*2.3. Statistical Analysis*

Data were collected in excel sheets, and statistical analyses were performed using GraphPad Prism 7.0 (GraphPad Software, Inc., San Diego, CA, USA). For descriptive statistics of patients and flaps, the mean of continuous data accompanied by the standard deviation and the median or mode with interquartile ranges for ordinal data were reported. The paired two-sided Wilcoxon–Mann–Whitney-Test and the two-sided chi-square test were computed to assess differences in categorical and dichotomous variables, respectively. *p*-values of <0.05 were regarded statistically significant.

#### **3. Results**

Between January 2010 and March 2021, 46 patients underwent sternal defect reconstruction with free TFL flaps. Patients included 17 women (37%) and 29 men (63%), with a mean age of 67 ± 11 years (range: 38 to 85 years). Sternal osteomyelitis in all 46 patients was confirmed upon microbiology, clinical presentation, histology, and computed tomography. The defect etiologies were as follows: wound infection and sternum osteomyelitis after CABG with use of the left internal mammary artery (LIMA, *n* = 35; 76%), CABG with use of both IMAs (*n* = 1; 2%), valve replacement (VR, *n* = 5; 11%), or combined valve replacement and LIMA-CABG (*n* = 5, 11%). The median ASA classification was 3 (range: 2 to 4). Demographic data of patients, individual risk factors, chronic conditions, the El Oakley classification, as well as the results of microbiological examinations are summarized in Table 1.

**Table 1.** Patient demographics and comorbidities, El Oakley and Wright classification, and microbiological examination (SD = Standard Deviation).


**Table 1.** *Cont.*


A total of 16 patients (35%) underwent secondary reconstructions at our institution following failed prior attempts. Of these, 11 patients (*n* = 11; 24%) had previously undergone unsuccessful bilateral pedicled pectoralis major flaps at the referring cardiosurgical departments. Five patients, in detail, two VRAM and three bilateral pectoralis major flaps, developed partial flap losses requiring further free flap surgery at our institution (*n* = 5; 11%). These 16 patients underwent an average of 3 ± 2 debridements and negative pressure wound therapy cycles prior to free flap surgery. The mean operation time (OT) was 387 ± 120 min (range: 212 to 695 min), which included a mean flap ischemia time of <sup>63</sup> <sup>±</sup> 16 min (range: 32 to 91 min). While the mean sternal defect size was 194 <sup>±</sup> 43 cm<sup>2</sup> , the mean skin paddle surface of the TFL flap was 205 <sup>±</sup> 38 cm<sup>2</sup> , with a mean flap length of 24 ± 3 cm and a mean flap width of 8 ± 1 cm. Single-stage AVLs were necessary in 22 cases to provide reliable recipient vessels (Figure 1). Operative characteristics are presented in Table 2.



126

skin paddle surface of the TFL flap was 205 ± 38 cm<sup>2</sup>

**Figure 1.** A 62-year-old male patient with DSWI and sternal osteomyelitis after a bilateral CABG procedure. The resulting defect measured 22 × 8 cm (**A**). Sternal reconstruction with a TFL flap and an AVL was planned. The AVL was created between the subclavian artery and vein in an end-toside technique using a greater saphenous vein graft (**B**). Subsequently, arterial and venous end-toend anastomoses with the TFL flap pedicle were performed (**C**,**D**). The patient's recovery was uneventful, and he was discharged 13 days postoperatively. Stable defect reconstruction without any sign of wound healing disorder or recurrent infection three month after surgery (**E**). **Figure 1.** A 62-year-old male patient with DSWI and sternal osteomyelitis after a bilateral CABG procedure. The resulting defect measured 22 × 8 cm (**A**). Sternal reconstruction with a TFL flap and an AVL was planned. The AVL was created between the subclavian artery and vein in an end-to-side technique using a greater saphenous vein graft (**B**). Subsequently, arterial and venous end-to-end anastomoses with the TFL flap pedicle were performed (**C**,**D**). The patient's recovery was uneventful, and he was discharged 13 days postoperatively. Stable defect reconstruction without any sign of wound healing disorder or recurrent infection three month after surgery (**E**).

Chronic obstructive pulmonary disease (COPD) 20 (44%)

BMI (Body Mass Index) (kg/m<sup>2</sup>

Obesity (BMI > 30 kg/m<sup>2</sup>

El Oakley and Wright classification

Microbiological examination of soft and bony tissue

Chronic kidney disease (CKD) 25 (54%) Diabetes mellitus (DM) 30 (65%) Active smoker at time of surgery 13 (28%)

> I - II - IIIA 7 (15%) IIIB 11 (24%) IVA 2 (4%) IVB - V 26 (57%)

*Staphylococcus aureus* 17 (37%)

*Staphylococcus epidermidis* 14 (30%) *Enterococcus faecalis* 10 (22%) *Escherichia coli* 7 (15%)

A total of 16 patients (35%) underwent secondary reconstructions at our institution following failed prior attempts. Of these, 11 patients (*n* = 11; 24%) had previously undergone unsuccessful bilateral pedicled pectoralis major flaps at the referring cardiosurgical departments. Five patients, in detail, two VRAM and three bilateral pectoralis major flaps, developed partial flap losses requiring further free flap surgery at our institution (*n* = 5; 11%). These 16 patients underwent an average of 3 ± 2 debridements and negative pressure wound therapy cycles prior to free flap surgery. The mean operation time (OT) was 387 ± 120 min (range: 212 to 695 min), which included a mean flap ischemia time of 63 ±

, the mean

, with a mean flap length of 24 ± 3 cm and

Methicillin-resistant *Staphylococcus aureus* 6 (13%)

Multiresistant Gram-negative bacteria 9 (20%)

16 min (range: 32 to 91 min). While the mean sternal defect size was 194 ± 43 cm<sup>2</sup>

recipient vessels (Figure 1). Operative characteristics are presented in Table 2.

a mean flap width of 8 ± 1 cm. Single-stage AVLs were necessary in 22 cases to provide reliable

) 29 ± 6

) 19 (41%)

#### *3.1. Postoperative Complications*

The immediate postoperative course was uneventful in 38 of 46 patients (83%). Eight patients (17%) experienced acute microvascular complications or progredient hematoma at the recipient-site, which required emergency free flap re-exploration. In detail, acute microvascular compromise was observed during re-exploration in three cases (7%) in the form of acute arterial (*n* = 2; 4%) or venous thrombosis (*n* = 1; 2%). In this context, the use of single-stage AVLs did not increase the risk of microvascular thrombosis (*n* = 1/21 vs. *n* = 2/22; odds ratio: 0.50; 95% confidence interval: 0.03 to 4.6; *p* = 0.9). Postoperative hematoma evacuation was necessary in five patients (11%) within the first three days after flap surgery. All flaps could be salvaged. The further postoperative course was complicated in 6 patients (13%). In three patients, partial flap necroses of the most distal TFL parts were observed (7%). In these three patients, further secondary reconstructions with an intercostal anterior perforator flap (*n* = 1; 2%), a pedicled VRAM flap (*n* = 1; 2%), and two opposing local rotational flaps (*n* = 2; 4%) were performed. Wound dehiscence of the TFL flap with the need for debridement and secondary split-thickness skin-grafting (SSTG) was necessary in three patients (7%). Surgical donor-site complications occurred in five patients (11%), including three cases of impaired wound healing (7%), one case of donor-site infection (2%), and one case with the need for hematoma evacuation (2%). Of these patients, two received SSTG (4%) and three underwent successful secondary wound closure (7%). Eventually, all donor-sites healed satisfactorily. Table 3 shows the distribution of surgical complications. The average hospital stay was 34 ± 12 days, with a mean length of postoperative hospital stay of 23 ± 14 days. Postoperatively, 37 patients (80%) were monitored at the intensive care unit (ICU) for an average of 6 ± 9 days. Postoperative medical complications were seen in six patients (13%). These included: paralytic ileus (*n* = 1; 2%), postoperative delirium (*n* = 1; 2%), and cardiovascular instability with severe hypotension (*n* = 2; 4%). Two patients (4%) died due to respiratory failure and cardiovascular instability within the first 30 days post-surgery. The 1-year mortality rate was 17% (*n* = 8), but these deaths were not related to flap surgery. At the time of discharge, all successfully treated patients presented with stable soft-tissue conditions and without a sign of recurrent sternal infection.

### *3.2. Follow-Up Examinations*

Follow-up donor-site examinations were performed 34 ± 19 months after free flap surgery, including a total of 28 patients (61% follow-up rate). From the initial cohort of 46 patients, 8 patients (17%) had died, and 10 patients (22%) could not be reached. At the TFL donor-sites, the ROM of hip and knee joints revealed no restrictions when compared to the contralateral healthy sides (Table 4). Three patients subjectively described a weakness

in knee extension (*n* = 3; 11%); however, muscle strength was not notably impaired in any patient (donor-site: median of 5, range of 4 to 5 vs. healthy-site: median of 5, range of 4 to 5; *p* = 0.25). Herniation of the quadriceps muscle was not seen in any patient at follow-up. The patient-reported satisfaction showed an overall good result for both functional and aesthetic outcomes at the donor-site (function: median of 2, range of 1 to 4; aesthetic: median of 2, range of 1 to 4). With respect to general health and satisfaction, the SF-36 questionnaire (physical component summary, median of 44, range: 33 to 57; mental component summary, median of 29, range: 19 to 40) as well as the LEFS (median of 54, range: 35 to 65) revealed satisfactory results. The mean donor-site scar length was 24.8 ± 3.3 cm. Scar examinations revealed a median VSS score of 3 (range: 2 to 7) at the donor-site and a median of 3 (range: 2 to 5) at the corresponding recipient-site. In accordance, the POSAS showed an overall good satisfaction with scar quality and scar appearance at the donor-site (median: 24, range: 18 to 34) and recipient-site (median: 23, range: 19 to 31) with comparable results for the patients' and observers' scale (Table 4).

**Table 3.** Postoperative complications.


**Table 4.** Follow-up examination (VSS = Vancouver Scar Scale, POSAS = Patient and Observer Scar Assessment Scale, SD = Standard Deviation).


#### **4. Discussion**

In the present study, we report on our treatment of a selective group of patients with DSWI and sternal osteomyelitis after cardiac surgery resulting in large sternal defects. The patients in this study had multiple previous surgeries (El Oakley III to V), with their overall morbidity leading to a median ASA score of 3. Nowadays, microsurgical free flap transfer is a safe and reliable procedure, with failure rates ranging between 1 and 6% [17–19]. However, it is technically complex, time-consuming, and often debilitating on multimorbid patients due to their diminished physical reserves [20–22]. A particularly challenging situation arises when free flaps become the only remaining reconstructive option because defects cannot be closed with local or pedicled flaps. Nevertheless, a persisting disfiguring,

painful, and infected defect is hardly ever an alternative for the patient's remaining life span. In order to avoid additional complications, any reconstructive procedure in these patients needs to be as safe and reliable as possible. Hereby, operative time should be kept as short as possible, with the designated flap being "easily accessible". Pursuing the objective of optimizing the treatment of these critically ill patients, we have increasingly been choosing the free TFL flap with its large skin paddle, which makes it ideal for the reconstruction of extensive three-dimensional sternal defects.

The TFL muscle is a weak flexor and lateral rotator of the thigh. The muscle is dispensable, and its absence usually causes no remarkable functional deficit or donor-site morbidity [9]. This was confirmed by the results of our clinical follow-up examinations. Hereby, muscle strength and range of motion were not considerably impaired. However, our method of muscle strength assessment may not have been precise enough to discriminate between the donor-site and uninjured side. The LEFS, a well-established and sensitive tool to measure lower extremity functional impairment [14], also did not reveal considerable impairment in any patient (median LEFS = 54) when compared to normative median values (LEFS = 66) for patients older than 65 [23]. Donor-site morbidity in this study was negligible when compared to other flaps, such as the rectus abdominis muscle flap, for example [24,25]. The TFL flap can comprise a maximal skin paddle three times the size of the muscle [26]. Therefore, it combines the freedom of abundant skin coverage and a strong fascial layer with the versatility of a microvascular pedicle. The fascia is a highly vascularized semirigid layer, which provides additional structural support to cover large defects [9,27]. The TFL flap is easily accessible and can be harvested in a supine position, thus eliminating the need for lengthy intraoperative repositioning, when compared to flaps from the thoracodorsal system. The operating time can be kept short by working in a two-team approach. Another advantage of the large muscular TFL flap is its suitability for anastomoses to AVLs, as opposed to fasciocutaneous perforator flaps, such as the anterior lateral thigh (ALT) flap. In this context, Henn and colleagues stated that ALT flaps in combination with AVLs might be prone to microvascular complications due to an elevated flow resistance of the small-caliber perforator in comparison to the low-resistance conditions of the vein graft used in AVLs [28]. Because of these findings, we refrain from using fasciocutaneous perforator flaps in combination with AVLs and only use muscle-based flaps in these scenarios. In line with this notion, the combination of TFL flaps with AVLs showed no increased risk of microvascular thrombosis in this study.

Certainly, we want to point out that, in general, the full armamentarium of reconstructive procedures is required to cover sternal defects. Hereby, the most prevalent reconstructive options usually comprise pedicled muscle flaps, as they provide well-vascularized tissue with enough bulk to fill the defect cavity. The pedicled VRAM flap, LD flap, and bilateral pectoralis major flap have been the method of choice for decades [5,6,8]. In this context, it is recommended to cover cranial sternal wounds with pectoralis major flaps, whereas VRAM flaps are of better use to cover caudal sternal wounds. Davison and colleagues compared the outcome of 41 modified pectoralis major flaps against 56 pedicled VRAM flaps and reported equal wound dehiscence rates (VRAM: 14.2% vs. PM: 14.6%) [4]. However, it has to be considered that in the majority of patients developing DSWI after cardiovascular surgery, the LIMA is harvested for CABG. Therefore only the RIMA can be used for a cranially pedicled VRAM flap [8]. Furthermore, the most distal part of the pedicled flap, which is of paramount importance to the reconstructive outcome, especially in larger defects, is at risk of impaired perfusion, with a higher risk for wound dehiscence and infection [29]. We agree with Davison and colleagues that both pedicled flaps are straight forward and easy to harvest; nevertheless, the free TFL flap showed lower rates of wound dehiscence and partial flap necrosis in this study.

The latissimus dorsi muscle is the largest human muscle and is ideal to cover larger sternal defects. Even when the insertion to the humeral bone is detached, the maximum arc of rotation of a pedicled LD flap is an important limitation, particularly when a midline exceeding defect must be reconstructed [29]. This can put the most distal part of the flap

at risk of impaired perfusion. Spindler and colleagues published a study of 106 patients undergoing sternal reconstruction with a pedicled myocutaneous LD flap. Besides no total flap loss, they reported 35% revision surgeries because of wound healing disorders, hematoma, or persistent infection [30]. In comparison to their results, the number of revision surgeries for partial flap necrosis (*n* = 3), wound healing disorders (*n* = 3), and hematomas (*n* = 5) was lower in our cohort (24%). According to the current literature, the free LD transfer can show reliable results, with encouragingly low rates of revision surgeries or serious complications [31]. However, raising a free LD flap to cover sternal defects has some disadvantages that need to be considered. First, patients must be repositioned intraoperatively, and the flap must be banked in the axilla. Second, the latissimus dorsi muscle, as the rectus abdominis muscle, is an auxiliary breathing muscle; thus, sacrificing this muscle can affect breathing mechanisms in these already multimorbid patients [30]. While some authors state that the greater omentum (OM) flap is a valuable alternative to muscle flaps, we do not consider the OM flap as a first line procedure [32]. Kolbenschlag and colleagues presented a study of 50 patients undergoing sternal defect reconstruction with a pedicled OM flap and reported high surgical complication rates and high donor-site morbidity, with one patient even developing acute intestinal incarceration [33]. Therefore, the OM flap should be considered a last resort backup option rather than a first-line treatment. Furthermore it should be considered than an increasing defect size can be related to a higher incidence of partial flap necrosis of pedicled flaps, leading to a higher rate of revisional surgeries and impaired postoperative recovery [34]. In this context we can recommend the myocutaneous TFL flap as a workhorse flap for extensive sternal defect reconstruction.

Despite the promising nature of our data, which highlight the feasibility of the free TFL flap for the reconstruction of large sternal defects in a one-stage procedure, our study has important limitations that need to be discussed. First, our study is limited by its retrospective nature and involvement of several different surgeons and their various flap planning routines. Second, the study comprises a relatively small number of patients. Therefore, we conclude that a larger cohort of patients and a longer follow-up period are necessary to gain more reliable data regarding the choice of the reconstructive approach in this context. Third, the follow-up response rate was low, and follow-up times varied greatly, which may have resulted in different stages of rehabilitation, wound healing, and scarring assessed.

#### **5. Conclusions**

In conclusion, the free TFL flap represents a valuable option for sternal reconstruction in critically ill patients with large defects and a history of previously failed reconstructive procedures. It is associated with an encouragingly low morbidity at the corresponding donor-site. We therefore regard the free TFL flap, in combination with AVLs if needed, as a workhorse flap for sternal reconstruction, rather than merely a backup option.

**Author Contributions:** A.K.B., F.F., B.T. and C.A.R. designed the study. F.F. carried out data acquisition, performed the statistical analysis and data interpretation, and drafted the first version of the manuscript. G.H., S.A.M. and E.-M.R. supported the manuscript writing. L.H., E.G., U.K. and C.A.R. helped in data interpretation and manuscript revision. A.K.B. and C.A.R. further participated in the conception of the study, interpretation of data, as well as manuscript preparation. All authors have read and agreed to the published version of the manuscript.

**Funding:** No funding was received for this article. None of the authors has a financial interest in any of the products, devices, or drugs mentioned in this manuscript.

**Institutional Review Board Statement:** This study was approved by the local ethics committee (Mainz, Germany) under the IRB approval reference number 2021-15577. The need for consent was deemed unnecessary according to German national regulations (ethical committee of Rhineland palatinate, Germany). The study has been performed in accordance with the guidelines and regulations of the Declaration of Helsinki.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

**Conflicts of Interest:** The authors declare no conflict of interest.
