Introduction

Free-flap reconstruction following head and neck (H&N) tumor ablation has become a routine practice with advancements in microsurgical techniques1. However, daunting postoperative complications after free-flap reconstruction, such as deprivation of eating, speaking, and breathing functions, have been reported2. Numerous practices have been introduced to improve surgical outcomes after free-flap reconstruction, especially risk factor evaluation3. 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 studies4,5,6,7.

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 radiotherapy4,5. In those cases, the quality of life after the operation is not satisfactory due to repeated readmissions or revision operations8,9,10. 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.

Patients and methods

Inclusion criteria

All published original studies of either randomized control trials or retrospective cohorts mentioning DM and any certain complications following H&N free-flap reconstructions were included. Articles by the same authors or based on the same study population should be verified as separate studies or designated to analyze a different complication before being included.

Search strategy

A search of electronic databases, namely PubMed Central, Embase, MEDLINE, and the Cochrane Library, from January 2005 to April 2020 was conducted; the search terms were “DM”, “H&N,” AND “free flaps.” This was further supplemented with cross-referencing the bibliographies from the papers identified by the search or other relevant articles. Only full-text manuscripts in the English language were considered for inclusion. For studies with overlapping periods of the same targeted population, the one with the largest cohort was retained, while additional consideration for inclusion was made if the overlapping paper mentioned a specific complication that was not previously covered. Articles without available documentation of the diabetic or nondiabetic case numbers and respective complication rates or without an odds ratio comparing risks in diabetic to those in nondiabetic patients were further excluded.

Data extraction

After selecting the relevant literature, the primary data collection was performed and further reviewed by a second author to ensure accuracy. The data collected were as follows: definition of the reconstructed regions, study population and years, numbers of total patients and patients with DM, and rates of complications in both patients with and without DM. Various complications (surgical and medical), surgical complications, return to the operating room, free-flap failure/necrosis, surgical site infections, dehiscence, hematomas, coagulation-related complications (e.g., hematoma, bleeding, requiring transfusion or thrombosis), fistulas, plate exposures, other complications, readmissions, and mortality were included in the complication analysis.

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, I2, and Tau2 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-analysis5,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37. 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.

Figure 1
figure 1

The flow diagram shows the process of enrolling studies for the meta-analysis.

Full size image
Table 1 Patient Characteristics of the studies included in the meta-analysis (n = 27).
Full size table
Figure 2
figure 2

The funnel plot of all studies included.

Full size image

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).

Figure 3
figure 3

The postoperative complications forest plot: total and surgical complications. *Eskander32.

Full size image

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).

Figure 4
figure 4

The individual surgical complications forest plot, part I: return to the operating room, flap failure/necrosis, and surgical site infection. *Eskander32.

Full size image
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).

Surgical site infections

A total of 11 studies evaluated surgical site infections, with 5488 patients enrolled; 773 of them had DM. The pooled results included 8 studies with RR. There was a trend of increasing surgical site infections in the DM group, nearly reaching statistical significance (RR = 1.401, 95% CI 0.987–1.987, p = 0.059). After pooling the results from the 5 studies that reported OR, DM significantly increased surgical site infections (OR = 2.414, 95% CI 1.619–3.599, p < 0.001) (Fig. 4).

Dehiscence

Four studies were included to evaluate dehiscence rates, with 473 DM patients among 3311 total subjects. All complication rates were expressed as RR. The pooled results showed that DM did not significantly increase dehiscence rates (RR = 1.162, 95% CI 0.814–1.660, p = 0.408) (Fig. 5).

Figure 5
figure 5

The individual surgical complications forest plot, part II: dehiscence, fistula, plate exposures, hematoma, and coagulation-related complications. *Eskander32.

Full size image
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).

Figure 6
figure 6

Quality aspects: readmissions, mortality, and other complications. *Eskander32, **Eskander35. Other complications include myocardial infarction, cerebrovascular event, deep vein thrombosis, pulmonary embolism, pneumonia, urinary tract infection, and septic shock.

Full size image
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 healing8,9,10. However, retrospective cohort studies from different institutions and even different time periods within the same database could end up with contradicting conclusions4,5.

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 region38. Consisting of merely five studies, this analysis seemed relatively short of conclusiveness38. Further investigations were performed by Cupato et al.39; these authors gathered 16 studies for meta-analysis. They concluded that DM significantly increased flap failures and local region complications39. This complication analysis was performed to include all major H&N surgeries instead of limiting the analysis to free-flap reconstructions39. 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 subjects5,12,40,41. Some studies concluded that patients with DMS were at increased risk and others did not5,12,40,41. 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 patients8. 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 outcomes42. 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 healing10. Higher infection rates in DM patients have been previously reported from various types of surgical procedures6. In our 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 suggested43. 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 infection44. 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 patients45.

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 reconstructions23,46.

Pharmacologic prophylaxis with antiplatelets, anticoagulants, and volume expanders has been commonly prescribed after free flap transfers in DM patients due to their impaired microcirculation47. 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 surgery47.

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 population48,49. For instance, the high prevalence of tobacco and alcohol use in those patients imposes higher cardiovascular diseases, which could be exacerbated during surgery50,51. Additionally, blood transfusion during or after surgery might also aggravate heart failure, respiratory distress, and pneumonia52. Aside from cancer recurrence or metastasis, the most common causes of mortality after free-flap reconstruction were cardiac, pulmonary, and infectious etiologies53. 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.