Inuence Factors of Sternal Healing and Comparison of Different Sternal Closure Methods(cid:0)A Retrospective Analysis

Background We determined the factors affecting sternum healing after a median sternotomy based on a retrospective analysis and compared the stability of different sternal closure techniques. Methods We collected information involving patients who underwent a median sternotomy in Changzhou First Hospital from 2014–2019 and had chest CT examinations 1–24 months postoperative, Main outcome includes the average of sternal healing score at the ve specic anatomic levels and the adverse event of transverse displacement of sternum.


Abstract Background
We determined the factors affecting sternum healing after a median sternotomy based on a retrospective analysis and compared the stability of different sternal closure techniques.

Methods
We collected information involving patients who underwent a median sternotomy in Changzhou First Hospital from 2014-2019 and had chest CT examinations 1-24 months postoperative, Main outcome includes the average of sternal healing score at the ve speci c anatomic levels and the adverse event of transverse displacement of sternum.

Conclusion
In summary, advanced age and internal mammary artery interception are risk factors that affect sternal healing based on imaging ndings. this new method of sternal closure provides an effective way of ensuring sternal stability of both sternal plates and reduces the risk for complication after cardiac surgery in high-risk patients.

Background
Median sternotomy is one of the most common incision methods in heart surgery [1]. It provides excellent exposure of the heart,which is helpful for surgeons to operate [2]. Currently, the main focus of research involves the complications from median sternotomy, while sternum healing has not attracted widespread attention. Unlike the signi cant morbidity and mortality of postoperative sternal complications [3][4][5], poor sternal healing is more common in elderly patients [6], but most cardiothoracic surgeons overlook this problem because of its low risk. The morbidity of chronic pain after cardiac surgery, which has an adverse impact on health-related quality of life, is up to 56% [7]. A prospective study conducted by H. K. Gjeilo[8] showed that 3.8% patients had chronic pain after 5 years. Prophase of sternal healing is related to sternal complications. One of the cause of postoperative deep sternal wound infections is malabsorption of sternal hematoma which leads to failure to form an effective dead space [9][10].
Through the analysis of the factors affecting the sternum healing, it is helpful to take appropriate surgical methods and postoperative care that accelerate the prophase of sternal healing and reduce the incidence of complications for the high-risk groups with poor sternum healing. The purpose of the study was to investigate determinants of sternal healing and estimate the effect of titanium plate compared to wire cerclage on sternal motion after cardiac surgery.

Study design
This retrospective study enrolled 122 patients who were examined with chest CT one to twenty-four months after cardiac surgery at the Third a liated Hospital of soochow University, Changzhou, China.The chest CT imaging included the whole sternum allows us to evaluate sternal healing after operation. Inclusion criteria included patients aged 18-80 years undergoing elective cardiac surgery via a median sternotomy. Exclusion criteria included severe chronic obstructive pulmonary disease, history of taking immunosuppressive drug or autoimmune disease, history of osteoporosis and bone dysplasia, history of second thoracotomy, hemodialysis and placement of an aortic balloon regurgitation device during perioperative, history of renal insu ciency, complications from median sternotomy. All patients reexamined by chest CT scans one week after cardiac surgery and sternal was reducted well.
The baseline characteristics of patients were as follows: gender, EF, history of hypertension, history of pulmonary hypertension, internal mammary artery injury, a history of diabetes, body mass index (BMI), age, operative time, postoperative time, xation method, the cortical thickness (the average thickness of the anterior and posterior sternal cortex at the midline of the ve speci c anatomic levels), the thickness of the sternum (the average thickness of the sternum at the midline of the ve speci c anatomic levels).
Main outcome included the average of sternal healing score at the ve speci c anatomic levels and the adverse event of transverse displacement of sternum.

Surgical methods
The sternum was closed by one of the two following methods: titanium plates (Sternal Fixation System; Was-ton Medical Appliance Co. Ltd.), in which four plates are used in the rst, second, third, and fourth or fth intercostal spaces; and closure of the sternum with 5-7 metal wires (Surgical Stainless Steel Suture; Ethicon, LLC) as a single interrupted suture or a gure-of-eight suture. On the method of closure, the patients were divided into three groups(sternum plate groups, wire groups and mixed xation groups).
The surface of sternum was smeared with nonabsorbable bone wax to stop bleeding. A few patients used a new xation method. One half of the titinium plate was placed on one side only where the region had been xed with a 8-shaped steel wire, and then, the titinium plate around the 8-shaped steel wire allowing a rm approximation of the two sternal halves, followed by placement of the other half of the plate. The con guration of the plate around the wire prevents wire cutting through the bone and provides 360-degree stabilization. After all plates and wires were secured, the soft tissue was closed with 2 − 0 Polybutylate coated braided Polyester Suture and the skin was closed with 4 − 0 Polybutylate coated braided Polyester Suture (ETHIBOND EXCEL, Shanghai, China).

Outcome measures
The endpoint of the study was sternal healing which determined by independend radiologists using CT scans and a validated evaluation method. Five speci c anatomic levels (Manubrium, Top of aortic arch, Aortopulmonary window, Main pulmonary artery, Aortic root)were selected by two different radiologists with senior title for evaluation. Radiologists scored independently each location using a 6-point scale(0: no sign of healing, 1: minimal healing, 2: mild healing, 3: moderate healing, 4: partial synthesis, 5: complete synthesis). Sternal union was de ned as a mean score of ≥ 3.After kapper-test,the average value of scores from two radiologists was taken as the nal result. Cortical thickness and sternal thickness were also measured manually by the two radiologists through CT scan results.

Statistical analysis
In order to assess the risk factors for poor sternal healing, we divided the patients into two groups: a short term healing group (the S group) and a long term healing group (the L group). Multiple linear regression were performed to identify risk factors for poor sternal healing at the S group and the L group.
Continuous data were presented as mean ± standard deviation and evaluated with t-test. Categorical data, summarized as a number(%), were evaluated with Chi square test and Fisher test. Except that a signi cance level of 0.1 was used in the univariate linear regression analysis to select the variables for the multiple linear regression, p-values less than 0.05 were considered to indicate statistical signi cance. We made an propensity score matching (PSM) [11] to control bias by "MatchIt" R package [12]. The age was the adjusted variable used to PSM for patients by a nearest methods with a case-control ratio of 1:3 ( Fig. 1). All statistical analyses were performed by R software (version 3.6.3).

Results
Participants and characteristics 122 patients were enrolled, including 77 males and 45 females, with an average age of 57.83 ± 11.80 years. The mean BMI was 61.6 ± 11.3 kg. 64 patients underwent Off pump coronary artery bypass grafting. Isolated valve procedures was performed in 58 patients. The mean CT examination time after surgery was 12.53 ± 8.43 months. The two radiologists determined the healing score based on the sternal CT scan (Fig. 2). The results were analyzed by a Kapper-test. The results showed that the scoring method was reliable and repeatable (Table 1). There was a signi cant bias in the research data because surgeons always choose an appropriate sternal closure method according to the speci c situation of patients (Table 2). To control bias, we made an propensity score matching (PSM) and 102 patients were selected for the subsequent analyses. Table 3 displayed a summary of the demographic and clinical characteristics of those patients after PSM.   Factors in uencing sternal healing Figure 3 showed the rate of good sternal healing over time from 1 to 24 months. The average total sternal healing score in the S gropu was 2.56 ± 1.52. 56% patients showed good healing. In the L group, the mean total sternal healing score was 2.63 ± 1.12 and good healing was observed in 62.5% patients. Poor healing was more frequently found in the lower sternum than others. Table 4 showed the results of the analyses of the in uence factors for sternal healing score. In the S group, Sternal score was only correlated with postoperative time (HR = 0.18, 95%CI:0.135-0.225, p < 0.001). In the L group, old people had a higher risk of poor sternal healing than young (Age, HR=-0.028,95%CI:-0.05-0.006, p = 0.013).
Patients with left internal mammary artery grafting (LIMAG) had a high risk of poor sternal healing (HR=-0.444, CI:-0.869-0.019, p = 0.045). It should be noted that BMI was not statistically signi cant. The Q-Q plot was uesd to verify the normality and linear relationship of linear regression (Fig. 4).

Transverse displacement
In this study, we found transverse separation of sternum in 17 patients. This problem existed in 16 patients with titanium plates xation and 1 patients with steel wire xation, while there was no one with mixed xation. The transverse displacement mainly occurred in the lower sternum (The levels of main pulmonary artery and aortic root). Only one patient's sternum had a score of 2 at the level of transverse deviation, and the rest had a score of 0. Figure 5 showed the model of sternum after median sternotomy. The case group included patients closed sternum with steel wire xation along and the patients with mixed xation. The patients who used titinium plates xation were selected as control group. By binary logistic regression, It could be found that steel wire xation was a favorable factor to prevent the transverse displacement of sternum (HR = 0.122, 95%CI:0.007-0.651, p = 0.047).

Discussion
Bone healing is mainly affected by macroscopic factors (biomechanics and blood supply) as well as microscopic factors (such as molecular biology) [13]. BitKover found that there was no sternal healing 3 months after median sternotomy in a prospective computed tomography scan study [14]. Advanced age as an independent risk factor is not only related to sternal postoperative complications, but also affects sternum healing. Studies have con rmed the importance of immune mechanisms and in ammatory response in bone healing [15]. The cells that is important in in ammatory responses (macrophages, T cells, and mesenchymal stem cells) are related to age. The number and activity of osteochondrocytes and their progenitor cells in the bone marrow of the elderly are lower than young. In addition, advanced age is a high-risk factor for vascular diseases which is closely related to the ability of blood vessels to transport blood ow and it leads to poor blood perfusion of sternum. The choice of bypass vessels is important during coronary artery bypass surgery. Puskas[16] con rmed that transplantation of the internal mammary artery as a bypass vessel signi cantly improved the long-term survival rate. In clinical practice, almost all cardiothoracic surgeon choose the internal mammary artery as the rst choice for transplantation. Interception of the internal mammary artery will not only increase the chance of sternal complications, such as mediastinal infection and sternal opening, but also affect sternum healing [17]. BMI > 35kg/m² as an independent risk factor in uencing the score of sternum have been reported in previous literature [18], but the BMI of Asians is generally lower than that of Europeans [19]. In this study, only ve patients had a BMI greater than 30kg/m². All of them were xed with sternum plate, so this study failed to get positive results related to BMI. During follow-up we found that a small number of patients with dysplasia of the lower sternal segment had di culty in achieving sternal healing based on imaging scan after surgery. The cortical bone and the thickness of the sternum was used as an indicator to measure sternum development. We found that whether the thickness of the sternum or the cortical bone thickness was not related to sternal healing score.
Several forces such as breathing and coughing act on the sternum and load the sternum through a combination of lateral shear and transverse shear. Losanoff et al. [20] con rmed that the lateral tension of the thorax is mainly concentrated in the lower part of the sternum. It is due to the con uence of multiple ribs in the lower sternum and the greater mobility of the chest wall in the area during breath and exercise [21]. Therefore, healing of the lower sternum is poor. Minimal anteroposterior movement of the sternal halves will lead to the result that the cortical bone on one side of the sternum enter the cancellous bone of the other side of sternum. That can affect bone fusion even without frank dissociation [22]. Therefore, the method of sternal closure seems to be important to long-term sternal healing. Mechanical studies demonstrated that rigid plate xation of the sternum results in superior mechanical properties compared with wire xation [23]. COL David J. Cohen[24] compared the biomechanics of different sternal closure techniques and found that the plates were stiffer than the gure-of-eight wire constructs in the transverse shear direction.
The main reasons for the differences between the results of this study and the conclusion of COL David J. Cohen are as follows: The sternum plates used in the two experiments were different. Failure of the wire system usually involves the wire cutting into the bone under loads. Within the maximum shear force that steel wire can bear, the stability of eight-shaped steel wire xation sternum may be stronger than that of sternum plate. The vitro experiment was di cult to imitate the complex biomechanics of human body completely by the force of simple direction.
Titinium plates provide more stability than wire cerclage alone, but it lack the posterior sternal stabilization. 360-degree rigid sternal xation with combination of plates and wire cerclage described here has been performed in 2 patients to date, with good sternal healing (sternal healing score ≥ 3). Taylor M. James once proposed a similar method to x the sternum and this method had been employed in 40 patients, with a zero incidence of deep sternal wound infection [25]. Wire cerclage along with rigid sternal plates that helps stabilize the sternum not only in the posterior plane but also in the lateral plane may reduce the amount of mechanical stress put on the plate. Owing to the complication of this method, we recommend considering this technique for especially high-risk patient who are old people, or those with morbid obesity. According to biomechanics, we can use this method to x the lower part of the sternum, while the upper part of the sternum could be xed with titanium plate or eight-shape wire alone (Fig. 5).
The limitations of this study inherent was a retrospective review. There was obvious selection bias regarding whether a patient received titanium plate reinforcement, because this technique was usually used in old people,or those with morbidly obesity. We attempted to control for this bias by PSM. Lastly, because of the small number of patients xed sternum with this new method, our ability to examine its stiff was limited.
In summary, advanced age and internal mammary artery interception are risk factors that affect sternal healing based on imaging ndings. this new method of sternal closure provides an effective way of   Flow chart for the data screening Rate of good sternal healing over time from 1-24 months Figure 4 If the two distributions are similar, the Q-Q diagram tends to fall on y = X-ray. If the two distributions are linearly correlated, the points tend to fall on a straight line on the Q-Q graph.