Chepeha DB, Annich G, Pynnonen MA, Beck J, Wolf GT, Teknos TN, Bradford CR, Carroll WR, Esclamado RM. Pectoralis Major Myocutaneous Flap vs Revascularized Free Tissue TransferComplications, Gastrostomy Tube Dependence, and Hospitalization. Arch Otolaryngol Head Neck Surg. 2004;130(2):181-186. doi:10.1001/archotol.130.2.181
To evaluate the factors related to surgical complications, rate of gastrostomy tube (G-tube) dependence, and hospitalization in patients undergoing reconstruction with a pectoralis myocutaneous flap vs a soft-tissue revascularized flap.
Quasi-experimental case series with a historic control group.
A total of 179 patients (138 men and 41 women) with a mean (SD) age of 58 (14) years treated between January 1, 1986, and December 31, 1995, with a pectoralis flap (108 patients) or a revascularized free flap (71 patients).
Inclusion criteria were first or second extirpation, reconstruction with soft-tissue flap, or defect including the upper aerodigestive tract. Exclusion criteria were secondary reconstruction, or reconstruction for salvage of a complication.
Although the major complication rate was not significantly different according to reconstructive approach, hypopharyngeal defects had a significantly higher major complication rate of 30% (6/20) compared with 8% (13/159) for other defect sites (P<.003). The minor complication rate was higher in the pectoralis group, at 57% (62/108), than in the revascularized flap group, at 21% (15/71) (P<.001). G-tube dependence was higher in the pectoralis group at 42% (40/96), in contrast to the revascularized flap group at 16% (10/63) (P<.001). G-tube dependence was 25% higher in patients who underwent salvage surgery after radiation (42% [30/72]) than in patients treated with postoperative radiation (17% [12/69]) (P<.004). Revascularized flaps helped ameliorate the effects of radiation before surgery; 56% (23/41) of the patients who received pectoralis flaps were G-tube dependent, while the rate of G-tube dependence in the revascularized flap group was 23% (7/31) (P<.004). Hospitalization was longer in the pectoralis group (14 days) than the revascularized flap group (12 days) (P<.006).
Patients who undergo reconstruction with a pectoralis flap have significantly higher minor complication rates, a higher rate of G-tube dependence, and longer hospitalization than patients who undergo reconstruction with a soft-tissue revascularized flap.
The pectoralis major myocutaneous flap (PMMF) transformed head and neck reconstruction by providing a source of vascularized tissue, which resulted in decreased complication rates and improved functional outcome.1- 5 During the past decade, revascularized free tissue transfer (RFTT) has been performed with increasing frequency in an attempt to improve the functional and aesthetic results of life-altering head and neck extirpations.6- 9 Despite the increased use of the RFTT, the exact role of this reconstructive approach has yet to be established. Previous publications comparing these 2 reconstructive approaches have been limited by either small patient populations or unbalanced comparison groups.10- 13
The current study was undertaken to provide information to assist the head and neck surgeon in choosing a reconstructive approach based on statistical analysis of demographic, clinical, and treatment variables. The hypothesis was that soft-tissue RFTT reduces surgical complications, rate of gastrostomy tube (G-tube) dependence, and duration of hospitalization compared with PMMFs.
A review was conducted to evaluate the effect of PMMF vs soft-tissue RFTT on surgical complications, rate of G-tube dependence, and duration of hospitalization. The chart review included patients who underwent a reconstructive procedure after an oncologic extirpation from January 1, 1986, through December 31, 1995. This period was chosen to reduce historical bias, which can favor the most recent techniques.14 Individual surgeon bias was expected to be limited, as most of the PMMFs were performed by surgeons facile with RFTT.
Patients were included if they underwent a first or second extirpation and required only soft tissue for reconstruction of their defect. Patients were excluded if they underwent secondary reconstruction, received a flap to salvage a complication, or had data missing relating to treatment or outcome variables. The inclusion and exclusion criteria were designed to optimize the comparison of these 2 reconstructive groups. These eligibility criteria resulted in a population of 179 patients.
Major complications, minor complications, and duration of hospitalization were analyzed for the entire sample group. The PMMF was performed in 108 patients and a soft-tissue RFTT was performed in 71 patients. The mean (SD) patient age was 58 (14) years. There were 138 men (70%) and 71 women (30%). The clinical demographic variables between the 2 groups were well balanced and showed no statistically significant differences (Table 1). The rate of G-tube dependence was analyzed in 159 patients, a subset of the 179 patients, whose head and neck reconstruction involved the upper aerodigestive tract.
The explanatory variable (surgical intervention) was either a PMMF or a soft-tissue RFTT. The technique of elevation of the PMMF involved either a curvilinear submammary incision (performed in some men and all women) or a direct incision from the lateral clavicle to the lateral superior part of the skin paddle. The portion of the skin paddle that was random varied with the body habitus, location of the defect to be reconstructed, and size of the skin paddle. Guiding principles were minimization of tension on the pedicle, minimization of shear of the skin paddle on elevation and inset, placement of additional stay sutures on the pectoralis muscle on inset, and denervation of the pectoralis muscle.
Patients who underwent soft-tissue fasciocutaneous and myocutaneous RFTT formed the comparison group. Of these patients, 30 (42%) received a radial free forearm flap, 18 (25%) received a rectus flap, 14 (20%) received a lateral arm flap, 5 (7%) received a latissimus flap, and 4 (6%) received a scapular flap. An operating microscope, frame clamp, and 9.0 or 10.0 nylon suture were used to perform each RFTT. Dextran 40 was administered during revascularization and was maintained at 25 to 30 mL/h, with aspirin starting on postoperative day 1 at a dose of 81 mg.
Minor complications were defined as those that required treatment such as packing, drainage, or medication, but did not require a procedure in the operating room. Minor complications included wound infection, partial flap necrosis, partial flap dehiscence, fistula, plate exposure, hematoma, wound abscess, seroma, and postoperative hemorrhage. If complications such as partial flap necrosis or flap dehiscence did not require any treatment whatsoever, they were not included as a complication. If any wound infection or breakdown resulted in a fistula, then "fistula" was coded as the outcome in an effort to represent the overall outcome of a series of complications in an individual patient.
Major complications were defined as those that required a procedure in the operating theater or resulted in death. If a fistula caused a carotid rupture leading to death, the complication was coded as death.
G-tube dependence was defined as present in patients who required a G-tube for part or all of their nutritional needs. Patients were classified as eating by mouth if they were able to have their G-tube removed. This measure of functionality of the reconstruction was chosen because it was reliably documented in the chart. In addition, rate of G-tube dependence is a very reliable indication of swallowing function. A small subset of patients who were able to maintain full nutrition by mouth retained their G-tube so that they could obtain formula feeds at a reduced cost. These patients were coded as eating by mouth.
Duration of hospitalization was measured in days. A larger proportion of PMMFs were performed at the beginning of the study period and a larger proportion of the RFTTs were performed at the end of the study period. All inpatient admissions to the otolaryngology service from 1986 through 1995 were evaluated by case mix and duration of hospitalization to assess the risk of historical bias. During this period, the mean duration of hospitalization on the otolaryngology service was stable in a range between 4.6 and 5.1 days. Although there was a small change in case mix toward higher-acuity cases at the end of the study period, the duration of hospitalization remained stable overall. The narrow range of "duration of hospitalization" for the otolaryngology service suggests that historical bias is minimized in this sample. The date of surgery was run as an independent variable in each of the multivariable equations presented in this article and was not significant in any instance.
The independent variables used in the analysis include the reconstructive approach (PMMF or RFTT), date of surgery, age (years), sex, defect, radiation (preoperative, postoperative, or none), stage (stage I-II or stage III-IV), albumin level, and total red blood cell transfusion. The "preoperative radiation" group underwent salvage surgery after a recurrence rather than planned postradiation surgery. The defect was coded to include each involved site so that both the extent and the location could be evaluated. The defects could involve up to all of 5 possible sites: oral tongue, oral cavity other than oral tongue, oropharynx, hypopharynx, and cutaneous (skin). Because of the large number of independent variables under study, more extensive breakdown of the defect coding would not be statistically robust. Partial flap dehiscence, partial flap necrosis, and fistula were also used as independent variables to evaluate their contribution to the rate of G-tube dependence. Major and minor complications were used as independent variables to evaluate their contribution to duration of hospitalization.
Univariate data were tabulated for all the dependent and independent variables. Bivariate comparisons of PMMF and RFTT were measured for each outcome (major complications, minor complications, G-tube dependence, and hospitalization). The χ2 statistic and Fisher exact test were used for the bivariate outcomes and the Mann-Whitney test was used to compare duration of hospitalization.
Logistic regression analysis was used for all binary outcomes, and multivariable linear regression analysis was used for duration of hospitalization. Backward stepwise analysis was used starting with all the independent (predictor) variables. Equations were chosen that minimized error, minimized the number of variables, maximized the P value, and were most clinically appropriate. Although each independent variable in a given equation may not be independently significant, the strongest equation as outlined above was chosen. Duration of hospitalization was not normally distributed but, when transformed with the natural log, it showed an adequate fit with residual analysis. Therefore, the regression for duration of hospitalization was run with the log-transformed duration of hospitalization.
Data were stored in a FileMaker Pro 3.0 relational database (FileMaker, Inc, Santa Clara, Calif) and exported as a comma-delimited text to SAS version 6.12 (SAS Institute Inc, Cary, NC). The data were then analyzed on a Macintosh G3 Computer (Apple Computer, Inc, Cupertino, Calif).
Overall, the major and minor complication rates were lower in the RFTT group. The hypopharynx site was associated with the highest complication rate. The PMMF group had a significantly higher rate of G-tube dependence and a significantly longer duration of hospitalization. Because of the size of and balance between the 2 reconstructive groups, the results should be statistically robust.
Nineteen (11%) of 179 patients experienced a major complication. Patients who underwent a PMMF experienced a 12% (13/108) major complication rate, compared with an 8% (6/71) major complication rate in the RFTT group; however, this difference was not statistically significant (Table 2). The only significant independent (predictor) variable for a major complication was a defect involving the hypopharynx, which was associated with a 30% (6/20) complication rate (P<.003) (Table 3). The odds ratio (95% confidence interval) of developing a major complication for patients with a hypopharyngeal defect compared with all other defect locations in patients in this cohort was 4.8 (1.6-14.6) (P<.006) (Table 4). Multivariable logistic regression did not identify additional independent (predictor) variables associated with major complications. It is likely that a number of factors in addition to the hypopharyngeal site, such as surgical management, anesthetic management, preexisting comorbidities, and nutritional status (albumin was a variable in the logistic regression), contributed to major complications. Given that there were 19 major complications, it is beyond the scope of this study to evaluate additional independent variables.
Of the 179 patients, 77 (43%) experienced minor complications. The minor complication rate was 57% (62/108) in the PMMF group and 21% (15/71) in the RFTT group (P<.001). A breakdown of the minor complication group is contained in (Table 5). Nine subgroups composed the minor complication group. The rates of wound infection, partial flap necrosis requiring medical treatment, and partial flap dehiscence requiring medical treatment were all higher in the PMMF group and were all statistically significant. The odds ratio (95% confidence interval) of developing a minor complication in a PMMF was 5.0 (2.5-10.0) times greater than with an RFTT. Although additional variables did not generate a better model, female sex and hypopharynx site were the next 2 most important predictors of minor flap complications (Table 4).
Of the 159 patients whose head and neck reconstruction involved the upper aerodigestive tract, 50 (31%) were G-tube dependent. The rate of G-tube dependence was higher in the PMMF group at 42% (40/96) compared with 16% (10/63) in the RFTT group (P<.001) (Table 6).
The rate of G-tube dependence was higher in patients who underwent preoperative radiation compared with patients who underwent postoperative radiation. Overall, 89% (141/159) of patients underwent preoperative or postoperative radiation. Only 18 patients received surgery as the only treatment modality. G-tube dependency in patients who underwent radiation preoperatively was 42% (30/72), whereas only 17% (12/69) of patients who received postoperative radiation therapy were G-tube dependent (P<.004) (Table 6).
The RFTT helped ameliorate some of the effects of radiation therapy. In all irradiated patients (preoperative or postoperative), the RFTT group experienced a substantially lower rate of G-tube dependence (16% [9/56]) than the PMMF group (39% [33/85]) (P<.001). Of the 72 patients who received radiation before surgery, the rate of G-tube dependence in the PMMF group was 56% (23/41), which was significantly higher than the 23% (7/31) rate of G-tube dependence for patients who underwent reconstruction with an RFTT (P<.004). Patients who underwent reconstruction with a PMMF and who had postoperative radiation also had a higher rate of G-tube dependence (23% [10/44]) than those who had RFTT and postoperative radiation therapy (8% [2/25]) (Fisher exact test, 2-tailed, P = .19) (Table 7). Overall, the rate of G-tube dependence was significantly higher in patients who received radiation therapy before surgery and was 2 to 3 times higher in patients who underwent reconstruction with a PMMF.
A defect involving the oropharynx site also adversely affected the rate of G-tube dependence. Of the 35% (55/159) of the patient population with an oropharyngeal defect, 47% (26/55) were G-tube dependent, compared with 23% (24/104) of patients whose defect did not include the oropharyngeal site (P<.02) (Table 6). Interestingly, there was little difference in reconstructive approach with respect to G-tube dependence in the oropharynx defect group. In contrast, patients with nonoropharynx defect sites had a higher rate of G-tube dependence in the PMMF group (36% [20/56]) than in the RFTT group (8% [4/48]) (P<.001) (Table 7).
The rate of G-tube dependence in patients with advanced stage III to IV disease was also higher (Table 6). Stage III to IV disease affected 77% (123/159) of the patient population. Of these patients with high-stage disease, 45% (34/76) in the pectoralis group, but only 15% (7/47) in the RFTT group, were G-tube dependent (P<.001) (Table 7).
Stepwise descending multivariable logistic regression for G-tube dependence was performed on the independent variables listed in the "Methods" section. A 4-variable equation including reconstructive approach, radiation, oropharynx defect, and stage (P<.001, −2 log 163.7) was selected as the most appropriate for this study. The logistic regression statistically controlled for these variables and produced an odds ratio (95% confidence interval) for G-tube dependence for each variable as follows: PMMF, 3.9 times (1.7-9.3 times) higher risk; preoperative radiation, 4.4 times (2.0-10.1 times) higher risk; oropharynx site, 2.7 times (1.2-5.8 times) higher risk; higher stage III to IV disease, 2.0 times (0.8-5.2 times) higher risk (Table 4). The PMMF, preoperative radiation, and oropharynx site were highly significant independent predictors (P<.002, P<.004, and P<.01, respectively) of increased rate of G-tube dependence. Although the variable "stage" strengthened the equation, it was not an independent predictor of G-tube dependence (P = .15).
The mean duration of hospitalization for the total patient sample was 15.6 days. The median duration of hospitalization for the PMMF group was 14 days, significantly longer than the 12 days for the RFTT group (P<.006).
The multivariable linear regression for duration of hospitalization, although significant (P<.001), had a low R2 (0.09). This brings into question the real strength of the association between reconstructive approach and hospitalization. This relatively weak equation showed that when major and minor complications were combined as a single variable, this variable was the most important factor in determining duration of hospitalization (Table 4).
Patients who received a PMMF experienced higher complication rates, increased G-tube dependence, and longer duration of hospitalization. Despite the apparent poor outcomes in patients who underwent reconstruction with PMMFs, the data suggest that PMMFs may have a role in smaller defects in the oropharynx in patients not previously treated with radiation when surgery is the sole treatment modality.
The value of this study in the context of existing literature is its large patient population and the balance of treatment, demographic, and clinical variables. Previous studies have evaluated smaller patient populations, have not been as evenly balanced (particularly with respect to treatment variables), or have not accounted for the location of the defect.
The complication rate experienced in this patient population is similar to that in the published literature.4,8,9,15- 17 The finding that the complication rate in the PMMF group was significantly higher is also supported by current literature.10- 12 The higher complication rate in PMMF results from increased incidence of partial flap dehiscence and necrosis that further result in wound abscess and fistula. To overcome these complications, healing by secondary intention and/or reoperation may be required, which can prolong hospitalization and decrease the ability of the reconstructive head and neck surgeon to control the functional outcome.
A previous study found that, for posterior oral cavity and oropharyngeal defects, the rate of G-tube dependence was 85% in the PMMF group and 39% in the RFTT group (P = .002).13 Few studies have thoroughly reviewed the impact of PMMF compared with RFTT on oral intake. Rate of G-tube dependence was selected as the outcome variable to evaluate oral intake because it is a reliable objective measure of function and it unfortunately occurs with sufficient frequency to analyze most independent variables of interest. The results were intuitively satisfying as they confirmed several widely held surgical beliefs. Patients have better swallowing function if they undergo reconstruction with an RFTT, receive postoperative radiation as opposed to surgical salvage for a recurrence after radiation, and their defect is smaller or does not involve the oropharynx. It is interesting that reconstructive approach affected each important independent predictor of a rate of G-tube dependence with the exception of the oropharyngeal defect. The high rate of G-tube dependence among patients with an oropharynx defect may relate more to the inherent difficulties with restoring the complex neuromuscular function of this region, which coordinates the swallow and protects both the airway and the nasopharynx. These data suggest that functional restoration of the oropharynx is independent of reconstructive approach and requires more than contouring and volume restoration.
We hypothesized that the higher rate of G-tube dependence in the PMMF group may be related to scar formation from partial flap necrosis, flap dehiscence, or fistula. Contrary to our hypothesis, these complications were not independent predictors of swallowing function and did not represent the most important variables in the logistic regression. These findings corroborate the clinical impression of surgeons whose flap of choice is the PMMF and who believe that flap dehiscence and partial flap necrosis heal without sequelae. This still leaves a question as to why G-tube dependence is higher in PMMF. Perhaps it is due to the intrinsic characteristics of the flap, such as its small volume, limited ability to fold, limited axis of rotation, and tendency for downward pull. If these are the limitations, a latissimus dorsi musculocutaneous rotational flap would provide an alternative to the PMMF.
Hospitalization was significantly shorter in the RFTT group. Although PMMFs were performed during the entire 9-year study period, RFTT replaced PMMF as the reconstructive approach of choice during this time. In most cases, the same surgeons who performed PMMF were also microvascular surgeons and, as a result, should have had similar approaches to inpatient management. In addition, the mean duration of hospitalization for otolaryngology patients was stable during this period, as described in the "Methods" section. The median hospitalization duration was 14 days for the PMMF group and 12 days for the RFTT group (P<.006). The present duration of hospitalization for patients after head and neck surgery is much shorter.
The impact of reconstructive approach on hospitalization can also be examined by reviewing the complication rate. As previously discussed, the PMMF has a higher complication rate. If it is accepted that patients are rapidly discharged unless they have a complication, it can be concluded that the significantly longer hospitalization for patients in the PMMF group may apply during the present day.
Soft-tissue RFTT significantly reduced surgical complications, rate of G-tube dependence, and duration of hospitalization in patients undergoing major extirpative surgery for head and neck cancer. Although RFTT significantly reduced the minor complication rate, there was no significant difference in the major complication rate by reconstructive approach. The predictor for major complications was the hypopharynx defect site. Logistic regression of swallowing function showed that, in addition to reconstructive approach, preoperative radiation and an oropharynx defect were key independent predictors for increased rate of G-tube dependence. Stage III to IV disease was an important variable in the logistic regression but not an independent predictor for increased rate of G-tube dependence. The PMMF may have a role in small oropharyngeal defects in patients who receive surgery as their sole treatment modality. Even so, the controlled odds ratio of G-tube dependence would be 4 times greater. When a PMMF was used for a defect outside the oropharynx in a patient with stage III to IV disease who had preoperative radiation, a complication was 5 times more likely, G-tube dependence was 13 times greater, and the hospital stay was significantly longer. Data from this study strongly suggest that soft-tissue reconstruction with RFTT is a better approach than the PMMF; however, it does not conclude that revascularized free tissue is the only approach to head and neck reconstruction. If a reconstructive surgeon does choose to perform an axial rotational flap, other options such as the latissimus flap, which may have fewer limitations than the PMMF, should be considered.
Corresponding author and reprints: Douglas B. Chepeha, MD, MSPH, Department of Otolaryngology–Head & Neck Surgery, 1500 E Medical Center Dr, Taubman Center 1904, University of Michigan, Ann Arbor, MI 48109-0312 (e-mail: firstname.lastname@example.org).
Submitted for publication July 3, 2002; final revision received May 2, 2003; accepted May 20, 2003.
Shan Baker, MD, and Michael Sullivan, MD, contributed records of patients who underwent PMMF. Julia Shin-Jung Lee, MS, Department of Biostatistics at the University of Michigan, Ann Arbor, evaluated the data contained in this article.