Kerstin M. Stenson, Danial J. Haraf, Harold Pelzer, Wendy Recant, Merrill S. Kies, Ralph R. Weichselbaum, Everett E. Vokes. The Role of Cervical Lymphadenectomy After Aggressive Concomitant ChemoradiotherapyThe Feasibility of Selective Neck Dissection. Arch Otolaryngol Head Neck Surg. 2000;126(8):950–956. doi:10.1001/archotol.126.8.950
To evaluate the necessity, technical feasibility, and complication rate of neck dissection performed on patients with head and neck cancer after 5 cycles of concomitant chemoradiotherapy (CRT) and to justify a selective neck dissection (SND) approach and define the optimal timing of post-CRT neck dissection.
Design and Setting
Retrospective analysis in an academic university medical center.
Sixty-nine eligible patients with advanced (stage III and IV) head and neck cancer who have undergone 1 of 4 CRT protocols. Patients ranged in age from 36 to 75 years, and surgical procedures were performed over a 4-year period. Follow-up ranged from 6 to 64 months.
Neck dissection (most commonly unilateral SND) performed within 5 to 17 weeks after CRT completion.
Main Outcome Measures
Complication rate and incidence of positive pathology (viable cancer) in pathologic neck dissection specimens.
Seven (10%) of 69 patients developed wound healing complications, 4 (6%) of whom required surgical intervention for ultimate closure. There were no wound infections. Other complications occurred in 11 (16%) of 69 patients and included need for tracheotomy, nerve transection and paresis, and permanent hypocalcemia. Twenty-four (35%) of 69 patients revealed microscopic residual disease. Ten (50%) of 20 patients with N3 neck disease had positive pathology, whereas 14 (36%) of 39 patients with N2 disease had viable carcinoma in the dissection specimen (P = .09 by χ2 analysis). There was no significant relation between radiologic complete response or partial response and residual microscopic cancer. In 1 patient, disease recurred in the neck after dissection. Mean follow-up time was 30.3 months.
(1) Neck dissection for patients with N2 or greater neck disease after CRT is necessary to eradicate residual disease. (2) The complication rate of SND after CRT with hyperfractionated radiotherapy is low. (3) SNDs are technically feasible when performed within the "window" between the acute and chronic CRT injury (4-12 weeks). (4) SNDs, rather than more radical procedures, appear to be therapeutically appropriate in this group of patients because of the low incidence of disease recurrence in the neck.
MULTIMODALITY treatment for patients with advanced head and neck cancer has been studied for years.1,2 Early in the surgical literature, preoperative chemotherapy and radiotherapy were combined with radical surgery with low toxicities and improved survival times.2- 4 More recently, the goals of multimodality therapy have encompassed organ preservation through less radical surgery in addition to increased survival.5- 9
Concomitant chemoradiotherapy (CRT) is a promising approach that treats micrometastatic systemic disease while enhancing the locoregional efficacy of radiotherapy.10- 12 The ultimate goals of this strategy are organ and function preservation, elimination of systemic micrometastases, and improved survival. Studies have shown the feasibility of such treatments, with both standard and hyperfractionated radiotherapy.13- 16 Although organ preservation is achieved in most patients, toxicities and the concerns of increased surgical complications after CRT remain high. At the University of Chicago, Northwestern University, and University of Illinois, neck dissection for N2 or greater disease is frequently performed after CRT because of the possibility of residual gross or microscopic disease. The purpose of this report is to study the necessity, the technical feasibility, and the complication rate of neck dissection after CRT with hyperfractionated (twice-daily) radiation. Other comments will focus on the timing and the type of neck dissection performed.
Two hundred thirty-seven patients following several CRT protocols from July 1993 through March 1999 were evaluable for this study. One hundred seventy-seven were staged with N2 or N3 neck disease. Sixty-nine patients had neck dissections (Table 1). Chemotherapeutic drug combinations included paclitaxel, fluorouracil, and hydroxyurea (with or without granulocyte colony-stimulating factor [GCSF]); and cisplatin, fluorouracil, and hydroxyurea (with or without GCSF). All drugs are radiosensitizers. All CRT protocols consisted of 5 cycles (week on, week off) of chemotherapy plus concomitant twice-daily radiation. Total radiation dose was 6650 to 7550 rads (6.65 to 7.55 Gy). No patient received previous radiotherapy or surgery prior to starting treatment. Patients 41, 44, and 45 received previous chemotherapy before CRT. All patients had stage III or IV head and neck cancer and underwent operative staging of their tumors prior to therapy. Sixty-nine patients underwent planned neck dissection after CRT. Fifty-nine patients underwent surgery at the University of Chicago, 54 of the patients were operated on by the corresponding author (K.M.S.). Sixteen patients were women and 59 were men. They ranged in age from 36 to 75 years. All histologic findings were squamous cell carcinoma, with the exception of patients 50 (adenoid cystic carcinoma), 59 (poorly differentiated mucoepidermoid carcinoma), and 68 (adenocarcinoma). Posttreatment computed tomographic (CT) scans were done 4 weeks after completion of CRT to assess radiologic response or progression of disease. A designation of complete response (CR) was given if there was no radiologic evidence of disease at the primary site and neck areas. Partial response (PR) was assigned to patients who had radiologic evidence of disease at the primary site or neck area. Before neck dissection, biopsy of the primary site and frozen section pathologic analysis were performed. Neck dissection was performed after pathologic documentation of complete response (ie, negative for carcinoma) at the primary site. If a patient had positive pathology at the primary site, he or she was not eligible for this study, since resection of the primary site and neck would be required. The CT scans were referred to regularly to aid in selection of neck dissection type. For example, if tissue planes between carotid sheath and sternocleidomastoid muscle were not discrete or if the jugular vein could not be identified continuously from the thoracic inlet to the skull base, a modified or radical neck dissection was planned.
No patient had upper aerodigestive tract entry. Fifty patients underwent unilateral selective (n = 42) or modified radical neck dissection type II or III (n = 8) after CRT17,18 (Table 2). Four patients (12, 24, 47, and 68) underwent radical neck dissection or modified radical neck dissection type I (68) with pectoralis myofascial flap placement over exposed carotid artery. One patient (18) underwent thyroid lobectomy alone. Fourteen patients (23, 26, 27, 33, 46, 50, 53, 58, 61-64, 66, and 69) underwent bilateral selective neck dissections (SNDs), 1 of whom (26) also underwent thyroid lobectomy for a large goiterous nodule and another (27) who also underwent total thyroidectomy. A total of 56 patients underwent SND (levels I-III, I-IV, or rarely II-V), preserving sternocleidomastoid muscle, internal jugular vein, and 11th cranial nerve. With respect to SND, level V was dissected only if there was gross disease present in this location before treatment.
Forty-seven patients had N2 neck disease (3 with N2a, 23 with N2b, and 21 with N2c), and 20 patients had N3 neck disease. Two patients with N1 disease (18 and 42) underwent post-CRT surgery (thyroid lobectomy and SND, respectively) and were included in this series because of residual suspicious findings on posttreatment CT scan. Patient 18 had a residual thyroid mass with previous (pre-CRT) carcinoma-positive fine needle aspiration. Patient 42 had a suspicious level 2 node on posttreatment CT scan. No other patient with N1 disease underwent post-CRT neck dissection. Every effort was made to document thyroid function (via thyrotropin and total thyroxine) and provide supplemental thyroid hormone if necessary. All patients were given intraoperative dexamethasone sodium phosphate with a rapid postoperative taper to prevent upper airway edema. All patients were given intravenous antibiotics (ampicillin sodium–sulbactam sodium) or clindamycin phosphate (if penicillin allergic) during the operation and hospital stay and were discharged with a regimen of oral or gastrostomy tube amoxicillin–clauvulanate potassium or clindamycin. All operations were performed in a university hospital resident teaching environment. Wound healing complications were documented in the outpatient chart, recorded in the database, and appropriately treated. Other surgical complications were dictated in operative notes and recorded. Length of time between completion of CRT and neck dissection was calculated for each patient and recorded. Pathology reports of neck dissection specimens were reviewed and recorded in the database.
Statistical studies were performed using either χ2 or Fisher exact analyses.
Thirty patients had a radiologic (CT scan) CR, and 33 patients had PR (Table 3). Six patients did not have a radiologic response reported. The mean time elapsed between end of CRT and neck dissection was 8.01 weeks (range, 5-17 weeks). Every effort was made to perform neck dissection within 8 weeks after completion of CRT. Several patients were operated on after this period to allow for healing of neck skin wet desquamation/radiation injury. Follow-up time ranged from 6 to 64 months. No patients developed primary wound infections.
Sixty-one patients had no postoperative wound healing complications. There were 7 patients (10%) who experienced wound complications. Three patients (4%) (14, 28, and 58) had suture line dehiscences (1 secondary to underlying seroma) that healed with local wound care. Four patients (6%) (6, 13, 61, and 63) experienced a major wound dehiscence, requiring additional surgery (usually pectoralis flap) for closure and ultimate healing. These major wound dehiscences were secondary to underlying hematoma (1 patient; skin flaps did not adhere after evacuation) and overall poor wound healing and dehiscence with active neck extension (3 patients). No patients experienced a chylous fistula or flap necrosis. The average length of time from CRT end to neck dissection in those patients who experienced wound healing complications was 8.14 weeks. This figure closely approximates the average length of time for those patients who did not experience complications (8.01 weeks).
Eleven patients (16%) experienced other complications not related to wound healing. Two patients (4 and 19) had inadvertent transection of the 11th cranial nerve. Both of these patients had normal shoulder function postoperatively due to preservation of the cervical motor nerve roots. One patient (31) had inadvertent transection of the 12th cranial nerve. Although she complains of tongue swelling on the side of the transection, she has maintained her weight with an oral diet. Patient 41 developed a right vocal cord paralysis and temporary Horner syndrome after dissection of the right side of the neck, although neither the vagus nerve nor sympathetic trunk was obviously injured. This patient's vocal cord is still paralyzed, although he has excellent compensation and speech. Patient 27, who underwent neck dissections and thyroidectomy, developed postoperative hypocalcemia, which did not resolve. Seven patients (23, 24, 27, 42, 43, 48, and 62) required postoperative tracheotomy secondary to oropharyngeal or supraglottic edema. One patient required urgent tracheotomy 2 hours after neck dissection. Five other patients required tracheotomy 3 to 8 months postoperatively because of progressive airway obstruction unrelated to primary site tumor recurrence. Total nonwound healing complications (nerve transection, hypocalcemia, tracheotomy tube placement) were 11 (16%) of 69 patients. Total complication rate (wound plus other complications) in this group of patients was 18 (26%) of 69 patients.
Twenty-four (35%) of 69 patients revealed microscopic residual disease. Forty-five patients showed no residual viable carcinoma in the pathology specimen (Table 4). Ten (50%) of 20 patients with N3 neck disease had positive pathology, while only 14 (29%) of 49 patients with N1,2 disease had viable carcinoma in the dissection specimen. No patient with N1 or N2a (n = 5) disease had positive pathology, leaving 13 (30%) of 44 patients with N2b,c disease with positive pathology. This difference between the N3 and N1,2 groups, ie, the likelihood of having positive pathology, is not statistically significant (P = .09). There was no gross tumor visible at surgery.
The relation between radiologic response (PR or CR) and pathologic response (positive or negative) was evaluated for all patients (6 patients did not have a radiologic response assigned) (Table 5). There was no significant relation between radiologic CR or PR and residual microscopic cancer.
Seven patients (20, 29, 43, 46, 48, 54, and 58) developed recurrent disease at the primary site. In only 1 patient (40) did the disease recur in the neck after neck dissection; the patient refused further treatment. Thirteen patients (1, 3, 12, 20, 26, 27, 28, 29, 30, 31, 45, 47, and 68) undergoing post-CRT neck dissection developed distant metastases, and 8 have died 5 to 10 months postoperatively. One patient (10) is lost to follow-up after 17 months. One patient (9) died of cardiac causes at an outside institution approximately 10 months after neck dissection, and another (5) died of unknown causes at home approximately 2½ years after neck dissection. Three other patients died at home, 1 (23) from a drug (cocaine) overdose and 2 (6 and 24) from presumed tracheotomy tube obstruction.
This study challenges the impression held by some that surgery after radiation or CRT may be associated with high complication rates. In addition, SND appears to be technically feasible and an adequate surgical procedure for removal of post-CRT residual microscopic disease. This is reflected by our low complication rate and low recurrence rate in the neck. We, as others, have confirmed that surgery (without pharyngeal entry) can be safe after CRT. This concept represents an important paradigm, since CRT is likely to become more widely used in the next few years.
There is varied information in the literature regarding the management of the neck in patients undergoing multimodality therapy.19- 22 Although most reports agree on indications of posttreatment dissection (ie, residual neck disease regardless of N status or for N2 or greater disease), optimal timing (how long posttreatment is safe or technically feasible) and type of dissection (radical vs selective) are parameters that are in need of clarification.
Unlike previous reports, most neck dissections performed in this study were SNDs (supraomohyoid [levels I-III] or extended supraomohyoid [levels I-IV]) as opposed to radical neck dissections. Level V was dissected during the SND only if there was pretreatment gross disease in this location. We have found that despite preservation of the accessory nerve, its function appears to be more severely impaired when it is dissected through the posterior triangle. Only 1 patient had disease that recurred in the neck after dissection, suggesting that these "less than radical" dissections are effective in clearing the residual neck disease. In addition, by definition, SNDs preserve the sternocleidomastoid muscle, which naturally serves to protect the vascular sheath contents. This concept of organ-preserving neck dissection after CRT is echoed by Boyd et al,23 who support the philosophy of SND in postradiotherapy patients. Bhattacharyya24 has specifically studied this issue in his report of conservation neck dissections and preoperative radiotherapy on the quantitative recovery of lymph nodes from the neck. He found that the quantity of nodes excised in the "conservation" neck dissections is less than that for radical dissections but that the pathologic nodal yield is not significantly different. Although the amount of nodal yield decreased significantly for all types of dissections after radiotherapy, it had no significant effect on pathologic nodal yield.
In regard to timing of post-CRT surgery, surgeons may elect to operate on patients who have undergone multimodality therapy after the acute dermatitis has resolved. However, consistent published standards for the optimal timing of postradiotherapy or post-CRT surgery are lacking in the literature. Our data, as well as the data of others, indicate that a surgical "window" may exist between the resolution of the acute CRT injury and the onset of the chronic CRT injury in which technically feasible and safe neck dissections can be completed without major wound healing complications.6,25 This safe window is approximately 4 to 12 weeks after completion of CRT. This intuitive concept is supported by the work of Hopewell et al,26 who studied the cellular basis of vascular irradiation damage. They found that initial changes in vessels 2 to 4 months after radiation were associated with a loss of endothelial cells, which resulted in an abnormal proliferation of viable cells. This proliferation resulted in decreased size of the capillary bed from small vessel occlusion. Later changes in the vasculature were associated with a reduction of smooth muscle cells. Thus, surgical trauma before capillary occlusion may result in wounds that heal relatively normally, as evidenced by our low wound complication rate.
In our series of patients who underwent neck surgery alone after CRT, there were no wound infections. Although the literature suggests that antibiotics are unnecessary in "clean" surgery (without pharyngeal entrance), we routinely used antibiotics in the perioperative period. The justification of antibiotic use is based on the findings of Carrau et al,27 who found a trend (P = .09) suggesting the possible value of antibiotics for neck dissections. The total wound healing complication rate of 10% is comparable to historical complication rates of neck dissection in the literature25,28- 32 (Table 6). The nonwound healing complication rate may be higher and appears to be somewhat different than previous reports. The requirement for tracheotomy in 6 patients we believe is a result of the combination of CRT and surgical "insults" to the patient's damaged, previously normal tissues. The nerve transections were technical errors, not clearly related to the compromised tissue bed, and occurred early in our experience. Hypocalcemia after thyroidectomy and nerve paresis without gross nerve injury is most likely related to blood supply of the parathyroid glands and nerves.
We also reported that 35% of patients had positive pathology in their dissection specimens, with only 1 patient experiencing disease recurrence in the neck. This finding emphasizes the necessity of neck dissection after CRT to remove residual microscopic disease, independent of a complete or partial clinical or radiologic response. As previously discussed, other studies have supported planned neck dissections for N2 or greater disease.23,25
As multimodality, organ preservation strategies become more successful, the role of surgery is being redefined. We have found that neck dissection in patients with advanced head and neck cancer, who have a pathologic CR at the primary site, is associated with a low wound healing complication rate and a significant percentage of positive pathology in the pathologic specimen; neck dissection after CRT is both feasible and necessary. Future studies within this group of patients will focus on the correlation of positive nodal status with distant metastases and survival.
Conclusions from this analysis include the following. (1) Neck dissection for patients with N2 or greater neck disease after CRT is necessary to eradicate residual disease as evidenced by a 35% incidence of positive pathology. There is no association between radiologic PR and CR and residual microscopic cancer in this group of patients. (2) Complication rate after CRT with hyperfractionated radiotherapy is low. (3) SND within the proposed safe window of 4 to 8 weeks post-CRT is technically feasible. (4) SND, rather than more radical dissections, is technically and therapeutically appropriate in this group of patients because of low incidence of disease recurrence in the neck. We believe that our low complication rate is due to early surgery after CRT. These data support the need for surgery and continued investigation of CRT in organ-preserving strategies.
Accepted for publication February 22, 2000.
We acknowledge William Moran, MD, and Barry Wenig, MD, for their contribution to our clinical database.
Corresponding author: Kerstin M. Stenson, MD, University of Chicago, 5841 S Maryland, MC1035, Chicago, IL 60637.