Cite this article as: Gilmanov D, Farneti PA, Ferrarini M, Santarelli F, Murzi M, Miceli A et al. Full sternotomy versus right anterior minithoracotomyfor isolated aortic valve replacement in octogenarians: a propensity-matched study. Interact CardioVasc Thorac Surg 2015;20:732–42.

Full sternotomy versus right anterior minithoracotomyfor isolated aortic valve replacement in octogenarians:

a propensity-matched study†

Daniyar Gilmanov*, Pier Andrea Farneti, Matteo Ferrarini, Filippo Santarelli, Michele Murzi,

Antonio Miceli, Marco Solinas and Mattia Glauber

Department of Adult Cardiac Surgery, G. Pasquinucci Heart Hospital, Gabriele Monasterio Foundation, Massa, Italy

* Corresponding author. Gabriele Monasterio Foundation, G. Pasquinucci Heart Hospital, 305 Via Aurelia Sud, loc. Montepepe, 54100 Massa, Italy.Tel: +39-0585-493604; fax: +39-0585-493614; e-mail: drgilmanov@hotmail.it (D. Gilmanov).

Received 25 September 2014; received in revised form 18 January 2015; accepted 23 January 2015

Abstract

OBJECTIVES: Surgical aortic valve replacement (AVR) is increasingly performed in elderly patients with good perioperative outcomes andlong-term survival, resulting in significant health-related quality-of-life benefits. This study aimed to evaluate the outcome of patients aged≥80 years undergoing isolated AVR through a right anterior minithoracotomy (RAMT) and compare it with a full sternotomy (FS).

METHODS: Two hundred and eighty-three elderly patients aged 80 years or more underwent isolated AVR between February 2001 andSeptember 2013. With propensity score matching (1 : 1), the outcomes of patients having minimally invasive surgery (RAMT) were compared withthose in whom the FS approach had been employed (100 vs 100 patients). TAVRs and partial sternotomy cases were excluded from the analysis.

RESULTS: There were two conversions in the RAMT group. Operative times did not significantly differ in the two groups. Patients in the RAMTgroup received a larger-sized prosthesis (P < 0.001) and were more likely to receive sutureless valves (P < 0.001). Shorter time for extubation(P < 0.001) and shorter hospital length of stay (P = 0.005) were observed in the RAMT group. Zero vs 4 (4.0%) (P = 0.043) patients had post-operative stroke and 2 (2.0%) vs zero (P = 0.16) had a transient ischaemic attack in the RAMT versus FS group, respectively. We registered thesame rate of permanent pacemaker implant (P = 0.47) and that of new-onset atrial fibrillation (P = 0.28) for both groups. Six patients died,with no significant difference for in-hospital mortality (P = 0.68). No variable had a statistically significant predictive value for in-hospital mor-tality. RAMT patients were more likely to be discharged home directly or via rehabilitation (P = 0.031). FS, along with four other factors, inde-pendently predicted longer hospital stay. Though the median follow-up duration was longer in the FS group (59 vs 24 months, P < 0.001), thetwo groups had similar survival rates at 5 years (80 vs 81%, P = 0.37). Ten factors were associated with long-term survival by Cox regressionanalysis, and RAMT had no statistical impact (P = 0.38).

CONCLUSIONS: Minimally invasive AVR through right anterior minithoracotomy can be safely performed in patients aged ≥80 years withacceptable morbidity and mortality rates. It is an expeditious and effective alternative to full sternotomy AVR and might be associated withlower postoperative stroke incidence, earlier extubation and shorter hospital stay.

Keywords: Aortic valve replacement •Minimally invasive surgery • Biological heart valve prosthesis •Outcomes • Statistics, propensity matching

INTRODUCTION

Life expectancy in the Western world continues to increase, creatinga challenging demand in healthcare systems for future decades [1].

In Europe, the population older than 80 years of age is pro-jected to be 5.7% by 2020 and ultimately double in 2060(Available at: http://epp.eurostat.ec.europa.eu/statistics_explained/index.php/Population_projections), and Italy is not an outlier [2].

Overall, the current population will be living longer with morehealth problems than in the past. Furthermore, there is evidenceof a higher prevalence of cardiovascular disease with advancedage, and severe calcific aortic stenosis is the most frequent valvularheart disease found in octogenarians. Its prevalence rate steadilyincreases with age, from 2.5% at 75 years to 8.1% at 85 years; thus,the cardiac surgeon is likely to be challenged by an increasingvolume of elderly patients in need of heart surgery evaluation.Aortic valve replacement (AVR) significantly improves quality

of life and provides excellent short- and long-term outcomes.Previous reports on risk stratification and outcomes of heart surgery

†Presented at the 28th Annual Meeting of the European Association for Cardio-Thoracic Surgery, Milan, Italy, 11–15 October 2014.

© The Author 2015. Published by Oxford University Press on behalf of the European Association for Cardio-Thoracic Surgery. All rights reserved.

Interactive CardioVascular and Thoracic Surgery 20 (2015) 732–742 ORIGINAL ARTICLE – ADULT CARDIACdoi:10.1093/icvts/ivv030 Advance Access publication 10 March 2015

in the elderly have concluded with favourable results in survival andquality of life [3–5].

Although a difference between the so-called chronological andbiological age may often be found, the ability of the elderly patientto cope with a major physiological insult, such as heart surgery, islow due to existing comorbidity, limited functional reserve of vitalorgans, and reduced defence and adaptation capacity [6].

Surgical AVR has been shown to improve symptoms and survival,yet patients with advanced age are at an increased risk of surgicalcomplications or death [7–11], and therefore, a less-invasive treat-ment strategy is desirable in such patients.

Minimally invasive AVR (MIAVR) has been shown to offer nu-merous advantages. Several earlier reports favourably comparedright anterior minithoracotomy (RAMT) with full and partial ster-notomy [12–15], and in the experience of our centre the RAMTwas the most promising surgical approach for AVR [16].

However, there is still limited evidence on the performance ofminimally invasive AVR in the geriatric population; less is knownon RAMT outcomes in the elderly, and so far no direct compari-son of the matched population was performed with the conven-tional surgical approach. We hypothesized that RAMT versus fullmedian sternotomy could benefit octogenarians to the sameextent as younger patients. This study aimed to evaluate the out-comes of patients aged ≥80 years undergoing isolated AVRthrough RAMT and compare them to those of conventional fullsternotomy (FS).

MATERIALS ANDMETHODS

The clinical audit committee of the G. Pasquinucci Heart HospitalInstitutional Board approved the study to meet ethical and legalrequirements, and all patients gave their written informed consent.

This was a retrospective, observational, cohort study of pro-spectively collected data from 516 patients who underwent iso-lated AVR through RAMT between August 2004 and September2013. Of them, 116 were aged ≥80 years; 116 elderly patientsunderwent AVR through median sternotomy between February2001 and September 2013, and 51 through partial sternotomy.Thus, a total of 283 elderly patients undergoing isolated AVR wereavailable for analysis. However, in order to obtain a more homo-geneous population, we preferred to exclude partial sternotomycases from the study. Transcatheter aortic valve implantationswere excluded from the analysis as well. Two patients with conver-sion to FS were analysed as intent-to-treat.

With propensity score matching (1 : 1), the outcomes of patientshaving minimally invasive surgery (RAMT) were compared withthose in whom a full median sternotomy approach (FS) had beenemployed (100 vs 100).

Morrow septal myectomy, atrial fibrillation (AF) ablation, leftatrial appendage closure and anterior mitral leaflet decalcification(‘shaving’) performed through aortotomy were (minor) associatedprocedures.

The completed data collection forms were entered in localdatabases and included several sections completed by the anaes-thesiologists, cardiac surgeons, intensive care unit (ICU) personneland perfusionists involved in the care of the patients.

Hospital mortality included all deaths within 30 days of oper-ation irrespective of where the death occurred and all deathsin hospital after 30 days among patients who had not been dis-charged after the index operation. Prolonged ventilatory supportwas defined as pulmonary insufficiency requiring mechanical

ventilatory support for more than 24 h. Prolonged ICU stay meant3 days of ICU or more, immediately after surgery and afterwards.New-onset AF was defined by the documentation of AF of anyduration at any point in the postoperative period on a rhythmstrip or 12-lead electrocardiogram. Neurological complicationsincluded permanent (with duration more than 72 h, or strokes)and transient (transient ischaemic attacks or reversible cerebro-vascular accidents, CVAs) neurological events. Postoperative stroke(permanent CVA) was defined as any new permanent major(type II) neurological deficit that occurred anytime during thepostoperative hospitalization and/or new findings on computedtomography (CT) or magnetic resonance imaging (MRI), persist-ing for more than 72 h. All patients were seen 2–3 months post-operatively and, thereafter, were contacted for follow-up data.Follow-up information was obtained by telephone calls, e-mail,surface mail or interview. The median follow-up period was 33.7months (interquartile range, IQR, 21.8–69.5 months; cumulatedfollow-up 749 patient-years), and the follow-up data were 100%complete.

Anaesthetic and surgical techniques andpostoperative management

Anaesthetic and surgical techniques were standardized accordingto institutional protocols for all patients, and have been describedelsewhere [17].However, meticulous preoperative and postoperative care, in-

cluding aggressive early mobilization, is mandatory in elderlypatients to minimize complications and shorten postoperativestay. Nephrotoxic drugs are avoided when selecting prophylacticantibiotics. Central venous catheters are removed as soon as pos-sible to avoid sepsis. Enteral feeding is begun early postoperativelyand is used liberally.All surgeries were performed by eight senior surgeons of our

department, with no difference in operative technique, rigorouslyfollowing a unique internal protocol.MIAVR was performed through a right anterior thoracotomy

(6–7 cm) in the second intercostal space, without rib resection.Direct aortic cannulation was performed using low-profile cannulas

such as Easyflow (Sorin, Saluggia, Italy) or Straightshot (EdwardsLifeSciences, Irvine, California). Venous drainage was achieved with avariety of percutaneous venous cannulas such as BioMedicus multi-stage (Medtronic, Minneapolis, MN, USA), Quickdraw (EdwardsLifeSciences), RAP (Sorin) or Smartcanula (Smartcanula LLC, Lausanne,Switzerland) inserted through the (right) femoral vein into the venaecavae. The correct placement of the venous cannula was obtainedusing the Seldinger technique under transoesophageal echocardio-graphic guidance. Since vacuum-assisted cardiopulmonary bypass(negative pressure approximately −40 to −60 mmHg) was established,a left ventricular vent was placed through the right superior pulmonaryvein, and the patients were cooled down to 34 or 35°C if a suturelessor rapid deployment prosthesis was expected to be implanted. Theascending aorta was clamped with the Cygnet cross-clamp (NovareSurgical Systems, Cupertino, California) or with the Glauber’s aorticdetachable clamp (Cardiomedical GmbH, Langenhagen, Germany;distributed by the Sorin Group). Antegrade cardioplegic solutionwas given into the aortic root or selectively into the coronary ostiausing warm blood cardioplegia or cold crystalloid solution (CustodiolKoehler Chemie, Alsbach-Haenlein, Germany). In all cases of MIAVR,the surgical field was flooded with carbon dioxide at a flow of 0.5–1.0l/min. A standard implantation technique for both sutured (using a

ORIG

INALARTICLE

D. Gilmanov et al. / Interactive CardioVascular and Thoracic Surgery 733

pledgeted, interrupted suture technique for stented valves and con-tinuous running sutures for stentless valves) and sutureless prostheseswas implemented.

Intraoperative transoesophageal echocardiography was usedroutinely for the assessment of cardiac function, proper position-ing of the percutaneous venous cannula, surgical results evalu-ation and confirmation of the air removal process. At the end ofsurgery, patients were transferred to the ICU and managedaccording to the unit protocol.

Statistical analysis

Patients’ demographic and operative data were summarized asmean ± standard deviation for normally distributed continuousvariables, median (IQR 25th–75th percentiles) for non-parametriccontinuous variables or proportion/prevalence for categoricalvariables, as appropriate. The Kolmogorov–Smirnov test was usedto check the normality/skewness of continuous variables data inthe subgroups before further analysis, and appropriate statisticaltests have been chosen accordingly. Differences between sub-groups were compared using the χ2 test (two-tailed) for categoric-al variables (univariate analyses) and the Student’s t-test or theMann–Whitney U-test or Kruskal–Wallis test, as appropriate, forcontinuous variables.

To reduce the effect of selection bias and potential confound-ing in this observational study, we developed a propensity scoreanalysis. All the variables listed in Table 1 were included in theanalysis. A propensity score, indicating the predicted probabilityof receiving MIAVR treatment, was then calculated by the use of anon-parsimonious multiple logistic regression analysis from thelogistic equation for each patient. Finally, we used the propensityscore to match RAMT to FS patients (1 : 1 match). We used 5 to 1digit matching, searching for the nearest neighbour, to identifythe matched patients.

The composite event score included following complications:necessity for prolonged (over 24 h) mechanical ventilation support,reoperation for any cause, perioperative acute myocardial infarction,stroke, transient ischaemic attack, new-onset dialysis, multisystemfailure, atrioventricular block requiring pacemaker implantation, car-diorespiratory arrest, need for percutaneous pericardial drainage,major arrhythmia and death during the index hospitalization.

Survival was evaluated using the Kaplan–Meier method, and theMantel–Cox log-rank test was used to compare the survival curves.

We explored a total of 84 factors for inclusion in a multivariatemodel using univariate testing with a significance level of P < 0.10used for selection. Significance was subsequently assessed in themultivariate model at the P < 0.05 level. Binary logistic regressionanalysis of predictor variables for 30-day mortality, perioperativestroke and prolonged hospital stay was performed with estimateodds ratios (ORs) and 95% confidence intervals (CIs) for each ofthe independent variables in the model displayed.

The Cox multivariable proportional hazards regression modelbased on 93 pre- and postoperative variables was developed toidentify independent predictors of follow-up mortality in thestudied population and to assess the role of surgical approach inlate survival for both matched and unmatched patients. Hazardratios with 95% CIs were calculated for each of the significant riskfactors. All reported probability values are two-tailed, and prob-ability values of less than 0.05 were considered to indicate statis-tical significance. All statistical analysis was performed with SPSSsoftware, version 19.0 (SPSS, Inc., Chicago, IL, USA).

RESULTS

Within a 9-year study period (August 2004–September 2013), 516patients underwent AVR through RAMT. We selected 116 of them,aged ≥80 years. One hundred and sixteen elderly patients under-went AVR through median sternotomy between February 2001and September 2013 (Fig. 1). Baseline demographic and pre-operative data of the study population are presented in Table 1.Before propensity matching, patients in the RAMT group wereslightly older, had lower prevalence of arterial hypertension, AF,left ventricular dysfunction and pulmonary hypertension, andwere less likely to have a critical preoperative status.Following propensity score elaboration (−2 log likelihood of

the regression model = 279.6; P = 0.97 for lack of fit by Hosmer–Lemeshow), 100 pairs of patients have been matched (Table 1). Inthe matched cohorts, there was no longer any significant differencebetween the two groups for any covariate. In the resulting cohorts,there were 2 (2.0%) and 0 patients with previous heart surgery(RAMT versus FS, P = 0.16).There were 2 (2.0%) conversions in the RAMT group due to

intraoperative bleeding and technical impossibility of percutan-eously cannulating venae cavae, and these cases have been ana-lysed as intent-to-treat.In 3 (3.0%) vs 4 (4.0%) patients, at least one complementary pro-

cedure was performed (RAMT versus FS, P > 0.99). The distributionof the prosthesis type received by the patient, operative times andother intraoperative data for the two groups of patients are givenin Table 2. Generally, larger aortic valve prostheses have been im-planted in the RAMT group (23 vs 21 mm, P < 0.001), with ahigher prevalence of sutureless valves (71 vs 4%, P < 0.001).Operative times did not significantly differ in the two groups.The main clinical outcomes of the propensity-matched cohorts

are presented in Table 3. The duration of ICU stay was similarbetween the groups (P = 0.27); 79 (79%) in the RAMT group vs 72(72%) patients in the FS group were discharged from the ICUwithin 24 h (P = 0.25).However, shorter time for extubation (P < 0.001) and shorter hos-

pital length of stay (P = 0.005) were observed in the RAMT group.Six patients died, and there was no significant difference for

in-hospital mortality between the two groups (P = 0.68). A similar inci-dence of low cardiac output syndromewas observed as well (P = 0.42).Zero vs 4 (4.0%) (P = 0.043) patients had postoperative stroke and 2

(2.0%) vs 0 (P = 0.16) had a transient ischaemic attack in the RAMTversus FS group, respectively. We registered the same rate of perman-ent pacemaker implant (P = 0.47) and that of new-onset AF (P = 0.28).One patient with postoperative aortic dissection and one

patient with aorto-ventricular junction detachment in the RAMTgroup were urgently reoperated after the index intervention[2 (2.0%) vs 0 in the FS group, P = 0.16].One (1.0%) patient in the RAMT group had a surgical wound

dehiscence and 1 (1.0%) patient in the FS group had a deepsternal wound infection (P > 0.99).When forced into a multivariable logistic regression model

(−2 log likelihood = 20.55, P > 0.999 for goodness of fit by Hosmer–Lemeshow), among all available variables only prolonged (>24 h)artificial ventilation significantly correlated (OR 69.4; 95% CI 6.48–743.1, P < 0.001) with perioperative stroke, while no significantassociation was found with the surgical approach (P = 0.996).Even though several variables were associated by a logistic

binary regression model (P = 0.47 for lack of fit by Hosmer–Lemeshow; −2 log likelihood = 0.0) with in-hospital death, none of

D. Gilmanov et al. / Interactive CardioVascular and Thoracic Surgery734

them reached the level of statistical significance (P = 0.97–0.99 foreach: angina, postoperative cardiorespiratory arrest, chronic ob-structive pulmonary disease, assisted ventilation time and gastro-intestinal complications).

In the multivariable logistic regression model (−2 log likelihood =233.9, P = 0.874 for lack of fit by Hosmer–Lemeshow), five factorswere independently associated with prolonged hospital stay (7days or more). These factors included: FS approach (OR 1.88, 95%CI 0.99–3.56, P = 0.054), implanted aortic valve prosthesis size

(OR 0.84, 95% CI 0.71–1.0, P = 0.049), early (within 12 h) extuba-tion (OR 0.41, 95% CI 0.18–0.91, P = 0.028), cardiorespiratoryarrest (OR 0.08, 95% CI 0.006–1.06, P = 0.055) and new-onset AF(OR 3.16, 95% CI 1.53–6.58, P = 0.002).Roughly half of the patients in the RAMT group were discharged

from the hospital within 6 postoperative days versus one-third inthe FS group (P = 0.009; Table 3).Seventy-two (72%) patients in the RAMT group vs fifty-four

(54%) patients in the FS group were discharged directly home or

Table 1: Baseline preoperative data (variables included in the model for propensity matching)

Variable Before propensity matching After propensity matching

FS (n = 116) RAMT (n = 116) P-value FS (n = 100) RAMT (n = 100) P-value

Age (years) 82.0 (80.9–83.6) 82.9 (81.4–84.3) 0.016 81.9 (80.9–83.6) 82.9 (81.4–84.5) 0.1282.4 ± 2.1 83.0 ± 2.1 0.036 82.5 ± 2.2 83.0 ± 2.1 0.09

Female 75 (65%) 72 (62%) 0.68 67 (67%) 64 (64%) 0.66Height (cm) 161.5 (155.8–165.0) 162.0 (158.0–170.0) 0.62 162.5 (157.3–168.0) 163.0 (157.3–170) 0.58Weight (kg) 68.0 ± 12.0 68.8 ± 11.5 0.61 67.5 ± 11.8 68.1 ± 11.0 0.73Body mass index (kg/m2) 25.6 ± 3.7 25.7 ± 3.6 0.82 25.6 ± 3.8 25.6 ± 3.5 0.98Obesity 15 (13%) 19 (16%) 0.46 11 (11%) 15 (15%) 0.40Arterial hypertension 98 (85%) 86 (74%) 0.052 85 (85%) 76 (76%) 0.11Hypercholesterolaemia 58 (50%) 71 (61%) 0.086 53 (53%) 61 (61%) 0.25Diabetes mellitus type II 13 (11%) 18 (16%) 0.34 12 (12%) 17 (17%) 0.32Smoking history 31 (27%) 23 (20%) 0.21 27 (27%) 17 (17%) 0.09Family history of CAD 27 (23%) 33 (28%) 0.37 23 (23%) 29 (29%) 0.33Extracardiac arterial vascular disease 21 (18%) 22 (19%) 0.87 19 (19%) 16 (16%) 0.58Cerebral arterial vascular disease 11 (9.5%) 8 (6.9%) 0.47 10 (10%) 6 (6%) 0.30Previous TIA 1 (0.9%) 1 (0.9%) >0.99 1 (1%) 1 (1%) >0.99Previous stroke 2 (1.7%) 1 (0.9%) >0.99 2 (2%) 1 (1%) >0.99History of cardiac arrhythmia 35 (30%) 30 (26%) 0.47 26 (26%) 28 (28%) 0.75Preoperative atrial fibrillation 30 (26%) 16 (14%) 0.02 22 (22%) 16 (16%) 0.28Previous pacemaker implant 5 (4.3%) 4 (3.4%) >0.99 3 (3%) 3 (3%) >0.99Chronic renal failure 6 (5.2%) 7 (6.0%) 0.78 3 (3%) 4 (4%) >0.99Dialysis 0 0 n.a. 0 0 n.a.Infective endocarditis 0 1 (0.9%) >0.99 0 1 (1%) >0.99Chronic obstructive pulmonary disease 23 (20%) 13 (11%) 0.07 15 (15%) 11 (11%) 0.40With severe obstruction 9 (7.8%) 6 (5.2%) 0.42 8 (8%) 5 (5%) 0.39Chronic bronchodilator use 7 (6.0%) 3 (2.6%) 0.20 6 (6%) 2 (2%) 0.28

Previous interventional cardiological procedure 7 (6.0%) 13 (11%) 0.16 4 (4%) 10 (10%) 0.10Redo surgery 0 2 (1.7%) 0.50 0 2 (2%) 0.50Previous CABG 0 0 n.a. 0 0 n.a.Previous valvular surgery 0 2 (1.7%) 0.50 0 2 (2%) 0.50

Previous myocardial infarction 2 (1.7%) 7 (6.0%) 0.17 1 (1%) 1 (1%) >0.99Angina pectoris 31 (27%) 21 (18%) 0.12 24 (24%) 17 (17%) 0.22Previous or current malignancy 1 (0.9%) 3 (2.6%) 0.62 1 (1%) 3 (3%) 0.62Preoperative critical statusa 4 (3.4%) 0 0.044 0 0 n.a.Liver cirrhosis 2 (1.7%) 1 (0.9%) >0.99 2 (2%) 1 (1%) >0.99Congestive heart failure 26 (22%) 27 (23%) 0.88 22 (22%) 21 (21%) 0.86NYHA classification 2.5 ± 0.69 2.4 ± 0.62 0.19 2.4 ± 0.66 2.4 ± 0.62 0.74Functional class 3 or 4 by NYHA 59 (51%) 45 (39%) 0.065 48 (48%) 39 (39%) 0.20Aortic transvalvular gradient (mmHg) 83.2 ± 26.9 82.0 ± 23.2 0.74 84.8 ± 27.6 80.5 ± 21.7 0.27Left ventricular ejection fraction (%) 55 (50–60) 60 (55–65) 0.036 55 (50–60) 60 (55–60.8) 0.16Left ventricular ejection fraction 0.5 and less 24 (21%) 10 (8.6%) 0.009 15 (15%) 10 (10%) 0.29Left ventricular ejection fraction 0.3 and less 4 (3.4%) 1 (0.9%) 0.37 2 (2%) 1 (1%) >0.99Aortic valve pathologyStenosis 95 (82%) 105 (90%) 0.14 84 (84%) 89 (89%) 0.44Mixed lesion 15 (13%) 9 (8%) 11 (11%) 9 (9%)Regurgitation 6 (5%) 2 (2%) 5 (5%) 2 (2%)

Severe pulmonary hypertension 23 (20%) 11 (10%) 0.026 17 (17%) 11 (11%) 0.22Urgency or emergency 15 (13%) 26 (22%) 0.058 12 (12%) 13 (13%) 0.83EuroSCORE version I 8.50 (8–10) 8 (8–10) 0.64 8 (8–9) 8 (8–9) 0.84Logistic EuroSCORE version I 10.74 (8.44–16.1) 10.74 (8.44–16.2) 0.47 10.49 (8.44–14.61) 10.22 (8.44–14.61) 0.72

CABG: coronary artery bypass grafting; CAD: coronary artery disease; EuroSCORE: European System for Cardiac Operative Risk Evaluation; FS: full sternotomy;n.a.: not available; NYHA: New York Heart Association; RAMT: right anterior minithoracotomy; TIA: transient ischaemic attack.aAt least one of the following: intubated and ventilated, counterpulsated, resuscitated or oligoanuric patient.

ORIG

INALARTICLE

D. Gilmanov et al. / Interactive CardioVascular and Thoracic Surgery 735

to a rehabilitation centre (P = 0.031), while twenty-six (26%) vsfourty-three (43%) patients, respectively, needed further medicalcare in other hospital institutions.

During the median follow-up period of 33.7 months (IQR, 21.8–69.5 months; cumulated follow-up 749 patient-years), 35 latedeaths were registered, and 12 of them occurred in the RAMTgroup. Leading causes were cardiac—15 (of them, 2 valve-relateddeaths), followed by pulmonary—6, neurological—3 and infectivecauses—2.

Though the median follow-up duration was longer in the FSgroup: 59.2 (28.1–94.2) vs 24.2 (16.8–33.2), P < 0.001, the two groupshad a similar Kaplan–Meier survival rate at 1, 3 and 5 years (94 vs88%, 88 vs 81% and 80 vs 81%, respectively, P = 0.37; Figs 2 and 3).

Ten variables were associated with follow-up mortality by theCox multivariable proportional hazards regression model (−2 loglikelihood = 193.8; χ2 statistic 104.7; P < 0.001 for overall model fit),as illustrated in Table 4. These variables included higher height,smoke and stroke history, infective endocarditis, congestive heartfailure, higher logistic EuroSCORE, prolonged assisted ventilation,respiratory dysfunction, pleural effusion with need of drainage/puncture and longer hospital stay. In this model, for both the un-matched (P = 0.93) and matched (P = 0.38) population, the RAMTapproach did not exert any statistical impact on late survival (Fig. 4).

DISCUSSION

Increased life expectancy constitutes one of the greatest achieve-ments in human history. The rapidly increasing ageing population,frequently affected by various cardiovascular pathologies and

associated comorbidities, has prompted the call for a betterunderstanding of the needs and outcomes of elderly patientsundergoing cardiac surgery [9, 10].A growing body of evidence on the outcome of surgical and

transcatheter AVR of elderly patients has become available overthe last decade [3, 18–20].Following encouraging results of the numerous studies above,

growing numbers of elderly patients are referred at the presenttime for surgery; hence, surgical AVR is increasingly performedin elderly patients with good perioperative outcomes and long-term survival, and results in significant functional improvementand health-related quality-of-life benefits [3–5]. Advanced age,however, may pose an indication for less-invasive procedures,particularly in those patients aged >80 years, who are considereda vulnerable population due to associated comorbidities andfrailty.As shown in Fig. 1, there has been a continuous increase in the

number of octogenarian patients who had undergone isolatedAVR in our centre since 2001, and the proportion of MIAVR hadan explosive trend in 2011, with the introduction of suturelessprostheses.

Figure 1: Year-by-year distribution of octogenarian patients operated for isolatedaortic valve replacement. Relative height of each column corresponds to a totalnumber of patients operated on in the specific time frame. Numbers indicatepercentage proportion (%) of RAMT cases performed in a specific year (blue partof each column). RAMT: right anterior minithoracotomy; S: full sternotomy.

Table 2: Intraoperative data

Variables FS (n = 100) RAMT(n = 100)

P-value

Implanted prosthesisdiameter (mm)

21 (21–23) 23 (23–25) <0.001

Prosthesis size distribution19 mm 11 (11%) 1 (1%) <0.00121 mm 40 (40%) 22 (22%)23 mm 30 (30%) 35 (35%)25 mm 14 (14%) 40 (40%)27 mm 5 (5%) 2 (2%)

Prosthesis typeStented or stentlessbioprosthesis

96 (96%) 29 (29%) <0.001

Sutureless 4 (4%) 71 (71%)ATS 3F Enable suturelessprosthesis

2 (2%) 2 (2%) <0.001

Carpentier-EdwardsPerimount pericardialbioprosthesis

43 (43%) 24 (24%)

Edwards Intuity aorticvalve system

2 (2%) 1 (1%)

Medtronic Mosaicporcine bioprosthesis

17 (17%) 5 (5%)

Perceval suturelessprosthesis

0 68 (68%)

Sorin Freedom Solostentless bioprosthesis

31 (31%) 0

Mitroflow pericardialbioprosthesis

5 (5%) 0

CPB time (min) 94 (79–116) 92 (78–121) 0.91ACC time (min) 63 (52–78) 59 (48–76) 0.45Associated minor proceduresa 4 (4%) 3 (3%) >0.99CPB over 100 min 46 (46%) 38 (38%) 0.18ACC time over 70 min 36 (36%) 36 (36%) >0.99

ACC: aortic cross-clamping; CPB: cardiopulmonary bypass; FS: fullsternotomy; RAMT: right anterior minithoracotomy.aRadiofrequency ablation of atrial fibrillation, anterior mitral leafletdecalcification (‘shaving’), left atrial appendage closure or septalmyectomy by Morrow.

D. Gilmanov et al. / Interactive CardioVascular and Thoracic Surgery736

The present study demonstrates that the majority of patientsaged 80 or above admitted for AVR had pre-existent comorbidity,were taking one or more medications and had functional limita-tions of their illness by NYHA class. A considerable part of thepatients in this study required additional healthcare servicesbeyond their admission.

Importantly, the significant difference in the prolonged hospitalstay between the propensity-matched RAMT and FS groups,revealed by univariate analysis, was confirmed also by means ofthe multivariable regression model. Previously, reduced hospitalstay was demonstrated in elderly (over 65 years of age) MIAVR

patients compared with the conventional surgical approach in acase–control study by Sharony et al. [21], in a propensity-matchedpopulation by Glauber et al. [16], in a comparable population byLamelas et al. [14] and in 1 : 4 matched MIAVR patients byBrinkman et al. [13]. Shorter ICU stay and ventilation times, as well,have been already described by direct comparison studies [13, 14].Our internal protocols of patient management in the ICU do notprovide the possibility of transfer to a common ward in the nightand during a Sunday turn, unless particular circumstances occur(e.g. emergency case to be admitted in the ICU). Hence, thepatients are discharged from the ICU during the day and the localdatabase contains the ICU stay data expressed in days, rather thanin hours. While many published reports specify ICU stay in hours,the current study, in contrast to others, described the respectivedata by units of days. This might explain why an important param-eter of quality—the ICU stay (expressed in days)—did not differbetween the two study groups (FS versus RAMT), but the ventila-tion time (expressed in hours) did.The aortic valve prosthesis choice was left to the surgeon’s pref-

erence; yet, three important periods could be defined over thetime span of the current study. The first one (2001–2004) wascharacterized by availability of only a conventional stented pros-thesis; in the second one (2004–2010), a stentless prosthesis bythe Sorin Group was frequently used along with the stented pros-thesis. Superior haemodynamic performance of the stentlessvalves, especially in the patients with a small native annulus or fi-brotic/calcified root with reduced elasticity, was guiding motif ofsuch a novelty. The progressive development of transcathetertechnology anticipated an important paradigm shift in our depart-ment—from stentless to sutureless aortic valve prostheses inelderly patients.Thus, the third period came in early 2010, when a sutureless

prosthesis was implanted for the first time in our institution in anoctogenarian patient. The innovation of the sutureless prosthesisprovided a qualitative leap, as well, and thenceforth, the octogen-arian patients have been increasingly operated on in our depart-ment for isolated MIAVR (Fig. 1), while the number of FS cases hasconstantly declined.The majority of the patients in the RAMT group received

sutureless prostheses (Table 2), and this fact has probably deter-mined a larger aortic valve prosthesis size. In our direct daily prac-tice, the same native aortic annulus measured intraoperativelywith sizers of different prostheses (especially, with the sutured andsutureless ones) frequently allows for a larger-sized prosthesisversus the conventional one—usually, one unit higher (e.g. 23instead of 21 mm, etc.). However, the postoperative echocardio-graphic data on the effective orifice area were not available in allthe patients enrolled in the study and, therefore, the current re-search has no effective orifice area analysis. Though we mayspeculate that the size of the prosthesis could have an impact onpatient–prosthesis mismatch incidence, this issue was beyond thescope of the current study.Sharony et al. [21] in their comparative study reported similar

operative times in the MIAVR and FS groups: 110 vs 107 min(P = 0.87) for cardiopulmonary bypass and 78 vs 75 min (P = 0.34)for aortic cross-clamping. However, the data from analogouslydesigned studies that have been published by Ruttmann et al.[12], Brinkman et al. [13] (longer cross-clamping time, P < 0.001,but only a statistical trend to a longer perfusion time, P = 0.071)and Lamelas et al. [14] (P < 0.001 for both times, but regardingmixed—aortic and mitral—procedures) did not support thosefindings.

Table 3: Clinical outcome of propensity-matched patients

Variables FS(n = 100)

RAMT(n = 100)

P-value

ICU length of stay (days) 1 (1–2) 1 (1–1) 0.27Prolonged ICU stay 14 (14%) 9 (9%) 0.27Assisted ventilation time (h) 9 (7–12) 7 (5–9.8) <0.001Prolonged ventilation support 7 (7%) 4 (4%) 0.35Hospital length of stay (days) 7 (6–10) 7 (6–8) 0.005Hospital length of stay of7 days or more

70 (70%) 52 (52%) 0.009

Hospital length of stay of12 days or more

12 (12%) 7 (7%) 0.23

In-hospital mortality 4 (4%) 2 (2%) 0.68Low cardiac output syndrome 16 (16%) 12 (12%) 0.42Need for perioperative IABPplacement

0 1 (1.0%) >0.99

Reopening for bleeding orcardiac tamponade

4 (4%) 6 (6%) 0.52

Perioperative acute myocardialinfarction

0 1 (1%) >0.99

Infective complications 2 (2%) 5 (5%) 0.25Sepsis 0 0 n.a.Wound dehiscence or

surgical site infection1 (1%) 1 (1%) >0.99

Stroke 4 (4%) 0 0.043Transient ischaemic attack 0 2 (2%) 0.16Pulmonary complications/respiratory dysfunction

10 (10%) 14 (14%) 0.38

Pneumonia 0 2 (2%) 0.16Pleural effusion requiringpuncture

5 (5%) 4 (4%) 0.73

Perioperative acute kidneyinjury

11 (11%) 5 (5%) 0.12

New-onset CVVH/haemofiltration support

1 (1%) 0 >0.99

Complete AV block requiringPM implant

3 (3%) 5 (5%) 0.47

New-onset atrial fibrillation orflutter

34 (34%) 27 (27%) 0.28

Major arrhythmia 7 (7%) 10 (10%) 0.45Gastrointestinal complications 0 1 (1%) >0.99Postoperative confusion/delirium

12 (12%) 12 (12%) >0.99

Composite event0 82 78 0.671 10 142 4 53 3 24 0 15 1 0

AV: atrioventricular block; CVVH: continuous veno-venoushaemodialysis; FS: full sternotomy; IABP: intra-aortic balloon pump;ICU: intensive care unit; PM: pacemaker; RAMT: right anteriorminithoracotomy.

ORIG

INALARTICLE

D. Gilmanov et al. / Interactive CardioVascular and Thoracic Surgery 737

Similarly, our research group has reported earlier thatpropensity-matched patients in the RAMT group had longer car-diopulmonary bypass (121.6 ± 45 vs 107.1 ± 32.3 min, P = 0.003)and cross-clamping (86.9 ± 31.8 vs 72.1 ± 27.2 min, P < 0.0001)times, compared with conventional FS AVR [16].

We deem two aspects important to note with regard to opera-tive times observed in the present study. First, a considerable partof the patients in the FS group was operated on by the attendingsurgeons in the early 2000s, who were still under training, resulting

in relatively longer CPB and aortic cross-clamping times com-pared with average values. There was also a significant interper-sonal variability between the surgeons for dichotomized CPB(< or > than 100 and 90 min) and aortic cross-clamping (< or >than 70 and 60 min) duration (P < 0.001 for each comparison,calculated for the matched cohorts). Secondly, in one-third of theFS patients, the Sorin Freedom SOLO stentless prosthesis wasimplanted, which may have required a longer procedure, espe-cially within a small aortic annulus. A recent single-centre study

Figure 2: Overall survival by Kaplan–Meier for the unmatched population. Green line and numbers represent the right anterior minithoracotomy (RAMT) cohort(n = 116), and blue ones the full sternotomy (FS) cohort (n = 116). P = 0.59 by Mantel–Cox log-rank. Note the longer follow-up duration for FS. AVR: aortic valve replacement.

Figure 3: Overall survival by Kaplan–Meier for the matched population. Green line and numbers represent the right anterior minithoracotomy (RAMT) cohort (n = 100),and blue ones the full sternotomy (FS) cohort (n = 100). P = 0.37 by Mantel–Cox log-rank. Note the longer follow-up duration for FS. AVR: aortic valve replacement.

D. Gilmanov et al. / Interactive CardioVascular and Thoracic Surgery738

reported mean aortic cross-clamp times as long as 86.05 ± 34.2 minwith this prosthesis [22]. Kolh et al. [7] also indicated that stentlessprostheses might require a longer clamping time than stentedprostheses.

In the FS group (no stentless prostheses have been used in theRAMT group), there was only a statistically insignificant, weaktrend towards longer cross-clamping time with the Sorin FreedomSOLO prosthesis: 72.1 ± 25.4 compared with 64.8 ± 21.3 min(P = 0.14) in the rest of the patients.

Furthermore, the operative times in the MIAVR group—usuallylonger compared with conventional sternotomy—have been sig-nificantly improved by the use of sutureless valves, equallingthereby the disparity between the two groups.

The AVR crude early mortality rate of 4.0 and 2.0% for octogen-arian patients in the current study compares favourably with whatis reported in the literature: e.g., in-hospital mortality rate of 5.5%reported by Langanay et al. [19], 5.6% by the Carabello group [11],6.1% in the review by Biancari et al. [23] and 9% by Kolh et al. [7].

Even though only 2 patients died in the RAMT group versus 4in the FS group, this difference was not statistically significant inour study, and the trend is exactly the same as observed in arecent large review [23]. However, the above review regardedunselected patients of all ages, in contrast to our study.

Furthermore, postoperative stroke incidence reduction inpatients who had undergone MIAVR versus FS was reported byBiancari et al. [23], even though this was statistically insignificant.As far as we know, our report is the first direct comparative studyof minimally invasive and FS AVR in octogenarian patients to dem-onstrate the lower incidence of stroke with the RAMT approach.

No consistent association between the surgical approach and in-cidence of perioperative stroke was reported in the available litera-ture: one of the sources reported higher occurrence with MIAVR,which was, however, not statistically significant [14], and the othersreported lower or equal incidence with mini-AVR [13, 16, 23, 24].

In our study, this difference was statistically significant (P = 0.043),despite comparing only few events (0 cases vs 4 in the two study

groups). With regard to perioperative stroke predictors and multi-variable logistic regression analysis, it seems that a too smallnumber of events (four) was an important limitation of such an ana-lysis, so that the only inevitable aftereffect consequence of impairedcerebral function—prolonged ventilation in ICU—was correlatedwith acute irreversible cerebrovascular accident. Earlier, Sharonyet al. [21] did not reveal any significant associated risk factors forstroke by multivariate analysis in elderly patients.Interestingly, pacemaker implantation rate (P = 0.531) and

re-exploration for bleeding (P = 0.590) were also similar betweenthe MIAVR and FS patients in the above review by Biancari et al.[23]—with data which are very similar to ours.We observed a relatively better actuarial long-term survival in

this very elderly population—�80% at 5 years in both studygroups, compared with 72% by Saxena et al. [20], 77% by the Cohngroup [25] and 65.4% by the review of Vasques et al. [3].In the present study, 10 factors have been independently asso-

ciated with long-term survival by Cox regression analysis (Table 4),and RAMT had no statistical impact (Fig. 4). The protective

Table 4: Long-term mortality analysis by the Coxmultivariable proportional hazards regression model:independent predictors of mortality

Variable Hazard ratio[Exp (B)]

95%Confidenceinterval forhazard ratio

P-value

Lower Upper

Height 1.119 1.049 1.193 0.001Smoking history 0.13 0.032 0.493 0.003Stroke history 28.1 6.41 123.46 <0.001Infective endocarditis 99.3 9.57 1029.0 <0.001Congestive heart failure 4.3 1.92 9.67 <0.001Logistic EuroSCORE 1.055 1.004 1.110 0.035Prolonged artificialventilation

3.9 1.11 13.83 0.034

Respiratory dysfunction/pulmonarycomplications

4.5 1.65 12.36 0.003

Pleural effusion requesteddrainage/puncture

10.4 2.42 45.102 0.002

Hospital length of staylonger than 12 days

4.1 1.54 10.44 0.005

−2 log likelihood = 193.8; χ2 statistic 104.7; P<0.001 for overall model fit.

Figure 4: (A) Comparison between full sternotomy (FS) and right anterior mini-thoracotomy (RAMT) cohorts by the Cox multivariable proportional hazards re-gression model at the mean of covariates in the unmatched population(n = 232). Hazard ratio 0.96 (95% CI 0.40–2.29), P = 0.93. (B) Comparisonbetween full sternotomy (FS) and right anterior minithoracotomy (RAMT)cohorts by the Cox multivariable proportional hazards regression model at themean of covariates in the matched population (n = 200). Hazard ratio 0.61 (95%CI 0.21–1.76), P = 0.38. AVR: aortic valve replacement.

ORIG

INALARTICLE

D. Gilmanov et al. / Interactive CardioVascular and Thoracic Surgery 739

character of smoking history for long-term survival as illustrated inTable 4 deserves particular comment. We do not think that oneshould smoke to improve survival after AVR; on the contrary, weconsider it an important risk factor for cardiovascular health, asdemonstrated by worldwide population trials. Nevertheless, thepersons recruited in the current study and who smoked in thepast, maybe, had had good general conditions to allow themselvesto fall into the habit of smoking.

Consistently with the data reported by Leontyev et al. [8], we foundgood predictive force of the logistic EuroSCORE for late mortality.

Summarizing the survival data, one may observe that, in thepresent study, RAMT compared with FS allowed improved imme-diate outcomes in terms of morbidity, with faster recovery, withlower incidence of irreversible CVA, i.e. stroke, and did not com-promise survival rate: either the early or late one.

Frailty of the elderly patient might have influenced the differ-ence in the postoperative outcomes between the two approaches,previously observed in the non-selected population [12–14, 16].Post-discharge care requirement determined in the present studyby discharge destination was undoubtedly in favour of the MIAVRgroup (26.5% of survivors vs 44.3% in the FS group), suggestingthat a minimally invasive technique might have contributed to alower perioperative morbidity and a faster recovery, thus resultingin less need for continuous hospital care. Similar results, with abetter performance of MIAVR, were reported by Sharony et al.[21]: 53% in the MIAVR group were discharged home comparedwith 39% in the sternotomy group, but only 28% of patients inthat study were octogenarians. This is an important economicissue; however, an evaluation of cost-effectiveness was beyondthe scope and resources of the current study, but could be a prob-able point of address for future research.

From a system perspective, early resource utilization planning canoccur if we better understand this population’s predicted demand forhospital care beds and longer-term need for appropriate supportivecare, alternate level of care and rehabilitation or transition beds.

Study limitations

We acknowledge several limitations of our study. It is based on aretrospective analysis of our large, institutional, observational, pro-spectively collected database; thus, it reflects a single-centre ex-perience only and carries all the limitations that a retrospectiveanalysis design implies. The results are difficult to generalize asthis was a single-centre study. Propensity score analysis is simply amethod for reducing bias in observational studies and the match-ing was limited by the available variables.

We might have been unable to account for the influence of anyresidual unmeasured factors that could affect the adverse out-comes. However, chart review and data entry were performedaccording to prespecified definitions. Our internal institutional elec-tronic database contains EuroSCORE I-based records, instead ofthe newly introduced and more efficient and precise EuroSCORE II.As mentioned before, since 2005 there was an intrinsic biasfor surgical approach selection (preoperatively), favouring RAMTover FS. Consequently, year-by-year distribution of operatedpatients (Fig. 1) and follow-up duration in the two groups were noteven, as the MIAVR phase came later. The multivariable logistic re-gression model of stroke and in-hospital mortality predictors islimited by the low number of events, which could have affected theresult as shown by a wide CI.

Although the Cox regression model identified several inde-pendent predictors of late mortality, this analysis is also limited

by the low number of events associated with particularly lowprevalence variables (stroke history, infective endocarditis andpleural effusion requiring puncture; see Table 4), which couldhave affected the result as shown by a wide CI. We have utilizedall-cause mortality data (though reliably obtained from our data-base) rather than the more specific cardiac-related mortalities,and we did not address the relative incidence of non-fatal cardiac-related events. The series also encompasses our ‘learning curve’, asit includes our initial experience with MIAVR and, earlier, with FS,when the attending surgeons were still under training.Our database is lacking specific geriatric evaluation, e.g. frailty

and reduced mobility scores, which are important prognosticindices of early and late outcomes in senile patients.

CONCLUSIONS

MIAVR through RAMT can be safely performed in patients aged≥80 years with acceptable morbidity and early and late mortalityrates. It is an expeditious and effective alternative to FS AVR andmight be associated with lower postoperative stroke incidence,earlier extubation and shorter hospital stay.

Conflict of interest: Mattia Glauber and Marco Solinas disclose acommercial/financial relationship with the Sorin Group.

REFERENCES

[1] Christensen K, Doblhammer G, Rau R, Vaupel JW. Ageing populations: thechallenges ahead. Lancet 2009;374:1196–208.

[2] Mazzocchetti A, Caranci N, Addis A. Population ageing and health impli-cation. Thinking time trends in Emilia-Romagna Region. Recenti Prog Med2014;105:191–7.

[3] Vasques F, Messori A, Lucenteforte E, Biancari F. Immediate and lateoutcome of patients aged 80 years and older undergoing isolated aorticvalve replacement: a systematic review and meta-analysis of 48 studies.Am Heart J 2012;163:477–85.

[4] Stoica SC, Cafferty F, Kitcat J, Baskett RJ, Goddard M, Sharples LD et al.Octogenarians undergoing cardiac surgery outlive their peers: a case forearly referral. Heart 2006;92:503–6.

[5] Shan L, Saxena A, McMahon R, Wilson A, Newcomb A. A systematicreview on the quality of life benefits after aortic valve replacement in theelderly. J Thorac Cardiovasc Surg 2013;145:1173–89.

[6] Pretre R, Turina MI. Cardiac valve surgery in the octogenarian. Heart 2000;83:116–21.

[7] Kolh P, Kerzmann A, Honore C, Comte L, Limet R. Aortic valve surgery inoctogenarians: predictive factors for operative and long-term results. Eur JCardiothorac Surg 2007;31:600–6.

[8] Leontyev S, Walther T, Borger MA, Lehmann S, Funkat AK, Rastan A et al.Aortic valve replacement in octogenarians: utility of risk stratification withEuroSCORE. Ann Thorac Surg 2009;87:1440–5.

[9] Craver JM, Puskas JD, Weintraub WW, Shen Y, Guyton RA, Gott JP et al.601 octogenarians undergoing cardiac surgery: outcome and com-parison with younger age groups. Ann Thorac Surg 1999;67:1104–10.

[10] Alexander KP, Anstrom KJ, Muhlbaier LH, Grosswald RD, Smith PK, Jones RHet al. Outcomes of cardiac surgery in patients > or =80 years: results from theNational Cardiovascular Network. J Am Coll Cardiol 2000;35:731–8.

[11] Bakaeen FG, Chu D, Huh J, Carabello BA. Is an age of 80 years or greateran important predictor of short-term outcomes of isolated aortic valve re-placement in veterans? Ann Thorac Surg 2010;90:769–74.

[12] Ruttmann E, Gilhofer TS, Ulmer H, Chevtchik O, Kocher A, Schistek R et al.Propensity score-matched analysis of aortic valve replacement by mini-thoracotomy. J Heart Valve Dis 2010;19:606–14.

[13] Brinkman WT, Hoffman W, Dewey TM, Culica D, Prince SL, Herbert MA et al.Aortic valve replacement surgery: comparison of outcomes in matched ster-notomy and PORTACCESS groups. Ann Thorac Surg 2010;90:131–5.

[14] Lamelas J, Sarria A, Santana O, Pineda AM, Lamas GA. Outcomes of min-imally invasive valve surgery versus median sternotomy in patients age 75years or greater. Ann Thorac Surg 2011;91:79–84.

[15] Miceli A, Murzi M, Gilmanov D, Fugà R, Ferrarini M, Solinas M et al.Minimally invasive aortic valve replacement using right minithoracotomy

D. Gilmanov et al. / Interactive CardioVascular and Thoracic Surgery740

is associated with better outcomes than ministernotomy. J ThoracCardiovasc Surg 2014;148:133–7.

[16] Glauber M, Miceli A, Gilmanov D, Ferrarini M, Bevilacqua S, Farneti PAet al. Right anterior minithoracotomy versus conventional aortic valve re-placement: a propensity score matched study. J Thorac Cardiovasc Surg2013;145:1222–6.

[17] Glauber M, Farneti A, Solinas M, Karimov J. Aortic valve replacementthrough a right minithoracotomy. Multimed Man Cardiothorac Surg 2006;doi:10.1510/mmcts.2005.001826.

[18] Asimakopoulos G, Edwards M-B, Taylor KM. Aortic valve replacement inpatients 80 years of age and older. Survival and cause of death based on1100 cases: collective results from the UK Heart Valve Registry. Circulation1997;96:3403–8.

[19] Langanay T, Flécher E, Fouquet O, Ruggieri VG, De La Tour B, Félix C et al.Aortic valve replacement in the elderly: the real life. Ann Thorac Surg2012;93:70–8.

[20] Saxena A, Poh C, Dinh DT, Reid CM, Smith JA, Shardey GC et al. Early andlate outcomes after isolated aortic valve replacement in octogenarians: anAustralasian Society of Cardiac and Thoracic Surgeons Cardiac SurgeryDatabase Study. Eur J Cardiothorac Surg 2012;41:63–8.

[21] Sharony R, Grossi EA, Saunders PC, Schwartz CF, Ribakove GH, CullifordAT et al. Minimally invasive aortic valve surgery in the elderly: a case-control study. Circulation 2003;108(Suppl 1):II43–7.

[22] Altintas G, Diken AI, Hanedan O, Yurdakok O, Ozyalcin S, Kucuker SA et al.The Sorin Freedom SOLO stentless tissue valve: early outcomes after aorticvalve replacement. Tex Heart Inst J 2013;40:50–5.

[23] Biancari F, Martin M, Bordin G, Vettore E, Vinco G, Anttila V et al. Basicdata from 176 studies on the immediate outcome after aortic valve re-placement with or without coronary artery bypass surgery. J CardiothoracVasc Anesth 2014;28:1251–6.

[24] Tabata M, Umakanthan R, Cohn LH, Bolman RM III, Shekar PS, Chen FYet al. Early and late outcomes of 1000 minimally invasive aortic valveoperations. Eur J Cardiothorac Surg 2008;33:537–41.

[25] ElBardissi AW, Shekar P, Couper GS, Cohn LH. Minimally invasive aorticvalve replacement in octogenarian, high-risk, transcatheter aortic valveimplantation candidates. J Thorac Cardiovasc Surg 2011;141:328–35.

APPENDIX. CONFERENCE DISCUSSION

Scan to your mobile or go tohttp://www.oxfordjournals.org/page/6153/1to search for the presentation on the EACTS library

Dr C. Mestres (Abu Dhabi, United Arab Emirates): You showed that there weresome differences in hospital stay, but did this actually translate into some eco-nomic savings?

Dr Gilmanov: We have yet to perform any additional statistical analysisregarding the economic aspects and the relative hospital costs. We havedemonstrated that there was a shorter hospital stay in favour of direct homedischarge or dismissal to the rehabilitation facility, but possibly we will be ableto perform this analysis in the future to complete our work.

Dr S. Cicek (Istanbul, Turkey): I have two questions. My first question isregarding the incidence of postoperative stroke. Compared to the full sternot-omy group, it is much lower in your group, and this is in contrast to some otherstudies, especially because of the difficulties with deairing when using the mini-thoracotomy approach. What is your strategy for deairing? Do you use CO2?What kind of deairing procedures do you apply for the minithoracotomycases?

The second question, I might have missed it, but I didn’t see any data regard-ing the use of transfusion and blood products between the two groups. Haveyou observed decreased blood product usage in the minithoracotomy group?

Dr Gilmanov: As concerns the postoperative stroke incidence, actually wehad analysed only 4 cases in the full sternotomy group, and that is why we hadno statistical association with the preoperative risk profile in these patients.That’s why only postoperative lung artificial ventilation was associated withperioperative stroke incidence, as an inevitable consequence of cerebrovascu-lar accident. So I would accentuate the low number of events to analyse, only4 cases.

With regard to the technical differences between the right anteriorminithoracotomy and the full sternotomy group, I put up one slide de-scribing the technical aspects of our right minithoracotomy approach inthe intraoperative setting. In contrast to many other centres practicingminimally-invasive approaches for aortic valve replacement, we are con-vinced that central cannulation and central antegrade perfusion are more

physiological with respect to retrograde perfusion performed via the femoralgroin artery, and we are still doing so as in our previous years’ experience. So Ithink that at least a contribution of antegrade perfusion was an importantfactor to a lower incidence of perioperative stroke in the right minithoracot-omy group.With respect to the deairing procedure, we usually put both the ventricular

and aortic vents to optimize the deairing procedure. Usually we use CO2 floodin the operative field to reduce air microparticles in the heart chambers, and Ithink that our careful strategy to perform accurate cannulation and to do themost accurate purse-string placement are important to reduce the incidenceof stroke. It may be defined as a less-touch technique. We usually do thepurse-strings and carefully cannulate the ascending aorta with a low systemicpressure.Another important issue on the perioperative stroke regards the different

phases in our experience. Most of the patients submitted to the standard con-ventional approach to aortic valve replacement were operated 6, 7 years ago,from 2001 to 2008.Dr Mestres: I think this is a bad problem in methodology because there are

consecutive series and the matching was only one to one. We understand theproblem.Dr Gilmanov: Regarding the second question, the blood transfusion?

In previous studies, our centre has demonstrated that minimally-invasiveapproaches, both ministernotomy with right anterior thoracotomy as well asright minithoracotomy compared with full sternotomy, are important factorsto reduce blood transfusion, but we did not have sufficient data to includethem in the current study, that is why I did not present them. But we areconvinced that shorter incisions without excessive procedures within thethorax might contribute to lower blood loss and consequently to a reducedneed for blood transfusion.

eComment. The underestimated role of frailty in cardiac surgery

Authors: Daniel Hernandez-Vaquero, Rocio Diaz and Jacobo SilvaDepartment of Cardiac Surgery, Central University Hospital of Asturias, Oviedo, Spaindoi: 10.1093/icvts/ivv097

© The Author 2015. Published by Oxford University Press on behalf of the EuropeanAssociation for Cardio-Thoracic Surgery. All rights reserved.

We read with great interest and admiration the article by Gilmanov et al. [1] com-paring right anterior minithoracotomy with full sternotomy in octogenarians under-going aortic valve replacement (AVR). To do this, they developed a rigorouspropensity score analysis including 40 variables in the model (Table 1) [1]. However,and despite studying an octogenarian cohort, where frailty is prevalent, no measuresof frailty or ‘biological age’ (in spite of ‘chronological age’) were considered. Frailtyhas become a high-priority theme in cardiovascular medicine. Frailty from theFrench ‘frêle’, meaning of little resistance, is a biological syndrome that reflects a stateof decreased physiological reserve and vulnerability to stressors. Cardiac surgery isone of the biggest acute iatrogenic stressors to which the patient’s resiliency will de-termine their postoperative course [2]. The Frailty ABCs (Frailty Assessment BeforeCardiac Surgery) prospective study showed that slow 5-m gait speed was associatedwith a 3-fold increase in postoperative mortality or major morbidity [3]. Gait speed,as the best measure of frailty, contributed incremental value above the Society forThoracic Surgeons risk score (area under the curve 0.70 for risk score alone versusarea under the curve 0.74 for risk score plus gait speed). Patients with a slow gaitspeed and a high risk score had a 43% incidence of mortality/morbidity, whereasthose with a normal gait speed and a low to intermediate risk score had only a 6% in-cidence. It is evident that frail patients who undergo cardiac surgery have higherrates of postoperative mortality, morbidity, prolonged length of stay, and need fordischarge to facilities. Conversely, cardiac surgeons and cardiologists are still wonder-ing whether frail patients who undergo less invasive interventions have improvedoutcomes, although this is at times taken for granted [2]. We think that this study [1] isa failed opportunity to shed light on whether minimally invasive AVR is a smallerstressor than conventional AVR and this leads to better outcomes in frail patients.

Conflict of interest: none declared.

References

[1] Gilmanov D, Farneti PA, Ferrarini M, Santarelli F, Murzi M, Miceli A et al. Fullsternotomy versus right anterior minithoracotomy for isolated aortic valve re-placement in octogenarians: a propensity-matched study. Interact CardioVascThorac Surg 2015;20:732–42.

ORIG

INALARTICLE

D. Gilmanov et al. / Interactive CardioVascular and Thoracic Surgery 741