Clinical consequences of head and neck free-flap reconstructions in the DM population

Diabetes mellitus (DM) is a common comorbidity and risk factor for postoperative complications in head and neck (H&N) microsurgical reconstructions. Our study focused on the association between DM and individual complications regarding both surgical and medical aspects. A meta-analysis of English-language articles comparing a series of complications between DM and non-DM H&N free-flap recipients was performed by comprehensive meta-analysis (CMA). Twenty-seven articles presented 14,233 H&N free-flap reconstructions, and a subset of 2329 analyses including diabetic cases was included for final analysis. Total postoperative (RR = 1.194, p < 0.001; OR = 1.506, p = 0.030) and surgical (RR = 1.550, p = 0.001; OR = 3.362, p < 0.001) complications were increased in DM subjects. Free-flap failure/necrosis (RR = 1.577, p = 0.001; OR = 1.999, p = 0.001) and surgical site infections (OR = 2.414, p < 0.001) were also increased in diabetic recipients. However, return to the operating room, dehiscence, fistulas, plate exposures, readmissions, and mortalities were not increased in DM patients. DM increased various complications in H&N free-flap reconstructions. Surgical indications should be cautiously evaluated, and aggressive treatments should be implemented for high-risk recipients.

Free-flap reconstruction following head and neck (H&N) tumor ablation has become a routine practice with advancements in microsurgical techniques 1 . However, daunting postoperative complications after free-flap reconstruction, such as deprivation of eating, speaking, and breathing functions, have been reported 2 . Numerous practices have been introduced to improve surgical outcomes after free-flap reconstruction, especially risk factor evaluation 3 . Among those factors associated with postoperative complications, diabetes mellitus (DM) is a very important risk factor in clinical practice. However, there have been differing reports about free-flap success rates, surgical site infections, and surgical outcomes in DM patients in previous studies [4][5][6][7] .
A lack of persuasive and clinical evidence impedes determining whether DM is contributory to complications following major microsurgical procedures. As tissues left after H&N tumor resections are often scarce, free-flap reconstruction is frequently used for instant coverage and functional restorations. Serious postoperative complications after free-flap reconstruction not only psychologically devastate patients with physical appearance or loss of basic function but also may delay crucial radiotherapy 4,5 . In those cases, the quality of life after the operation is not satisfactory due to repeated readmissions or revision operations [8][9][10] . Therefore, recognizing the risk of DM in association with free-flap reconstruction and the response to corresponding treatments before free-flap reconstruction are very important in real-world practice. To define the possible effects of DM in association with free-flap use for H&N reconstruction, we performed this meta-analysis to examine these associated complications in more detail.
Statistical analysis. The meta-analysis was conducted using Comprehensive Meta-Analysis (CMA) 2.0 Software, [Biostat, New Jersey, USA], using the inverse variance method for pooled relative risk. The dichotomous data were summarized using relative risk (RR) or odds ratio (OR) separately and with 95% confidence intervals (95% CIs). Statistical heterogeneity tests of Cochran's Q-value, I 2 , and Tau 2 were performed to determine whether a fixed or random effect model was adopted.

Results
Process outcomes. The search strategy identified 63 eligible studies, and 27 studies were ultimately included in the meta-analysis 5, . The selection process is demonstrated in the flow diagram ( Fig. 1), and the study characteristics are presented in Table 1. The funnel plot of all studies included is presented in Fig. 2.

Analysis of complications. Total complications.
A total of 6 studies were included for overall postoperative, surgical and other complications. A total of 3924 patients were enrolled, 466 of whom had DM. The pooled results were 6 studies with RR and 3 studies with OR. After multivariable analysis, DM significantly increased the total complication rate (RR = 1.194, 95% CI 1.089-1.310, p < 0.001 and OR = 1.506, 95% CI 1.040-2.181, p = 0.030) (Fig. 3).
Total and individual surgical complications. Total surgical complications. Ten studies that evaluated total surgical complications were identified; a total of 5064 patients were enrolled, 744 of whom were DM patients. The pooled results include 9 studies with RR and 3 studies with OR. The multivariate analysis results showed that DM significantly increased surgical complication rates (RR = 1.550, 95% CI 1.195-2.010, p = 0.001 and OR = 3.362, 95% CI 1.961-5.763, p < 0.001) (Fig. 3).
Return to the operating room. Five studies evaluating the rates of "return to the operating room" were included, with 413 DM patients among 2976 total subjects. All complication rates were expressed as RR. The pooled results showed that DM did not significantly increase the rate of "return to the operating room" (RR = 1.415, 95% CI 0.760-2.633, p = 0.273) (Fig. 4).
Flap failure/necrosis. Ten studies that evaluated flap failure or necrosis were included, with 9144 total patients and 1550 DM patients. The pooled results included 9 studies with RR and 3 studies with OR. The multivariate analysis results showed that DM significantly increased flap failure or necrosis (RR = 1.577, 95% CI 1.217-2.043, p = 0.001 and OR = 1.999, 95% CI 1.347-2.967, p = 0.001) (Fig. 4) Fistula. Two studies were included to evaluate fistula formation after surgery, with 63 DM patients among 723 subjects. All complication rates were expressed as RR. The pooled results showed that DM did not significantly increase fistula formation after the operation (RR = 4.323, 95% CI 0.172-108.961, p = 0.374) (Fig. 5).
Plate exposure. Two studies were included for the evaluation of plate exposures, with 67 DM patients among 665 subjects. All complication rates were expressed as RR. The pooled results showed that DM did not increase plate exposure rates (RR = 0.944, 95% CI 0.374-2.387, p = 0.904) (Fig. 5).
Hematoma. Four studies were included for the evaluation of hematoma, with 176 DM patients among 1150 subjects. All complication rates were expressed as RR. The pooled results showed that DM did not significantly increase the rates of hematoma (RR = 3.338, 95% CI 0.742-15.025, p = 0.116) (Fig. 5).
Coagulation-related complications. Regarding coagulation-related complications (e.g., hematoma, bleeding, required transfusion, and thrombosis), seven studies were included in the meta-analysis. The complication rates were expressed as RR; there were a total of 3849 patients, 562 of whom were DM patients. The pooled results showed that there was a trend of increasing coagulation-related complications in the DM group (RR = 1.708, 95% CI 0.918-3.178, p = 0.091) (Fig. 5).
Quality aspects. Readmissions. Two studies were included to evaluate the rates of readmission, with 340 DM patients among 2702 subjects. All complication rates were expressed as RR. The pooled results showed that DM did not significantly increase readmission rates (RR = 1.211, 95% CI 0.870-1.686, p = 0.256) (Fig. 6).  www.nature.com/scientificreports/ Mortality. Two studies were included to evaluate mortality, with 291 DM patients among 2610 subjects. All complication rates were expressed as RR. The pooled results showed that DM did not significantly increase mortality rates (RR = 1.999, 95% CI 0.347-11.530, p = 0.438) (Fig. 6).
Other complications. Two studies were included in the evaluation of other complications (e.g., myocardial infarction, cerebrovascular event, deep vein thrombosis, pulmonary embolism, pneumonia, urinary tract infection, and septic shock); a total of 2702 patients were enrolled, including 340 DM patients. All complication rates were expressed as RR. The pooled results showed that DM significantly increased the risk of other complications (RR = 1.284, 95% CI 1.109-1.487, p = 0.001) (Fig. 6).

Discussion
DM has been investigated for potentially increasing the risk of postoperative complications, such as vascular occlusions and thrombosis, that jeopardize free flap survival and result in immune disturbances associated with poor wound healing [8][9][10] . However, retrospective cohort studies from different institutions and even different time periods within the same database could end up with contradicting conclusions 4,5 .
Previous meta-analyses exploring the association between DM and H&N free-flap reconstructions had a few limitations. In 2015, Rosado et al. noted that DM was significantly related to more surgical complications following free-flap reconstructions in the H&N region 38 . Consisting of merely five studies, this analysis seemed relatively short of conclusiveness 38 . Further investigations were performed by Cupato et al. 39 ; these authors  www.nature.com/scientificreports/ gathered 16 studies for meta-analysis. They concluded that DM significantly increased flap failures and local region complications 39 . This complication analysis was performed to include all major H&N surgeries instead of limiting the analysis to free-flap reconstructions 39 . They also pooled different complications under the term "locoregional" 39 . None of these meta-analyses was explicitly designed to analyze individual complications. Additionally, neither one considered crucial complications nor quality indicators in their studies. For surgeons, a better understanding of which complication was at higher risk might be more helpful and provide insights into routine or pragmatic formulations in clinical practice. In this study, a total of 27 studies regarding H&N free-flap reconstructions in patients with DM were included in a meta-analysis in response to preceding studies with different opinions about free-flap failure or abnormal healing of the anastomoses in DM subjects 5,12,40,41 . Some studies concluded that patients with DMS were at increased risk and others did not 5,12,40,41 . Overall, we found that DM was associated with increasing total complications (surgical and other clinical sequelae), return to the operating room, and free-flap failure after the operation (Figs. 3 and 4). The underlying mechanisms might be due to the high prevalence of peripheral vascular diseases in DM patients 8 . Peripheral vascular diseases can result in fragile vessel conditions and precipitate free-flap failure. Currently, vascular mapping for evaluating vascular viability and anatomy by color Doppler, computed tomography (CT) angiography or magnetic resonance imaging (MRI) angiography might improve microsurgical outcomes 42 . Therefore, adopting angiography as part of the preoperative assessments might be beneficial and suggested for routine used in DM patients referred for free-flap operations.
DM has been demonstrated to alter the immune system with cytokine effects on local wound healing 10 . Higher infection rates in DM patients have been previously reported from various types of surgical procedures 6 . In our www.nature.com/scientificreports/ study, surgical site infection after free-flap reconstruction was significantly increased in DM patients (RR = 1.401, p = 0.059; OR = 2.414, p < 0.001) (Fig. 4). Therefore, hyperglycemia should be taken seriously and controlled cautiously to reduce surgical site infections in DM patients before and after surgery. The ideal target for glycemic control before and after surgery is still a debated issue. In some studies, maintaining HbA1c less than 8.5% (69 mmol/mol) for the general population undergoing surgeries is highly suggested 43 . However, there is no consensus about whether DM patients planning to undergo free-flap reconstructions should maintain stricter glycemic control. Currently, intensive blood sugar control of less than 150 mg/dL before and after surgery might reduce the risk of surgical site infection 44 . Both the Society of Thoracic Surgeons (STS) and the American Association of Clinical Endocrinologists and American Diabetes Association (AACE/ADA) endorse a glycemic range between 140 and 180 mg/dL for postoperative patients 45 .
A series of analyses according to the infection-related complications were further performed. Interestingly, DM was not associated with dehiscence, fistulas, or plate exposure (Fig. 5). Early empirical antibiotic usage or switching, surgical intervention and return to the operating room might control infection and deter such serious and long-term complications. However, there has been no consensus regarding infection prophylaxis and management in patients with DM following microsurgical reconstructions 23,46 .
Pharmacologic prophylaxis with antiplatelets, anticoagulants, and volume expanders has been commonly prescribed after free flap transfers in DM patients due to their impaired microcirculation 47 . In this study, of the risk of hematoma was not increasing in diabetic flap recipients (Fig. 5). However, there was a trend of increasing total coagulation-related complications (e.g., hematoma, bleeding, required transfusion, and thrombosis) in the DM group (RR = 1.708, 95% CI 0.918-3.178, p = 0.091) (Fig. 5). Therefore, surgeons should be cautious in dose adjustment and coagulation monitoring while using antiplatelets or anticoagulants in DM patients after surgery 47 .
Regarding the quality aspects, we found that DM did not significantly increase readmission rates or mortality after free-flap reconstruction. However, other common clinical complications, including myocardial infarction, cerebrovascular event, deep vein thrombosis, pulmonary embolism, pneumonia, urinary tract infection, and septic shock, were noted to be increased after surgery in DM patients. It might be that those patients with H&N malignancies seemed to suffer from more comorbidities than the general population 48,49 . For instance, the high prevalence of tobacco and alcohol use in those patients imposes higher cardiovascular diseases, which could be exacerbated during surgery 50,51 . Additionally, blood transfusion during or after surgery might also aggravate heart failure, respiratory distress, and pneumonia 52 . Aside from cancer recurrence or metastasis, the most common causes of mortality after free-flap reconstruction were cardiac, pulmonary, and infectious etiologies 53 . Therefore, pre-or postoperative care should be more attentive for DM patients undergoing H&N reconstructions. A thorough examination of cardiopulmonary function and comprehensive comorbidity treatments should be implemented before the operation. In addition to surgical teams, different specialties might be consulted beforehand to avoid serious medical consequences.

Conclusion
In summary, DM patients are prone to develop various complications after H&N free-flap reconstructions, and more aggressive strategies should be taken to ensure better outcomes. Our study results suggest practical ways for surgeons and oncologists to evaluate the risk of surgery in these patients. A patient-based and individual decision-making process should always be implemented and cautiously reviewed before free-flap reconstruction.