- Email: [email protected]
Free flap reconstruction of the head and neck: Analysis of 241 cases
Free flap reconstruction of the head and neck: Analysis of 241 cases BRUCE H. HAUGHEY, MB, ChB, EWAIN WILSON, MD, LUCIA KLUWE, MD, JAY PICCIRILLO, MD, JOHN FREDRICKSON, MD, DONALD SESSIONS, MD, and GERSHON SPECTOR, MD, St Louis, Missouri
OBJECTIVE: We undertook this study of free flap reconstruction of the head and neck to stratify patients and procedures, to determine how donor site preference changed over time, to assess medical and surgical outcomes, and to identify variables associated with complications. METHODS: We analyzed computerized medical records from 236 patients who underwent a total of 241 reconstructions at a tertiary academic medical center in St. Louis between 1989 and 1998. We created a more detailed retrospective database of 141 of those patients by using 48 perioperative variables and 7 adverse outcome measures. Multivariate statistical models were used to analyze associations between variables and outcomes. RESULTS: The fibula became the preferred donor site for mandibular reconstruction, and the radial forearm became the preferred donor site for pharyngoesophageal reconstruction. For the 241 procedures, the mortality rate was 2.1%, the median length of stay was 11 days, and the flap survival rate was 95%. Administration of more than 7 L of crystalloid during surgery and age over 55 were associated with major medical complications. Blood transfusion, prognostic comorbidity, and number of surgeons correlated with length of stay. Cigarette smoking and receipt of more than 7 L of crystalloid during surgery were associated with overall flap complications, and weight loss of more than 10% before surgery, more than one operating
From the Department of Otolaryngology–Head and Neck Surgery, Washington University School of Medicine. Presented at the Annual Meeting of the American Academy of Otolaryngology–Head and Neck Surgery, New Orleans LA, September 26-29, 1999. Reprint requests: Bruce H. Haughey, MBChB, FRACS, FACS, Director, Division of Head & Neck Surgical Oncology, Department of Otolaryngology, Washington University School of Medicine, Campus Box 8115, 660 S. Euclid, St. Louis, MO 63110; e-mail, [email protected]. Copyright © 2001 by the American Academy of Otolaryngology–Head and Neck Surgery Foundation, Inc. 0194-5998/2001/$35.00 + 0 23/1/116788 doi:10.1067/mhn.2001.116788 10
surgeon, and cigarette smoking were associated with major flap complications. CONCLUSIONS: Risk to patients and transferred tissue is low in free flap head and neck reconstruction. Age, smoking history, and weight loss should be considered during patient selection. Fluid balance should be considered during and after surgery. Division of labor for patient care should be carefully delineated among surgeons in a teaching setting. (Otolaryngol Head Neck Surg 2001;125:10-7.)
Free flap reconstruction of the head and neck was introduced with the jejunal free flap by Seidenberg in 1959.1 Because of improved knowledge of donor site anatomy2 and advances in microvascular surgery,3 it has become the most reliable and efficient method for restoring tissue to regions of the head and neck that have been resected because of diseases such as cancer or trauma.4,5 Advantages include the single stage procedures, innervated reconstructions, the versatility of flaps reconstructed soft tissue and bone and the ability to tailor repairs to complex 3-dimensional structures such as the tongue, and the fact that these are innervated reconstructions and single-stage procedures. In this article, we describe our use of this technique for over 10 years in a teaching setting: the Department of Otolaryngology at Washington University School of Medicine in St. Louis. We identify variables that influence medical and surgical outcomes by analyzing our series of 241 free flap transfers to the head and neck. The specific objectives of this retrospective study were to stratify patients and procedures, to determine whether donor sites changed over time, to assess medical and surgical outcomes, and to identify variables associated with complications. METHODS AND MATERIAL Data from all patients who underwent free flap reconstructions at our tertiary referral teaching center between 1989 and 1998 were used for this study with approval from the University’s Human Studies Committee. All reconstructions were managed by the corresponding author and resections by the corresponding author or another surgeon. Residents assisted with or performed the primary resections and neck dissec-
Otolaryngology– Head and Neck Surgery Volume 125 Number 1
HAUGHEY et al
Table 3. Donor sites
Table 1. Diagnoses Diagnosis
Site
N
Squamous cell carcinoma Other malignancy Osteoradionecrosis Trauma Congenital Fibrous dysplasia Mengioma Other Total
Radial forearm ± radius Fibula Jejunum Iliac crest Rectus abdominus Latissimus dorsi Serratus/rib Lateral thigh Scapula Total
184 30 14 5 2 1 1 4 241
Table 2. Recipient sites Site
Oromandibular Oral soft tissue Oropharyngeal Pharyngoesophageal Midface Facial soft tissue Skull base Other Total
11
N
%
73 43 41 35 14 14 8 13 241
30% 18% 17% 15% 6% 6% 3% 5% 100%
tions under supervision, isolating blood vessels at the recipient sites. A microvascular fellow took progressive responsibility for free tissue transfer over a 1-year training period. Because this study focused on perioperative outcomes, we followed patients for at least 30 days, entering records into a computerized database. We conducted a more detailed analysis, based on medical records, of the patients treated between 1992 and 1997. This smaller database provided comprehensive preoperative, intraoperative, and postoperative data for 139 patients and 141 free flap transfers. We studied 48 variables in these 141 cases. Demographic variables were age, race, gender, weight, and height. Clinical variables included defect site, tumor stage, reason for the flap, prior treatments such as radiation or surgery, donor site, recipient site, recent weight loss, medical comorbidity (assessed with the Kaplan-Feinstein scale6), tobacco and alcohol use, hematology values, serum biochemistry, and medication history. Intraoperative variables included the number of surgeons, length of anesthesia, volume and type of intravenous fluids, blood transfusions, and use of pressors. The database also contained postoperative outcome measures. These included 30-day survival, length of stay (which correlates with cost in our hospital), and medical complications. Surgical outcomes included flap survival, flap complications such as ischemia and flap loss, and donor site complications. A flap complication was defined as major if further surgery was necessary for that complication. Data
N
96 46 27 22 19 19 5 5 2 241
%
40% 19% 11% 9% 8% 8% 2% 2% 1% 100%
Table 4. Variables associated with major medical complications: univariate analysis (N = 141) Variable
Major comorbidity Age >55 Volume of crystalloid >7 L Reason for flap (bone reconstruction) Intraoperative pressors
P
0.007 0.009 0.017 0.028 0.077
was captured on a coding form and entered into a personal computer. It should be noted that we recorded minor medical complications, such as rashes and urinary tract infections, as well as life-threatening complications, such as myocardial infarction, infection, and gastrointestinal bleeding. Also, we defined flap complications broadly to include any condition at the recipient or donor site that possibly could be attributed to the flap, even though some (eg, hematomas, infections, and pharyngocutaneous fistulas) could have originated in the recipient site. One of the authors (JSP) performed the statistical calculations with the χ2 test for univariate analysis and the logistic regression model for multivariate analysis. RESULTS
Between 1989 and 1998, 236 patients underwent free flap reconstructions on the corrresponding author’s service at our center. There were 155 men and 81 women. However, 5 of the patients (3 men and 2 women) underwent a second resection and reconstruction during this period because their cancer recurred, increasing the total number of free flap reconstructions to 241. Therefore, 66% of the reconstructions were performed on males and 34% on females. Sixty-one percent of the reconstructions involved patients 55 years of age or younger; the remaining 94 (39%) were over 55. The age range was 16 to 91, with a median age of 62.
12
HAUGHEY et al
Otolaryngology– Head and Neck Surgery July 2001
Fig 1. Medical complications among 141 patients who underwent reconstructive head and neck surgery between 1992 and 1997. Eighty of these patients developed a total of 124 medical complications.
The 241 cases included 223 reconstructions after cancer surgery (93%). The remaining 18 reconstructions (7%) were performed after trauma, osteoradionecrosis, or removal of benign tumors such as large mandibular ameloblastomas. Table 1 lists the diagnoses. One hundred one (45%) of the cancer cases had received preoperative radiation therapy. One hundred twenty-nine (58%) of the 223 cancer cases underwent primary treatment, whereas 111 (46%) were reconstructed after salvage resection because a tumor had recurred or osteoradionecrosis had developed. The remaining 8% had been resected during previous surgery, often at another institution. Neck dissections were performed in 150 (62%) of the 241 cases: unilateral in 118 cases and bilateral in 32 cases. Table 2 indicates the recipient sites in the head and neck; they were predominantly the oral cavity and oropharynx. The major donor sites were the radial forearm (40%) and fibula (19%), although a range of other sites also were used (Table 3). Over time, however, donor site preferences changed. For mandibular reconstructions (N = 72) during the first half of the series (cases 1-120), 45% (14/31) of the reconstructions were performed with bone and soft tissue taken from the deep circumflex DCIA system (iliac crest), and 52% were performed with fibular material. During the second half of the
series (cases 121-241), the proportions changed to 15% (iliac crest) and 80% (fibula). There was also a distinct shift from using jejunal flaps to using tubed fasciocutaneous flaps for pharyngoesophageal reconstruction (N = 35): 88% jejunal and 12% fasciocutaneous during the first half of the series and 17% jejunal and 83% fasciocutaneous during the second. In most cases, the fasciocutaneous flaps were taken from the radial forearm, though some came from the lateral thigh. Outcome measures obtained from the complete data set of 241 cases were 30-day mortality, length of stay, and flap loss. Five patients (2.1%) died during the 30day postoperative period. Causes of death were a perforated bleeding duodenal ulcer, myocardial infarction, a sudden death that possibly resulted from tracheal tube plugging, hyperosmolar diabetic hyperglycemia, and adult respiratory distress syndrome. The median length of stay was 11 days, with a 5-day minimum and a 41day maximum. Two hundred twenty-eight (95%) of the 241 flaps survived completely. Ten flaps (4%) were completely lost, and 3 (1%) were partially lost. Of these 13, there were 6 repeat flap transfers for tissue loss and 7 that were managed without returning the patients to the operating room. The 141-case data set enabled us to analyze outcomes in more detail. It showed that 65 (46%) of these cases were managed entirely by the corresponding author
Otolaryngology– Head and Neck Surgery Volume 125 Number 1
HAUGHEY et al
13
Table 5. Variables associated with major medical complications: multivariate analysis (n = 141) 95% CI Variable
Category
Volume of crystalloid Age
Major comorbidity
<7 L >7 L ≤55 56-66 >66 No Yes
Adjusted odds ratio
1 5.08 1 14.15 20.52 1 4.34
Perioperative blood transfusion Recipient site (hypopharynx>oropharynx>oral cavity) Two or more surgeons Administration of albumin
Upper
Reference 1.60
P
0.006 16.20
Reference 1.65 2.37
121.26 177.93 Reference
0.90
Table 6. Variables associated with increased length of stay: univariate analysis (n = 141) Variable
Lower
0.02 0.01 0.067
20.78
Table 7. Variables associated with increased length of stay: multivariate analysis (n = 141) P
0.001 0.009 0.02 0.031
Variable
Number of surgeons Blood transfusion Prognostic comorbidity Type of colloid
P
0.002 0.004 0.013 0.340
(BHH) and that two or more surgeons were present during the remaining 76 cases (54%). When multiple surgeons were involved, the senior author performed only the free flap reconstruction. Subsequent surgery during the first 30 days was performed in 33 (23%) of the 141 cases. Of the 15 secondary reconstructions the most common type was a pedicled myocutaneous flap, usually the pectoralis major. Wound healing and/or carotid protection was successful in all of these procedures. The most common donor site complications were skin graft losses and infection: two required repeat surgery, one for debridement and application of a second skin graft (forearm), and the other for hematoma drainage (fibula).
(Kaplan-Feinstein level 3) age over 55, and administration of more than 7 L of crystalloid during anesthesia (Table 4). Using multivariate analysis, we determined that administration of more than 7 L of crystalloid during surgery was significantly associated with major medical complications (Table 5). The odds ratio was 5:1. Age was also a significant variable, and subjects between ages 56 and 66 were 14 times more likely to have a major medical complication develop than those 55 and under. The odds were more than 21:1 when patients 67 or older were compared with those 55 and under. Comorbidity did not reach statistical significance, though its presence appeared to increase the absolute risk more than 4-fold.
Medical complications
Length of stay
Eighty (57%) of the 139 patients developed one or more medical complications within 30 days of reconstructive surgery, and they had 124 complications in all. The most common were cardiovascular, respiratory, and central nervous system complications, as shown in Fig. 1. We classified medical complications on a 4point (0-3) scale, which assigns a score of 3 to lifethreatening complications. Using the χ 2 test, we compared frequencies of complications when a variable was either present or absent. When we considered major (severity level 3) medical complications, variables that reached statistical significance (P < 0.05) were major comorbidity
The median length of stay was 11 days, with a minimum of 5 days and a maximum of 41 days. The χ2 test associated perioperative blood transfusions, recipient site (hypopharynx vs oral cavity, oropharynx, or others), number of surgeons, and administration of albumin during surgery with length of stay (Table 6). For example, patients who underwent reconstruction of the hypopharynx stayed longer (median of 17 days) than the other sites (medians of 10-12 days). The multivariate analysis indicated independent association of blood transfusion, “prognostic” comorbidity (K-F = 3), and number of surgeons with increased length of stay (Table 7).
14
HAUGHEY et al
Otolaryngology– Head and Neck Surgery July 2001
Fig 2. Flap complications among 141 patients who underwent reconstructive head and neck surgery between 1992 and 1997. Forty of these patients developed a total of 64 complications.
Flap complications
These complications developed in 40 of 139 (29%) patients. The most common flap complications were fistula, wound dehiscence, and hematomas or seromas (Fig 2). There was no significant decrease in complication rate within the 5 years over which they were recorded. Table 8 summarizes the significant variables with the χ2 tests. Multivariate analysis of variables associated with any type of flap complication identified cigarette smoking during the 2 weeks before surgery as a significant variable. A large volume of crystalloid (>7 L) administered during surgery was also significant, but the involvement of more than one surgeon just failed to reach significance (Table 9). Loss of more than 10% of body weight before surgery was associated significantly with a major flap complication developing (one that required a return to the operating room), along with having more than one surgeon and smoking cigarettes. For major flap complications, multivariate analysis identified the same significant variables as univariate analysis: weight loss, having more than one surgeon, and smoking cigarettes (Table 10). DISCUSSION
We conclude that free tissue transfer is a reliable technique for head and neck reconstruction in a tertiary referral and teaching center. The perioperative mortality in our series of 241 patients was 2.1%, which is comparable with rates reported for other large series of free flap transfers to the head and neck (7.0%7, 2.0%8, 4.7%9, 6.3%10).
The donor sites in our series changed significantly over a decade. For mandibular reconstructions, we came to prefer the fibula over the iliac crest, mainly because it can be harvested easily under tourniquet control with little loss of blood. Also, its versatile cutaneous unit can be used for lining the aerodigestive tract and/or for skin coverage. It is, however, more prone to atherosclerosis in the arterial pedicle than the iliac crest or scapula. For repairing pharyngoesophageal defects, we now prefer fasciocutaneous flaps over jejunal flaps because, in our clinical judgment, there is less donor site morbidity. The possibility of surgically related abdominal hemorrhage or peritonitis is eliminated. Our detailed study of complications and statistical analyses identified several variables that associate with less satisfactory medical or surgical outcomes (Table 11). Administering a large volume of crystalloid during surgery was a risk factor for developing major medical complications and/or for any level of flap complication. Reasons for crystalloid administration include fluid replacement, transfusion with packed cells, and hypotension during surgery; interestingly, duration of anesthesia, which might co-vary with volume of crystalloids given, did not predict for complications, nor did the use of blood tranfusions or pressors. The pathologic results of increased crystalloid administration could include fluid overload and its cardiorespiratory consequences. Administration of a large volume of crystalloid during surgery was also associated with flap complications. One hypothesis is that excessive edema in the flap and/or the recipient site promotes swelling and mechan-
Otolaryngology– Head and Neck Surgery Volume 125 Number 1
HAUGHEY et al
15
Table 8. Variables associated with all flap complications: univariate analysis (n = 141) P (Any complication) 60/141
Variable
Cigarette smoking 2 weeks before surgery More than 1 surgeon Major (level 3) medical complications Alcohol during the 2 weeks before surgery Intraoperative pressors Large volume of crystalloid (>7 L) Weight loss >10%
0.001 0.041 0.042 0.057 0.083 0.149 0.419
P (Major complications) 8/141
Variable
Weight loss >10% More than 1 surgeon Cigarette smoking 2 weeks before surgery Large volume of crystalloid (>7 L) Intraoperative pressors Alcohol 2 weeks before surgery Major (level 3) medical complications
0.023 0.030 0.036 0.138 0.164 0.585
Table 9. Variables associated with any flap complication: multivariate analysis (n = 141) 95% CI Variable
Cigarette smoking Volume of crystalloid Number of surgeons Volume of blood lost Intraoperative pressors Alcohol consumption
Category
No Yes <7 L >7 L 1 2 or more <1 L >1 L No Yes No Yes
Adjusted odds ratio
1 5.33 1 2.75 1 2.19 1 0.48 1 2.32 1 1.14
ically stresses the pedicles. The 5 patients in the detailed, 141-case database whose flaps failed had a median of 9.5 (3.7-11.0) L of crystalloid given. The low, 3.7-L value was in a patient whose flap failed because of mechanical separation of the vein in an alcoholic withdrawal crisis. Edema in recipient sites and flaps might also disrupt the closure lines. In response to this association, we recommend to our anaesthetist colleagues that crystalloid volume be minimized within the limits of adequate vital organ perfusion. It is plausible that >7 L of crystalloid was a marker for anesthetic practice, a variable not included in our study. Another statistically significant variable for medical complications, advancing age, suggests that extra attention should be paid to the preoperative and postoperative management of older patients. A careful presurgical medical evaluation, preferably with the internal medical surveillance postoperatively, would be ideal in patients over 55 in view of the likelihood that major surgery will unmask physiologic vulnerability. Our study differs from an Australian study11 that failed to find a significant difference in postoperative complication rates between patients over and under 70 who were stratified
Lower
Upper
Reference 1.84
0.002 15.42
Reference 1.13
0.025 6.69
Reference 0.99
0.053 4.84
Reference 0.20
0.098 1.15
Reference 0.88
0.121 6.70
Reference 0.37
P
0.814 3.49
for preoperative comorbidity. However, in that study the absolute numbers of medical complications were still 29% in the under-70 group and 54% in the over-70 group. A British study found that older men were at risk for postoperative chest infection.9 Cigarette smoking within 2 weeks of surgery was also associated with flap complications. In this respect, our findings differ from those of several other series9,12 in which smoking did not appear to be a risk factor for flap complications; however, these series did not specify proximity of smoking to surgery. There is evidence, however, from other clinical literature and laboratory work that tobacco products inhibit healing and promote tissue loss in flaps.13 Therefore we recommend physicians discourage smoking in patients who are about to undergo free flap head and neck reconstruction. The finding that both length of stay and major flap complications increase significantly when more than one surgeon is involved is open to several interpretations. One possibility is that outcomes improve when the responsibility for preoperative selection and postoperative management decisions rests with one person. This finding encourages improved
16
Otolaryngology– Head and Neck Surgery July 2001
HAUGHEY et al
Table 10. Multivariate analysis of association with major flap complications (n = 8/141) 95% CI Variable
Category
Weight loss >10% Number of surgeons Cigarette smoking Volume of crystalloid
Adjusted odds ratio
No Yes 1 2 or more No Yes <7 L >7 L
Lower
1 3.62 1 4.22 1 3.23 1 2.33
Upper
Reference 1.21
P
0.022 10.83
Reference 1.22
0.023 14.78
Reference 1.08
0.036 9.68
Reference 0.76
0.138 7.23
Table 11. Summary of multivariate analyses Outcome Variable
Flap complication (any)
Cigarette smoking Volume crystalloid >7 L Two of more surgeons Weight loss (> 10%) Prognostic co-morbidity Blood transfusion Age >55 years
** **
Flap complication (major)
Medical complication (major) Length of Stay
** ** ** **
communication and peer review of cases selected for this type of surgery. Because there is no other study that controls for this variable, we cannot compare our results with other institutions. However, other centers do report comparable flap success rates with multiple surgeon teams.10 The association of patient weight loss with major flap complications may be explained by the effect of nutritional deficits on wound healing14; this continues to be a variable we try to eliminate by hyperalimentation before surgery. Pertinent negative findings were that preoperative irradiation had no association with medical or surgical outcomes even though radiotherapy is known to damage blood vessels and delay healing. A negative effect of irradiation has been reported in some studies9,15,16 but not in others.17 Duration of surgery did not influence medical or surgical outcomes; therefore, the adverse medical effects associated with high crystalloid volumes and two or more surgeons were unlikely to have resulted from prolongation of procedures. Our finding that age fails to affect flap outcomes agrees with several previous studies,7,17,18 but we demonstrated the significant effect of advancing age on major medical complications. Finally, the failure of all but major
**
** ** **
comorbidity using the Kaplan-Feinstein scale to predict complications probably speaks to case selection and our routine practice of seeking an internal medical consultation before free flap operations. This is at variance with another study that used different (Charleson) comorbidity index; this study did not distinguish complications as medical or surgical, as we did, but rather grouped complication severity and types.16 CONCLUSION
We will continue to offer microvascular free tissue transfer as the preferred method of reconstruction in appropriately selected patients with head and neck disease. Cognizance of potentially significant variables uncovered by this study should help optimize future outcomes and generates hypotheses for future investigation. Microvascular free flap transfers to the head and neck are known to be safe, but enhanced attention to patient selection, preparation of patients for surgery, fluid balance, and coordination between members of the care team should increase effectiveness and safety even further. REFERENCES 1. Seidenberg B. Immediate reconstruction of the cervical esopha-
Otolaryngology– Head and Neck Surgery Volume 125 Number 1
2. 3. 4.
5. 6. 7. 8. 9. 10.
gus by a revascularized isolated jejunal segment. Ann Surg 1959;149:162. Taylor GI, Razaboni RM. Michel Salmon’s anatomical studies. Book 1. Arteries of the muscles of the extremities and the trunk. St Louis: Quality Medical Publishing; 1994. O’Brien BMcC. Microvascular reconstructive surgery. Edinburgh, London, New York: Churchill Livingstone; 1977. Kroll SS, Schusterman MA, Reece GP. Comparison of the rectus abdominis free flap with the pectoralis major myocutaneous flap for reconstructions in the head and neck. Am J Surg 1992;164:615-8. Kroll SS, Schusterman MA, Reece GP. Costs and complications in mandibular reconstruction. Ann Plast Surg 1992;29:341-7. Kaplan MH, Feinstein AR. The importance of classifying initial co-morbidity in evaluating the outcome of diabetes mellitus. J Chron Dis 1974;27:387-404. Shestak KC, Jones NF. Microsurgical free tissue transfer in the elderly patient. Plast Reconstr Surg 1991;88:259-63. McNamara M, Pope S, Sadler A, et al. Microvascular free flaps in head and neck surgery. J Laryngol Otol 1994;108:962-8. Simpson KH, Murphy PG, Hopkins PM, et al. Prediction of outcomes in 150 patients having microvascular free tissue transfers to the head and neck. Br J Plast Surg 1996;49:267-73. Jones NF, Johnson JT, Shestak KC, et al. Microsurgical reconstruction of the head and neck: interdisciplinary collaboration
HAUGHEY et al
11. 12. 13. 14. 15.
16. 17. 18.
17
between head and neck surgeons and plastic surgeons in 305 cases. Ann Plast Surg 1996;36:37-43. Bridger AG, O’Brien CJ, Lee KK. Advanced patient age should not preclude the use of free-flap reconstruction for head and neck cancer. Am J Surg 1994;168:425-8. Reus WF, Colen LB, Straker DJ. Tobacco smoking and complications in elective microsurgery. Plast Reconstr Surg 1992;89:490-4. Chang LD, Buncke G, Slezak S, et al. Cigarette smoking, plastic surgery and microsurgery. J Reconstr Microsurg 1996;12:465-74. Steidelman WK, Digenis AG, Tobin GR. Impediments to wound healing. Am J Surg 1998;176(2A Suppl):395-475. Bengston BP, Schusterman MA, Baldwin BJ, et al. Influence of prior radiotherapy on the development of postoperative complications and success of free flaps in head and neck reconstruction. Am J Surg 1993;166:326-30. Singh B, Cordeiro PG, Santamaria E, et al. Factors associated with complications in microvascular reconstruction of head and neck defects. Plast Reconstr Surg 1999;103:403-11. Kroll SS, Robb GL, Reece GP. Does prior irradiation increase the risk of total or partial free flap loss? J Reconstr Microsurg 1998;14:263-8. Chick LR, Watson RL, Reus W, et al. Free flaps in the elderly. Plast Reconstr Surg 1992;1:87-94.
OBJECTIVE: We undertook this study of free flap reconstruction of the head and neck to stratify patients and procedures, to determine how donor site preference changed over time, to assess medical and surgical outcomes, and to identify variables associated with complications. METHODS: We analyzed computerized medical records from 236 patients who underwent a total of 241 reconstructions at a tertiary academic medical center in St. Louis between 1989 and 1998. We created a more detailed retrospective database of 141 of those patients by using 48 perioperative variables and 7 adverse outcome measures. Multivariate statistical models were used to analyze associations between variables and outcomes. RESULTS: The fibula became the preferred donor site for mandibular reconstruction, and the radial forearm became the preferred donor site for pharyngoesophageal reconstruction. For the 241 procedures, the mortality rate was 2.1%, the median length of stay was 11 days, and the flap survival rate was 95%. Administration of more than 7 L of crystalloid during surgery and age over 55 were associated with major medical complications. Blood transfusion, prognostic comorbidity, and number of surgeons correlated with length of stay. Cigarette smoking and receipt of more than 7 L of crystalloid during surgery were associated with overall flap complications, and weight loss of more than 10% before surgery, more than one operating
From the Department of Otolaryngology–Head and Neck Surgery, Washington University School of Medicine. Presented at the Annual Meeting of the American Academy of Otolaryngology–Head and Neck Surgery, New Orleans LA, September 26-29, 1999. Reprint requests: Bruce H. Haughey, MBChB, FRACS, FACS, Director, Division of Head & Neck Surgical Oncology, Department of Otolaryngology, Washington University School of Medicine, Campus Box 8115, 660 S. Euclid, St. Louis, MO 63110; e-mail, [email protected]. Copyright © 2001 by the American Academy of Otolaryngology–Head and Neck Surgery Foundation, Inc. 0194-5998/2001/$35.00 + 0 23/1/116788 doi:10.1067/mhn.2001.116788 10
surgeon, and cigarette smoking were associated with major flap complications. CONCLUSIONS: Risk to patients and transferred tissue is low in free flap head and neck reconstruction. Age, smoking history, and weight loss should be considered during patient selection. Fluid balance should be considered during and after surgery. Division of labor for patient care should be carefully delineated among surgeons in a teaching setting. (Otolaryngol Head Neck Surg 2001;125:10-7.)
Free flap reconstruction of the head and neck was introduced with the jejunal free flap by Seidenberg in 1959.1 Because of improved knowledge of donor site anatomy2 and advances in microvascular surgery,3 it has become the most reliable and efficient method for restoring tissue to regions of the head and neck that have been resected because of diseases such as cancer or trauma.4,5 Advantages include the single stage procedures, innervated reconstructions, the versatility of flaps reconstructed soft tissue and bone and the ability to tailor repairs to complex 3-dimensional structures such as the tongue, and the fact that these are innervated reconstructions and single-stage procedures. In this article, we describe our use of this technique for over 10 years in a teaching setting: the Department of Otolaryngology at Washington University School of Medicine in St. Louis. We identify variables that influence medical and surgical outcomes by analyzing our series of 241 free flap transfers to the head and neck. The specific objectives of this retrospective study were to stratify patients and procedures, to determine whether donor sites changed over time, to assess medical and surgical outcomes, and to identify variables associated with complications. METHODS AND MATERIAL Data from all patients who underwent free flap reconstructions at our tertiary referral teaching center between 1989 and 1998 were used for this study with approval from the University’s Human Studies Committee. All reconstructions were managed by the corresponding author and resections by the corresponding author or another surgeon. Residents assisted with or performed the primary resections and neck dissec-
Otolaryngology– Head and Neck Surgery Volume 125 Number 1
HAUGHEY et al
Table 3. Donor sites
Table 1. Diagnoses Diagnosis
Site
N
Squamous cell carcinoma Other malignancy Osteoradionecrosis Trauma Congenital Fibrous dysplasia Mengioma Other Total
Radial forearm ± radius Fibula Jejunum Iliac crest Rectus abdominus Latissimus dorsi Serratus/rib Lateral thigh Scapula Total
184 30 14 5 2 1 1 4 241
Table 2. Recipient sites Site
Oromandibular Oral soft tissue Oropharyngeal Pharyngoesophageal Midface Facial soft tissue Skull base Other Total
11
N
%
73 43 41 35 14 14 8 13 241
30% 18% 17% 15% 6% 6% 3% 5% 100%
tions under supervision, isolating blood vessels at the recipient sites. A microvascular fellow took progressive responsibility for free tissue transfer over a 1-year training period. Because this study focused on perioperative outcomes, we followed patients for at least 30 days, entering records into a computerized database. We conducted a more detailed analysis, based on medical records, of the patients treated between 1992 and 1997. This smaller database provided comprehensive preoperative, intraoperative, and postoperative data for 139 patients and 141 free flap transfers. We studied 48 variables in these 141 cases. Demographic variables were age, race, gender, weight, and height. Clinical variables included defect site, tumor stage, reason for the flap, prior treatments such as radiation or surgery, donor site, recipient site, recent weight loss, medical comorbidity (assessed with the Kaplan-Feinstein scale6), tobacco and alcohol use, hematology values, serum biochemistry, and medication history. Intraoperative variables included the number of surgeons, length of anesthesia, volume and type of intravenous fluids, blood transfusions, and use of pressors. The database also contained postoperative outcome measures. These included 30-day survival, length of stay (which correlates with cost in our hospital), and medical complications. Surgical outcomes included flap survival, flap complications such as ischemia and flap loss, and donor site complications. A flap complication was defined as major if further surgery was necessary for that complication. Data
N
96 46 27 22 19 19 5 5 2 241
%
40% 19% 11% 9% 8% 8% 2% 2% 1% 100%
Table 4. Variables associated with major medical complications: univariate analysis (N = 141) Variable
Major comorbidity Age >55 Volume of crystalloid >7 L Reason for flap (bone reconstruction) Intraoperative pressors
P
0.007 0.009 0.017 0.028 0.077
was captured on a coding form and entered into a personal computer. It should be noted that we recorded minor medical complications, such as rashes and urinary tract infections, as well as life-threatening complications, such as myocardial infarction, infection, and gastrointestinal bleeding. Also, we defined flap complications broadly to include any condition at the recipient or donor site that possibly could be attributed to the flap, even though some (eg, hematomas, infections, and pharyngocutaneous fistulas) could have originated in the recipient site. One of the authors (JSP) performed the statistical calculations with the χ2 test for univariate analysis and the logistic regression model for multivariate analysis. RESULTS
Between 1989 and 1998, 236 patients underwent free flap reconstructions on the corrresponding author’s service at our center. There were 155 men and 81 women. However, 5 of the patients (3 men and 2 women) underwent a second resection and reconstruction during this period because their cancer recurred, increasing the total number of free flap reconstructions to 241. Therefore, 66% of the reconstructions were performed on males and 34% on females. Sixty-one percent of the reconstructions involved patients 55 years of age or younger; the remaining 94 (39%) were over 55. The age range was 16 to 91, with a median age of 62.
12
HAUGHEY et al
Otolaryngology– Head and Neck Surgery July 2001
Fig 1. Medical complications among 141 patients who underwent reconstructive head and neck surgery between 1992 and 1997. Eighty of these patients developed a total of 124 medical complications.
The 241 cases included 223 reconstructions after cancer surgery (93%). The remaining 18 reconstructions (7%) were performed after trauma, osteoradionecrosis, or removal of benign tumors such as large mandibular ameloblastomas. Table 1 lists the diagnoses. One hundred one (45%) of the cancer cases had received preoperative radiation therapy. One hundred twenty-nine (58%) of the 223 cancer cases underwent primary treatment, whereas 111 (46%) were reconstructed after salvage resection because a tumor had recurred or osteoradionecrosis had developed. The remaining 8% had been resected during previous surgery, often at another institution. Neck dissections were performed in 150 (62%) of the 241 cases: unilateral in 118 cases and bilateral in 32 cases. Table 2 indicates the recipient sites in the head and neck; they were predominantly the oral cavity and oropharynx. The major donor sites were the radial forearm (40%) and fibula (19%), although a range of other sites also were used (Table 3). Over time, however, donor site preferences changed. For mandibular reconstructions (N = 72) during the first half of the series (cases 1-120), 45% (14/31) of the reconstructions were performed with bone and soft tissue taken from the deep circumflex DCIA system (iliac crest), and 52% were performed with fibular material. During the second half of the
series (cases 121-241), the proportions changed to 15% (iliac crest) and 80% (fibula). There was also a distinct shift from using jejunal flaps to using tubed fasciocutaneous flaps for pharyngoesophageal reconstruction (N = 35): 88% jejunal and 12% fasciocutaneous during the first half of the series and 17% jejunal and 83% fasciocutaneous during the second. In most cases, the fasciocutaneous flaps were taken from the radial forearm, though some came from the lateral thigh. Outcome measures obtained from the complete data set of 241 cases were 30-day mortality, length of stay, and flap loss. Five patients (2.1%) died during the 30day postoperative period. Causes of death were a perforated bleeding duodenal ulcer, myocardial infarction, a sudden death that possibly resulted from tracheal tube plugging, hyperosmolar diabetic hyperglycemia, and adult respiratory distress syndrome. The median length of stay was 11 days, with a 5-day minimum and a 41day maximum. Two hundred twenty-eight (95%) of the 241 flaps survived completely. Ten flaps (4%) were completely lost, and 3 (1%) were partially lost. Of these 13, there were 6 repeat flap transfers for tissue loss and 7 that were managed without returning the patients to the operating room. The 141-case data set enabled us to analyze outcomes in more detail. It showed that 65 (46%) of these cases were managed entirely by the corresponding author
Otolaryngology– Head and Neck Surgery Volume 125 Number 1
HAUGHEY et al
13
Table 5. Variables associated with major medical complications: multivariate analysis (n = 141) 95% CI Variable
Category
Volume of crystalloid Age
Major comorbidity
<7 L >7 L ≤55 56-66 >66 No Yes
Adjusted odds ratio
1 5.08 1 14.15 20.52 1 4.34
Perioperative blood transfusion Recipient site (hypopharynx>oropharynx>oral cavity) Two or more surgeons Administration of albumin
Upper
Reference 1.60
P
0.006 16.20
Reference 1.65 2.37
121.26 177.93 Reference
0.90
Table 6. Variables associated with increased length of stay: univariate analysis (n = 141) Variable
Lower
0.02 0.01 0.067
20.78
Table 7. Variables associated with increased length of stay: multivariate analysis (n = 141) P
0.001 0.009 0.02 0.031
Variable
Number of surgeons Blood transfusion Prognostic comorbidity Type of colloid
P
0.002 0.004 0.013 0.340
(BHH) and that two or more surgeons were present during the remaining 76 cases (54%). When multiple surgeons were involved, the senior author performed only the free flap reconstruction. Subsequent surgery during the first 30 days was performed in 33 (23%) of the 141 cases. Of the 15 secondary reconstructions the most common type was a pedicled myocutaneous flap, usually the pectoralis major. Wound healing and/or carotid protection was successful in all of these procedures. The most common donor site complications were skin graft losses and infection: two required repeat surgery, one for debridement and application of a second skin graft (forearm), and the other for hematoma drainage (fibula).
(Kaplan-Feinstein level 3) age over 55, and administration of more than 7 L of crystalloid during anesthesia (Table 4). Using multivariate analysis, we determined that administration of more than 7 L of crystalloid during surgery was significantly associated with major medical complications (Table 5). The odds ratio was 5:1. Age was also a significant variable, and subjects between ages 56 and 66 were 14 times more likely to have a major medical complication develop than those 55 and under. The odds were more than 21:1 when patients 67 or older were compared with those 55 and under. Comorbidity did not reach statistical significance, though its presence appeared to increase the absolute risk more than 4-fold.
Medical complications
Length of stay
Eighty (57%) of the 139 patients developed one or more medical complications within 30 days of reconstructive surgery, and they had 124 complications in all. The most common were cardiovascular, respiratory, and central nervous system complications, as shown in Fig. 1. We classified medical complications on a 4point (0-3) scale, which assigns a score of 3 to lifethreatening complications. Using the χ 2 test, we compared frequencies of complications when a variable was either present or absent. When we considered major (severity level 3) medical complications, variables that reached statistical significance (P < 0.05) were major comorbidity
The median length of stay was 11 days, with a minimum of 5 days and a maximum of 41 days. The χ2 test associated perioperative blood transfusions, recipient site (hypopharynx vs oral cavity, oropharynx, or others), number of surgeons, and administration of albumin during surgery with length of stay (Table 6). For example, patients who underwent reconstruction of the hypopharynx stayed longer (median of 17 days) than the other sites (medians of 10-12 days). The multivariate analysis indicated independent association of blood transfusion, “prognostic” comorbidity (K-F = 3), and number of surgeons with increased length of stay (Table 7).
14
HAUGHEY et al
Otolaryngology– Head and Neck Surgery July 2001
Fig 2. Flap complications among 141 patients who underwent reconstructive head and neck surgery between 1992 and 1997. Forty of these patients developed a total of 64 complications.
Flap complications
These complications developed in 40 of 139 (29%) patients. The most common flap complications were fistula, wound dehiscence, and hematomas or seromas (Fig 2). There was no significant decrease in complication rate within the 5 years over which they were recorded. Table 8 summarizes the significant variables with the χ2 tests. Multivariate analysis of variables associated with any type of flap complication identified cigarette smoking during the 2 weeks before surgery as a significant variable. A large volume of crystalloid (>7 L) administered during surgery was also significant, but the involvement of more than one surgeon just failed to reach significance (Table 9). Loss of more than 10% of body weight before surgery was associated significantly with a major flap complication developing (one that required a return to the operating room), along with having more than one surgeon and smoking cigarettes. For major flap complications, multivariate analysis identified the same significant variables as univariate analysis: weight loss, having more than one surgeon, and smoking cigarettes (Table 10). DISCUSSION
We conclude that free tissue transfer is a reliable technique for head and neck reconstruction in a tertiary referral and teaching center. The perioperative mortality in our series of 241 patients was 2.1%, which is comparable with rates reported for other large series of free flap transfers to the head and neck (7.0%7, 2.0%8, 4.7%9, 6.3%10).
The donor sites in our series changed significantly over a decade. For mandibular reconstructions, we came to prefer the fibula over the iliac crest, mainly because it can be harvested easily under tourniquet control with little loss of blood. Also, its versatile cutaneous unit can be used for lining the aerodigestive tract and/or for skin coverage. It is, however, more prone to atherosclerosis in the arterial pedicle than the iliac crest or scapula. For repairing pharyngoesophageal defects, we now prefer fasciocutaneous flaps over jejunal flaps because, in our clinical judgment, there is less donor site morbidity. The possibility of surgically related abdominal hemorrhage or peritonitis is eliminated. Our detailed study of complications and statistical analyses identified several variables that associate with less satisfactory medical or surgical outcomes (Table 11). Administering a large volume of crystalloid during surgery was a risk factor for developing major medical complications and/or for any level of flap complication. Reasons for crystalloid administration include fluid replacement, transfusion with packed cells, and hypotension during surgery; interestingly, duration of anesthesia, which might co-vary with volume of crystalloids given, did not predict for complications, nor did the use of blood tranfusions or pressors. The pathologic results of increased crystalloid administration could include fluid overload and its cardiorespiratory consequences. Administration of a large volume of crystalloid during surgery was also associated with flap complications. One hypothesis is that excessive edema in the flap and/or the recipient site promotes swelling and mechan-
Otolaryngology– Head and Neck Surgery Volume 125 Number 1
HAUGHEY et al
15
Table 8. Variables associated with all flap complications: univariate analysis (n = 141) P (Any complication) 60/141
Variable
Cigarette smoking 2 weeks before surgery More than 1 surgeon Major (level 3) medical complications Alcohol during the 2 weeks before surgery Intraoperative pressors Large volume of crystalloid (>7 L) Weight loss >10%
0.001 0.041 0.042 0.057 0.083 0.149 0.419
P (Major complications) 8/141
Variable
Weight loss >10% More than 1 surgeon Cigarette smoking 2 weeks before surgery Large volume of crystalloid (>7 L) Intraoperative pressors Alcohol 2 weeks before surgery Major (level 3) medical complications
0.023 0.030 0.036 0.138 0.164 0.585
Table 9. Variables associated with any flap complication: multivariate analysis (n = 141) 95% CI Variable
Cigarette smoking Volume of crystalloid Number of surgeons Volume of blood lost Intraoperative pressors Alcohol consumption
Category
No Yes <7 L >7 L 1 2 or more <1 L >1 L No Yes No Yes
Adjusted odds ratio
1 5.33 1 2.75 1 2.19 1 0.48 1 2.32 1 1.14
ically stresses the pedicles. The 5 patients in the detailed, 141-case database whose flaps failed had a median of 9.5 (3.7-11.0) L of crystalloid given. The low, 3.7-L value was in a patient whose flap failed because of mechanical separation of the vein in an alcoholic withdrawal crisis. Edema in recipient sites and flaps might also disrupt the closure lines. In response to this association, we recommend to our anaesthetist colleagues that crystalloid volume be minimized within the limits of adequate vital organ perfusion. It is plausible that >7 L of crystalloid was a marker for anesthetic practice, a variable not included in our study. Another statistically significant variable for medical complications, advancing age, suggests that extra attention should be paid to the preoperative and postoperative management of older patients. A careful presurgical medical evaluation, preferably with the internal medical surveillance postoperatively, would be ideal in patients over 55 in view of the likelihood that major surgery will unmask physiologic vulnerability. Our study differs from an Australian study11 that failed to find a significant difference in postoperative complication rates between patients over and under 70 who were stratified
Lower
Upper
Reference 1.84
0.002 15.42
Reference 1.13
0.025 6.69
Reference 0.99
0.053 4.84
Reference 0.20
0.098 1.15
Reference 0.88
0.121 6.70
Reference 0.37
P
0.814 3.49
for preoperative comorbidity. However, in that study the absolute numbers of medical complications were still 29% in the under-70 group and 54% in the over-70 group. A British study found that older men were at risk for postoperative chest infection.9 Cigarette smoking within 2 weeks of surgery was also associated with flap complications. In this respect, our findings differ from those of several other series9,12 in which smoking did not appear to be a risk factor for flap complications; however, these series did not specify proximity of smoking to surgery. There is evidence, however, from other clinical literature and laboratory work that tobacco products inhibit healing and promote tissue loss in flaps.13 Therefore we recommend physicians discourage smoking in patients who are about to undergo free flap head and neck reconstruction. The finding that both length of stay and major flap complications increase significantly when more than one surgeon is involved is open to several interpretations. One possibility is that outcomes improve when the responsibility for preoperative selection and postoperative management decisions rests with one person. This finding encourages improved
16
Otolaryngology– Head and Neck Surgery July 2001
HAUGHEY et al
Table 10. Multivariate analysis of association with major flap complications (n = 8/141) 95% CI Variable
Category
Weight loss >10% Number of surgeons Cigarette smoking Volume of crystalloid
Adjusted odds ratio
No Yes 1 2 or more No Yes <7 L >7 L
Lower
1 3.62 1 4.22 1 3.23 1 2.33
Upper
Reference 1.21
P
0.022 10.83
Reference 1.22
0.023 14.78
Reference 1.08
0.036 9.68
Reference 0.76
0.138 7.23
Table 11. Summary of multivariate analyses Outcome Variable
Flap complication (any)
Cigarette smoking Volume crystalloid >7 L Two of more surgeons Weight loss (> 10%) Prognostic co-morbidity Blood transfusion Age >55 years
** **
Flap complication (major)
Medical complication (major) Length of Stay
** ** ** **
communication and peer review of cases selected for this type of surgery. Because there is no other study that controls for this variable, we cannot compare our results with other institutions. However, other centers do report comparable flap success rates with multiple surgeon teams.10 The association of patient weight loss with major flap complications may be explained by the effect of nutritional deficits on wound healing14; this continues to be a variable we try to eliminate by hyperalimentation before surgery. Pertinent negative findings were that preoperative irradiation had no association with medical or surgical outcomes even though radiotherapy is known to damage blood vessels and delay healing. A negative effect of irradiation has been reported in some studies9,15,16 but not in others.17 Duration of surgery did not influence medical or surgical outcomes; therefore, the adverse medical effects associated with high crystalloid volumes and two or more surgeons were unlikely to have resulted from prolongation of procedures. Our finding that age fails to affect flap outcomes agrees with several previous studies,7,17,18 but we demonstrated the significant effect of advancing age on major medical complications. Finally, the failure of all but major
**
** ** **
comorbidity using the Kaplan-Feinstein scale to predict complications probably speaks to case selection and our routine practice of seeking an internal medical consultation before free flap operations. This is at variance with another study that used different (Charleson) comorbidity index; this study did not distinguish complications as medical or surgical, as we did, but rather grouped complication severity and types.16 CONCLUSION
We will continue to offer microvascular free tissue transfer as the preferred method of reconstruction in appropriately selected patients with head and neck disease. Cognizance of potentially significant variables uncovered by this study should help optimize future outcomes and generates hypotheses for future investigation. Microvascular free flap transfers to the head and neck are known to be safe, but enhanced attention to patient selection, preparation of patients for surgery, fluid balance, and coordination between members of the care team should increase effectiveness and safety even further. REFERENCES 1. Seidenberg B. Immediate reconstruction of the cervical esopha-
Otolaryngology– Head and Neck Surgery Volume 125 Number 1
2. 3. 4.
5. 6. 7. 8. 9. 10.
gus by a revascularized isolated jejunal segment. Ann Surg 1959;149:162. Taylor GI, Razaboni RM. Michel Salmon’s anatomical studies. Book 1. Arteries of the muscles of the extremities and the trunk. St Louis: Quality Medical Publishing; 1994. O’Brien BMcC. Microvascular reconstructive surgery. Edinburgh, London, New York: Churchill Livingstone; 1977. Kroll SS, Schusterman MA, Reece GP. Comparison of the rectus abdominis free flap with the pectoralis major myocutaneous flap for reconstructions in the head and neck. Am J Surg 1992;164:615-8. Kroll SS, Schusterman MA, Reece GP. Costs and complications in mandibular reconstruction. Ann Plast Surg 1992;29:341-7. Kaplan MH, Feinstein AR. The importance of classifying initial co-morbidity in evaluating the outcome of diabetes mellitus. J Chron Dis 1974;27:387-404. Shestak KC, Jones NF. Microsurgical free tissue transfer in the elderly patient. Plast Reconstr Surg 1991;88:259-63. McNamara M, Pope S, Sadler A, et al. Microvascular free flaps in head and neck surgery. J Laryngol Otol 1994;108:962-8. Simpson KH, Murphy PG, Hopkins PM, et al. Prediction of outcomes in 150 patients having microvascular free tissue transfers to the head and neck. Br J Plast Surg 1996;49:267-73. Jones NF, Johnson JT, Shestak KC, et al. Microsurgical reconstruction of the head and neck: interdisciplinary collaboration
HAUGHEY et al
11. 12. 13. 14. 15.
16. 17. 18.
17
between head and neck surgeons and plastic surgeons in 305 cases. Ann Plast Surg 1996;36:37-43. Bridger AG, O’Brien CJ, Lee KK. Advanced patient age should not preclude the use of free-flap reconstruction for head and neck cancer. Am J Surg 1994;168:425-8. Reus WF, Colen LB, Straker DJ. Tobacco smoking and complications in elective microsurgery. Plast Reconstr Surg 1992;89:490-4. Chang LD, Buncke G, Slezak S, et al. Cigarette smoking, plastic surgery and microsurgery. J Reconstr Microsurg 1996;12:465-74. Steidelman WK, Digenis AG, Tobin GR. Impediments to wound healing. Am J Surg 1998;176(2A Suppl):395-475. Bengston BP, Schusterman MA, Baldwin BJ, et al. Influence of prior radiotherapy on the development of postoperative complications and success of free flaps in head and neck reconstruction. Am J Surg 1993;166:326-30. Singh B, Cordeiro PG, Santamaria E, et al. Factors associated with complications in microvascular reconstruction of head and neck defects. Plast Reconstr Surg 1999;103:403-11. Kroll SS, Robb GL, Reece GP. Does prior irradiation increase the risk of total or partial free flap loss? J Reconstr Microsurg 1998;14:263-8. Chick LR, Watson RL, Reus W, et al. Free flaps in the elderly. Plast Reconstr Surg 1992;1:87-94.