction model.

Statistical methods

Categorical variables were reported as number and percentage. Age was reported as mean and standard deviation (SD). Median and interquartile range of the surgery period was also reported. Length of follow-up was observed using reverse censoring method. Study population was randomly divided into two equal size groups. One group was used to identify predictors for mortality and to build the prediction model (learning group) and the second group was used to validate the model (validation group). The characteristics of the two groups were compared in order to evaluate the randomization. Chi-square test, independent samples T test, and Mann Whitney test were used to compare the characteristics between groups. Log-rank test was used to compare survival between groups.

Multivariable cox regression using backward method (likelihood ratio was used as criteria and p>0.1 was set as threshold value for removal). Demographic data, comorbidities, cardiac characteristics, and patients' status were included in the multivariable cox regression. Nomogram was built in order to introduce a simple tool for prediction of 5, 10, and 15-year survival. We also evaluated the baseline hazard at these time points and reported the coefficients of the variables that were included in the final model.

Calibration plots (predicted vs. observed values) were drawn in order to evaluate the prediction model. A predicted value indicates the predicted survival probability at a single time point, and observed value refers to the corresponding Kaplan-Meier survival estimate.

Harrell's concordance (C) statistic was used to evaluate the overall model discrimination ability. The area under the receiver operating characteristics curve was used to evaluate the discrimination ability of the model at these time points in both the learning and validation groups.

All statistical tests were two sided and p<0.05 was considered as statistically significant. SPSS (IBM SPSS Statistics for Windows, version 24, IBM corp., Armonk, NY, USA, 2016) and R (version 3.5.1, R Foundation for Statistical Computing, Vienna, Austria, 2018) were used for all statistical analysis. The following "R" packages were used: “rms” package (Harrell FE Jr. rms: Regression Modeling Strategies, ver. 5.1–2), “survival” package (Therneau TM. survival:Survival Analysis, ver. 2.42–6), "prodlim" package (Gerds TA. prodlim:Product-Limit Estimation for Censored Event History Analysis, ver. 2018.04.18), "dynpred" package (Putter H. Companion Package to "Dynamic Prediction in Clinical Survival Analysis", ver. 0.1.2), and "survivalROC" package (Heagerty PJ. survivalROC:Time-dependent ROC curve estimation from censored survival data, ver. 1.0.3).

Results

The study cohort included 2,935 patients who underwent BITA grafting between 1996 and 2011 (median January 2001). The median follow-up time was 15 years (IQR 12–15). One thousand four hundred and sixty eight patients were included in the learning group and 1,467 in the validation group. Comparisons of patients' characteristics are presented in Table 1.

Tab. 1. Comparison of patients' characteristics between the learning and the validation groups.
Comparison of patients' characteristics between the learning and the validation groups.

Six hundred twenty-six (42.6%) patients in the learning group and 593 (40.4%) patients in the validation group died during the follow-up (p = 0.437). The median survival time was not reached over the 15 years of follow up. The cumulative 5, 10, and 15 year survival rates were 0.861, 0.706, and 0.503, respectively for the learning group, and 0.865, 0.716, and 0.526, respectively, for the validation group (Fig 1).

Kaplan-Meir curve demonstrating the cumulative survival during the follow-up period in the learning and validation groups.
Fig. 1. Kaplan-Meir curve demonstrating the cumulative survival during the follow-up period in the learning and validation groups.

In multivariable analysis based on the learning group, older age, diabetes mellitus, COPD, CHF, CRF, old MI, EF≤30%, pre-operative intra-aortic balloon, and PVD were significant predictors for mortality in the learning group (Table 2).

Tab. 2. Predictors for mortality in the learning group and the multivariable cox regression coefficients.
Predictors for mortality in the learning group and the multivariable cox regression coefficients.

The calibration curves of the model presented good calibration at the three time points, in both the learning and validation groups (Fig 2 and Fig 3).

Calibration curve of the multivariable cox regression model based on the learning group.
Fig. 2. Calibration curve of the multivariable cox regression model based on the learning group.
Calibration curve of the prediction model based on the validation group.
Fig. 3. Calibration curve of the prediction model based on the validation group.

Harrell's C was 71.5% in the learning group. The AUCs of cox model for the learning group for 5, 10, and 15 years were 0.742, 0.755, and 0.762, and for the validation group 0.766, 0.763, and 0.770, respectively. The ROC curves are presented in S1 and S2 Figs.

Fig 4 shows a nomogram based on the cox model with prediction of mortality 5, 10, and 15 years after surgery. Table 3 indicates the points for each variable in the nomogram and Table 4 shows the total points for several main probabilities of mortality.

Nomogram presenting the prediction model for 5, 10 and 15-year survival.
Fig. 4. Nomogram presenting the prediction model for 5, 10 and 15-year survival.
Nomogram shows cox model for prediction of mortality 5, 10, and 15 years after surgery. Each parameter has corresponding values (points) that appear in the upper toolbar (also in Table 3). Summarized total points should be applied on the bottom scale ("Total Point") to obtain probability of mortality 5, 10 and 15 years after surgery. DM—Diabetes mellitus; COPD—Chronic obstructive lung disease; CHF—Congestive heart failure; CRF—Chronic renal failure; MI—Myocardial infarction; EF—Ejection fraction; IABP—Intra-aortic balloon; PVD—Peripheral vascular disease.
Tab. 3. Points that each variable represents in the nomogram.
Points that each variable represents in the nomogram.
Tab. 4. Predicted mortality for 5, 10, and 15 years after surgery, according to sum of points on the nomogram.
Predicted mortality for 5, 10, and 15 years after surgery, according to sum of points on the nomogram.

Discussion

Several risk scores were developed to evaluate the risk associated with early mortality after CABG [1719], and several risk scores were also developed to evaluate long term mortality [715]. However, these models evaluated late mortality of up to 8 years post-operatively. Similar prediction models were also developed in other medical areas, for example: a model to predict mortality in patients with prostate cancer [20], a model to predict pulmonary hypertension based on findings of CT pulmonary angiography [21], and a model to predict early relapse in patients with Crohn`s disease [22].

Several studies have shown significant benefit of BITA grafting compared to SITA grafting [16, 2326]. Therefore, models that were built according to patients who mostly underwent SITA grafting are limited in their applicability to predict outcomes of patients who underwent BITA grafting.

In the current study, we tried to develop a model that identifies predictors for long term survival, up to fifteen years after BITA grafting and use it as a simple validated prediction tool. This model may provide a longer prediction period than the previous ones [715], and also specifically addresses the sub group of patients treated with BITA, who were not included in previous studies.

In the current study, we divided all BITA patients operated in our department between 1966 and 2011 into two groups: One thousand four hundred and sixty-eight were the learning group and 1,467 were the validation group. The two groups had similar overall mortality and similar 5, 10, and 15-year survival.

Older age, DM, COPD, CHF, CRF, old MI, EF≤30%, pre-operative intra-aortic balloon, and PVD, were significant predictors for mortality in the current model. A previous model proposed by Wu et al. [7] for prediction of 7-year mortality after CABG included age, body mass index (BMI), EF, hemodynamic stability, left main coronary artery disease, CVD, PVD, CHF, malignant ventricular arrhythmia, COPD, DM, CRF, and previous CABG. Another previous model proposed by MacKenzie et al. [14] for prediction of 8-year mortality after CABG included age, BMI, COPD, DM, EF, gender, left main coronary artery disease, old MI, number of diseased vessels, prior CABG, PVD, CRF, patients' status (elective/urgent, salvage, emergent), and white blood cells. Patients with malignant ventricular arrhythmia were not referred to BITA in our center and therefore this risk factor was not included in our model. Additionally, data on patients' BMI was not available for a large proportion of the charts, and was therefore waived. However, most of the variables that were included in these previous studies were also included as potential predictors in our study.

The Harrell's C in the current model was close to that reported by MacKenzie et al [14]. The AUCs of the cox model for 5, 10, and 15-year survival for the learning group ranged between 0.742 and 0.762, and for the validation group between 0.766–0.770, respectively. The AUCs value for 5 years were close to that reported in previous studies [7,14].

This study has several limitations. First, in order to evaluate long term outcome, we had to use data on patients who underwent CABG between 1996 and 2011. There was an improvement in patient management which may only be partially reflected in the cohort. Second, due to the historical nature of the study, data on BMI and blood tests were available only in part of the patients' charts, and therefore these variables were not included as potential predictors. Third, we did not have data on specific cause of death and therefore all-cause mortality was used instead. Lastly, supplemental angiographic data that could have been included in the model was not available.

In conclusion, a simple to use validated model can be used to predict 5, 10, and 15-year mortality after CABG using the BITA grafting technique.

Supporting information

S1 Fig [tif]
ROC curves of the Cox model for 5, 10, and 15-year survival in the learning group.

S2 Fig [tif]
ROC curves of the Cox model for 5, 10, and 15-year survival in the validation group.


Zdroje

1. Loop FD, Lytle BW, Cosgrove DM, Stewart RW, Goormastic M, Williams GW, et al. Influence of the internal-mammary-artery graft on 10-year survival and other cardiac events. N Engl J Med 1986;314:1–6. doi: 10.1056/NEJM198601023140101 3484393

2. Lytle BW, Blackstone EH, Loop FD, Houghtaling PL, Arnold JH, Akhrasss R, et al. Two internal thoracic arteries are better than one. J Thorac Cardiovasc Surg 1999;117:855–72. doi: 10.1016/S0022-5223(99)70365-X 10220677

3. Buxton BF, Komeda M, Fuller JA, Gordon I. Bilateral internal thoracic artery grafting may improve outcome of coronary artery surgery. Risk-adjusted survival. Circulation 1998;98:II1–6 9852872

4. Endo M, Nishida H, Tomizawa Y, Kasanuki H. Benefit of bilateral over single internal mammary artery grafts for multiple coronary artery bypass grafting. Circulation 2001;104:2164–70. doi: 10.1161/hc4301.098283 11684625

5. Tector AJ, Amundsen S, Schmahl TM, Kress DC, Peter M. Total revascularization with T-grafts. Ann Thorac Surg 1994;57:8–9. doi: 10.1016/0003-4975(94)90356-5

6. Taggart DP, D'Amico R, Altman DG. Effect of arterial revascularization on survival: a systematic review of studies comparing bilateral and single internal mammary arteries. Lancet 2001;358:870–5. doi: 10.1016/S0140-6736(01)06069-X 11567701

7. Wu C, Camacho FT, Wechsler AS, Lahey S, Culliford AT, Jordan D, et al. Risk score for predicting long-term mortality after coronary artery bypass graft surgery. Circulation. 2012;125(20):2423–2430. doi: 10.1161/CIRCULATIONAHA.111.055939 22547673

8. Tanaka S, Sakata R, Marui A, Furukawa Y, Kita T, Kimura T. Predicting long-term mortality after first coronary revascularization:–the Kyoto model. Circ J. 2012;76(2):328–334. doi: 10.1253/circj.cj-11-0398 22094912

9. Shahian DM, O'Brien SM, Sheng S, Grover FL, Mayer JE, Jacobs JP, et al. Predictors of long-term survival after coronary artery bypass grafting surgery: results from the Society of Thoracic Surgeons Adult Cardiac Surgery Database (the ASCERT study). Circulation. 2012;125(12):1491–1500. doi: 10.1161/CIRCULATIONAHA.111.066902 22361330

10. Campos CM, van Klaveren D, Farooq V, Simonton CA, Kappetein AP, Sabik J, et al. Long-term forecasting and comparison of mortality in the Evaluation of the Xience Everolimus Eluting Stent vs. Coronary Artery Bypass Surgery for Effectiveness of Left Main Revascularization (EXCEL) trial: prospective validation of the SYNTAX Score II. Eur Heart J. 2015;36(20):1231–1241. doi: 10.1093/eurheartj/ehu518 25583761

11. Carr BM, Romeiser J, Ruan J, Gupta S, Seifert FC, Zhu W, et al. Long-Term Post-CABG Survival: Performance of Clinical Risk Models Versus Actuarial Predictions. J Card Surg. 2016;31(1):23–30. doi: 10.1111/jocs.12665 26543019

12. Kilpin M, Talwar A, Meneguzzi J, Tran L, Reid C, Hayward P. Two Long-Term Mortality Risk Models For Coronary Artery Bypass Graft Surgery Produced in American Populations Validated in an Australian Population. Heart Lung Circ. 2018;27(1):79–88. doi: 10.1016/j.hlc.2017.02.008 28389195

13. Karim MN, Reid CM, Huq M, Brilleman SL, Cochrane A, Tran L, et al. Predicting long-term survival after coronary artery bypass graft surgery. Interact Cardiovasc Thorac Surg. 2018;26(2):257–263. doi: 10.1093/icvts/ivx330 29049838

14. MacKenzie TA, Malenka DJ, Olmstead EM, Piper WD, Langner C, Ross CS, et al. Prediction of survival after coronary revascularization: modeling short-term, mid-term, and long-term survival. Ann Thorac Surg. 2009;87(2):463–572. doi: 10.1016/j.athoracsur.2008.09.042 19161761

15. Wu C, Camacho FT, Zhao S, Wechsler AS, Culliford AT, Lahey SJ, et al. Long-term mortality of coronary artery bypass graft surgery and stenting with drug-eluting stents. Ann Thorac Surg. 2013;95(4):1297–1305. doi: 10.1016/j.athoracsur.2012.11.073 23391171

16. Pevni D, Uretzky G, Mohr A, Braunstein R, Kramer A, Paz Y, et al. Routine use of bilateral skeletonized internal thoracic artery grafting: long-term results. Circulation 2008;118:705–12. doi: 10.1161/CIRCULATIONAHA.107.756676 18663087

17. Roques F, Nashef SA, Michel P, Gauducheau E, de Vincentiis C, Baudet E, et al. Risk factors and outcome in European cardiac surgery: analysis of the EuroSCORE multinational database of 19030 patients. Eur J Cardiothorac Surg. 1999;15(6):816–823. doi: 10.1016/s1010-7940(99)00106-2 10431864

18. Nashef SA, Roques F, Michel P, Gauducheau E, Lemeshow S, Salamon R. European system for cardiac operative risk evaluation (EuroSCORE). Eur J Cardiothorac Surg. 1999;16(1):9–13. doi: 10.1016/s1010-7940(99)00134-7 10456395

19. Shahian DM, O'Brien SM, Filardo G, Ferraris VA, Haan CK, Rich JB, et al. The Society of Thoracic Surgeons 2008 cardiac surgery risk models: part 1—coronary artery bypass grafting surgery. Ann Thorac Surg. 2009 Jul;88(1 Suppl):S2–22. doi: 10.1016/j.athoracsur.2009.05.053 19559822

20. Brockman JA, Alanee S, Vickers AJ, Scardino PT, Wood DP, Kibel AS, et al. Nomogram Predicting Prostate Cancer-specific Mortality for Men with Biochemical Recurrence After Radical Prostatectomy. Eur Urol. 2015;67(6):1160–1167. doi: 10.1016/j.eururo.2014.09.019 25301759

21. Aviram G, Shmueli H, Adam SZ, Bendet A, Ziv-Baran T, Steinvil A, et al. Pulmonary Hypertension: A Nomogram Based on CT Pulmonary Angiographic Data for Prediction in Patients without Pulmonary Embolism. Radiology. 2015;277(1):236–246. doi: 10.1148/radiol.15141269 25961630

22. Ziv-Baran T, Hussey S, Sladek M, Amil Dias J, Martin de Carpi J, Miele E, et al. Response to treatment is more important than disease severity at diagnosis for prediction of early relapse in new-onset paediatric Crohn's disease. Aliment Pharmacol Ther. 2018;48(11–12):1242–1250. doi: 10.1111/apt.15016 30450578

23. Buxton BF, Komeda M, Fuller JA, Gordon I. Bilateral internal thoracic artery grafting may improve outcome of coronary artery surgery. Risk-adjusted survival. Circulation. 1998;98(19 Suppl):II1–6. 9852872

24. Schmidt SE, Jones JW, Thornby JI, Miller CC 3rd, Beall AC Jr. Improved survival with multiple left-sided bilateral internal thoracic artery grafts. Ann Thorac Surg. 1997;64(1):9–14. doi: 10.1016/s0003-4975(97)00473-6 9236328

25. Lytle BW, Blackstone EH, Loop FD, Houghtaling PL, Arnold JH, Akhrass R, et al. Two internal thoracic artery grafts are better than one. J Thorac Cardiovasc Surg. 1999;117(5):855–872. doi: 10.1016/S0022-5223(99)70365-X 10220677

26. Taggart DP, D'Amico R, Altman DG. Effect of arterial revascularisation on survival: a systematic review of studies comparing bilateral and single internal mammary arteries. Lancet. 2001;358(9285):870–875. doi: 10.1016/S0140-6736(01)06069-X 11567701


Článek vyšel v časopise

PLOS One


2019 Číslo 10