Mature results of a Phase III randomized trial comparing concurrent chemoradiotherapy with radiation therapy alone in patients with Stage III and IV squamous cell carcinoma of the head and neck

Authors

  • David J. Adelstein M.D.,

    Corresponding author
    1. Department of Hematology and Medical Oncology, The Cleveland Clinic Foundation, Cleveland, Ohio
    • Desk T40, Department of Hematology and Medical Oncology, The Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, Ohio 44195
    Search for more papers by this author
  • Pierre Lavertu M.D.,

    1. Department of Otolaryngology and Communicative Disorders, The Cleveland Clinic Foundation, Cleveland, Ohio
    Current affiliation:
    1. Department of Otolaryngology and Head and Neck Surgery, University Hospitals of Cleveland, Cleveland, Ohio.
    Search for more papers by this author
  • Jerrold P. Saxton M.D.,

    1. Department of Radiation Oncology, The Cleveland Clinic Foundation, Cleveland, Ohio
    Search for more papers by this author
  • Michelle Secic M.S.,

    1. Department of Biostatistics, The Cleveland Clinic Foundation, Cleveland, Ohio
    Search for more papers by this author
  • Benjamin G. Wood M.D.,

    1. Department of Otolaryngology and Communicative Disorders, The Cleveland Clinic Foundation, Cleveland, Ohio
    Search for more papers by this author
  • John R. Wanamaker M.D.,

    1. Department of Otolaryngology and Communicative Disorders, The Cleveland Clinic Foundation, Cleveland, Ohio
    Current affiliation:
    1. Central New York Ear, Nose, and Throat Consultants, Syracuse, New York.
    Search for more papers by this author
  • Isaac Eliachar M.D.,

    1. Department of Otolaryngology and Communicative Disorders, The Cleveland Clinic Foundation, Cleveland, Ohio
    Search for more papers by this author
  • Marshall Strome M.D.,

    1. Department of Otolaryngology and Communicative Disorders, The Cleveland Clinic Foundation, Cleveland, Ohio
    Search for more papers by this author
  • Marjorie A. Larto R.N.

    1. Department of Hematology and Medical Oncology, The Cleveland Clinic Foundation, Cleveland, Ohio
    Search for more papers by this author

Abstract

BACKGROUND

The current study presents mature results from a Phase III randomized trial comparing radiation therapy and concurrent chemoradiotherapy in patients with resectable American Joint Committee on Cancer Stage III and IV disease.

METHODS

One hundred patients were randomized to receive either radiation therapy alone (Arm A) (at a dose of between 66–72 grays [Gy] at 1.8–2 Gy per day) and the identical radiation therapy with concurrent chemotherapy (Arm B) (5-fluorouracil, 1000 mg/m2/day, and cisplatin, 20 mg/m2/day, both given as continuous intravenous infusions over 4 days beginning on Days 1 and 22 of the radiation therapy). Primary site resection was planned for patients with residual or recurrent local disease. Cervical lymph node dissection was performed for regional persistent disease or recurrence, or if N2-3 disease was present at the time of presentation.

RESULTS

After completing all therapy including surgery, 82% of the patients in Arm A and 98% of the patients in Arm B had been rendered disease free (P = 0.02). At a median follow-up of 5 years (range, 3–8 years), the 5-year Kaplan–Meier projections for overall survival for Arm A versus Arm B were 48% versus 50% (P = 0.55). Kaplan–Meier projections for the recurrence free interval were 51% versus 62% (P = 0.04), projections for a distant metastasis free interval were 75% versus 84% (P = 0.09), projections for overall survival with primary site preservation were 34% versus 42% (P = 0.004), and projections for local control without surgical resection were 45% versus 77% (P < 0.001). Salvage surgery proved to be successful in 63% and 73%, respectively, of the Arm A and Arm B patients with primary site failure. Unrelated death while free of disease occurred in 22% and 32%, respectively, of Arm A and Arm B patients (P = 0.26).

CONCLUSIONS

The addition of concurrent chemotherapy to definitive radiation in patients with resectable Stage III and IV squamous cell carcinoma of the head and neck improves the likelihood of disease clearance, a recurrence free interval, and primary site preservation. However, overall survival does not appear to be improved, reflecting both effective surgical salvage after local recurrence and competing causes of death. Cancer 2000;88:876–83. © 2000 American Cancer Society.

The value of adding chemotherapy to definitive surgery and/or radiation in the treatment of squamous cell carcinoma of the head and neck has been evaluated extensively over the past 20 years. Despite initial promise, a survival advantage has been difficult to demonstrate, particularly after neoadjuvant or sequential treatment schedules.1, 2 As a result, chemotherapy generally has not been considered a part of standard management.However, several recent studies using concurrent schedules of radiation therapy and full doses of effective combination chemotherapy have suggested that there may indeed be a survival benefit resulting from the addition of chemotherapy to definitive management.3–6 Recent meta-analyses also appear to confirm this observation.7–9 However, these trials have not compared concurrent chemoradiotherapy with definitive surgical resection specifically, nor has there been adequate exploration of the role of salvage surgery after failure of these aggressive nonoperative treatments.

It should be noted that although survival remains the gold standard by which treatment success must be judged, in patients with squamous cell carcinoma of the head and neck survival is impacted by many factors. These patients often have significant comorbid illnesses, much of which relates to the tobacco and alcohol abuse associated with this disease.1 Furthermore, the likelihood of a second malignancy, particularly of the upper aerodigestive tract, is very high in this population.10, 11 Thus, identification of a survival benefit from an investigational treatment often is confused by nontumor-related mortality, and other endpoints of treatment success become important, such as the possibility of increasing resectability, decreasing distant metastases, or improving the likelihood of primary site preservation.

In 1997, we reported preliminary results from a Cleveland Clinic Foundation Phase III randomized trial that compared definitive radiation therapy with concurrent chemoradiotherapy in patients with Stage III and IV squamous cell carcinoma of the head and neck.12 Surgical salvage was planned for those patients failing initial nonoperative treatment. These initial results suggested an improvement in recurrence free survival and primary site preservation in those patients treated with chemotherapy, although no improvement in overall survival was identified. We now report mature results from this study to confirm these initial observations and to further detail patterns of failure and the success of salvage treatment.

MATERIALS AND METHODS

Eligibility for this clinical trial required a diagnosis of squamous cell carcinoma of the head and neck that was considered to be resectable for cure. Patients with primary sites in the nasopharynx, paranasal sinuses, and salivary glands were excluded. Patients with an unknown primary head and neck tumor also were excluded. T and N classifications were assigned according to the staging system of the American Joint Committee on Cancer,13 and only patients with Stages III and IV, nonmetastatic (M0) disease were eligible. All patients previously were untreated and had an Eastern Cooperative Oncology Group performance status of 0 or 1.

Pretreatment evaluation in all patients included a medical history, an examination under anesthesia with panendoscopy, and a chest radiograph. Computed tomography scans or magnetic resonance imaging of the involved head and neck region or other staging procedures for distant metastases were obtained if clinically indicated. Pretreatment laboratory evaluation included a complete blood count and serum chemistry tests including urea nitrogen, creatinine, calcium, phosphorous, alkaline phosphatase, aspartate aminotransferase, albumin, total protein, bilirubin, and uric acid. Adequate hematologic, renal, and hepatic function were required for patient entry. Patients with uncontrolled angina, active infection, or other uncontrolled malignancy also were deemed ineligible.

Although resectability is a subjective judgment, specific criteria for resectability were established for this study. Disease was not considered resectable if there was tumor or lymph node fixation to the carotid artery, prevertebral fascia, or base of the skull. It also was considered unresectable if total extirpation of the tumor with negative margins would preclude a functional reconstruction. Patients with T4 disease because of clear bone involvement also were excluded from this trial.

The study was approved and reviewed yearly by the Cleveland Clinic Foundation Institutional Review Board. Written informed consent was obtained from all patients prior to randomization and the start of therapy. Patient management was provided by a multidisciplinary team including head and neck surgeons and medical, radiation, and nurse oncologists. All patients underwent a pretreatment dental evaluation with appropriate care.

Eligible patients were stratified by lymph node status (N0-1 vs. N2-3) and were randomized between Arm A (radiation therapy alone, 66–72 grays [Gy] given in conventional daily 1.8–2-Gy fractions) or Arm B (identical radiation therapy with concurrent chemotherapy). This chemotherapy was comprised of 5-fluorouracil, 1000 mg/m2/day, and cisplatin, 20 mg/m2/day, both given as continuous intravenous infusions over 4 days beginning on Days 1 and 22 of the radiation therapy. The overall treatment time required for the completion of both treatment arms was identical.

The protocol called for a response assessment after approximately 55 Gy of radiation had been administered. At this time all patients again were seen by the head and neck surgeon and a decision was made about continuation of therapy. Those patients experiencing a clinical response to treatment completed radiation. However, radiation therapy was discontinued in nonresponders or in those with progressive disease and, after a 3–6-week recovery period, surgical resection was performed.

For those patients who completed definitive radiation therapy, 6–12 weeks were allowed for mucosal recovery prior to a formal response analysis. This analysis included an examination under anesthesia and biopsy when deemed appropriate by the clinical circumstances. A complete response required the complete disappearance of all clinical, radiographic, and, if applicable, pathologic evidence of disease. Any response less than complete was deemed a treatment failure and the patient underwent appropriate surgical resection, if possible.

Surgery, whether it was performed after premature termination of radiation therapy or after an incomplete response to the full course of radiation, was comprised of a primary site resection if any biopsy confirmed evidence of residual disease was noted at the primary site. Cervical lymph node dissection was performed in those patients undergoing primary site resection, or if any clinically palpable lymph nodes remained after radiation. It also was recommended for those patients with N2 or N3 disease at presentation regardless of clinical response. Primary site surgery was not performed in those patients achieving a complete response at the primary site. Any patient subsequently developing locoregional disease recurrence was considered for appropriate salvage surgery.

Chemotherapy administration required hospitalization for hydration and antiemetic therapy. No chemotherapy dosage modifications were made irrespective of nadir blood counts or the blood count at the time of treatment. Megavoltage radiotherapy was generated by a 6-megavolt linear accelerator. Opposed lateral fields generally were used with an electron beam boost given to selected lymph node regions as indicated. There were no planned breaks scheduled during the administration of the radiation therapy, nor were there any toxicity-mandated delays in the administration of the second course of chemotherapy.

Patients were monitored at least weekly during their therapy in an effort to manage treatment-induced side effects, particularly mucositis and myelosuppression. Neutropenia with fever mandated hospitalization and appropriate antibiotic therapy. Hospitalization also was required when mucosal injury precluded an adequate oral intake. Nasogastric or percutaneous endoscopic gastrostomy feeding tubes were placed as needed. Tracheostomies were performed in those with significant airway compromise either at presentation or during the course of their treatment.

At the completion of all therapy, patients were followed at regular and frequent intervals by all members of the multidisciplinary team. Careful clinical examination was performed at 2–3-month intervals and suspected local, regional, or distant recurrences were biopsied. Radiographic studies were performed as clinically indicated.

A sample size of 100 patients was selected in an effort to identify, with 80% power, a postulated 25% difference in 3-year survival between the 2 treatment arms. Survival times and times to specific events were calculated from the date radiation therapy was initiated. The results were analyzed as of November 1, 1998. No patient was lost to follow-up. Survival curves were constructed using the Kaplan–Meier method14 and compared using Wilcoxon tests. Patient group characteristics, toxicities, and responses were compared using the Fisher exact test, the chi-square test, or the Student t test when appropriate. Recurrence free interval, distant metastasis free interval, and the likelihood of local control without primary site surgery were considered censored at the time of patient death if no event had yet occurred. In calculating the overall survival with primary site preservation, only those patients who were alive with no primary site resection and with no residual or recurrent disease present at the primary site were considered censored. Multivariable methods, using Cox regression models, were used to simultaneously examine multiple predictive factors for overall survival and for overall survival with primary site preservation.

RESULTS

Between March 1990 and June 1995, 100 patients were entered on this clinical trial. Fifty patients were assigned randomly to each treatment arm. All patients were eligible and all patients were evaluable for toxicity, response, and survival. All patients were considered resectable. The clinical characteristics of these 100 patients and their tumors are detailed in Table 1 and were statistically equivalent between the 2 treatment arms. Tumor (T) and lymph node (N) distribution is presented in Table 2. There were 14 patients with Stage III disease and 36 patients with Stage IV disease per arm. Except for significantly more T1 patients on Arm A the treatment arms were well balanced. This imbalance in T distribution was most apparent among those patients with primary tumors of the oropharynx. Six of the 22 Arm A patients with oropharyngeal carcinoma had T1 primary tumors whereas there were no T1 tumors among the 22 Arm B patients with oropharyngeal carcinoma (P = 0.02).

Table 1. Patient Characteristics
  Arm A (n = 50)Arm B (n = 50)
  1. PS: Eastern Cooperative Oncology Group performance status.

Median age (yrs) (range)60 (37–76)58 (30–76)
RaceWhite4444
African American66
GenderMale4034
Female1016
PS02929
12121
Primary siteOral cavity04
Oropharynx2222
Larynx1917
Hypopharynx97
Table 2. Tumor (T) and Lymph Node (N) DistributionThumbnail image of

Toxicity due to this treatment has also been described previously.12 As expected, there was significantly more toxicity on the treatment arm containing chemotherapy, with statistically more National Cancer Institute Common toxicity Criteria Grade 3 or 4 neutropenia, thrombocytopenia, cutaneous reactions, and mucositis in the Arm B patients, 36% of whom required hospitalization because of neutropenia and fever. Patients receiving chemoradiotherapy experienced a mean 12.4% loss of body weight compared with a mean 6.3% loss of body weight in those treated with radiation therapy alone (P < 0.001). Feeding tubes were required in 32% of the Arm A patients compared with 58% of the Arm B patients (P < 0.01). However, it is important that there were no toxic deaths on either treatment arm.

Radiation therapy dose intensity was maintained on both treatment arms and a median of 58 days was required for the completion of treatment regardless of whether chemotherapy was given. Significant treatment breaks were uncommon. Although delayed in those patients on Arm B, gradual mucosal recovery occurred in all patients and generally was complete within 8–12 weeks.

Late toxicities included hypothyroidism in 38% of the patients on Arm A and 38% of those on Arm B. Second malignancies were identified in 20 patients: 16% of the Arm A patients and 24% of the Arm B patients (P = 0.32). These were of upper aerodigestive tract origin in 12% and 14% of patients, respectively.

A complete response at the primary site was achieved without surgery in 66% of the Arm A patients and 94% of the Arm B patients (P < 0.001). After completing all therapy including surgical resection, 82% of the Arm A patients and 98% of the Arm B patients were rendered disease free (P = 0.02).

With a minimum follow-up of 3 years and a median follow-up of 5 years (range, 3–8 years), 5-year Kaplan–Meier projections were calculated. The P values were not adjusted for the imbalance in T1 patients, which favored Arm A. Figure 1 depicts the Kaplan–Meier overall survival function. For Arm A the 5-year projected overall survival was 48% versus 50% for Arm B (P = 0.55). Figure 2 depicts the projected 5-year recurrence free intervals of 51% versus 62% for Arm A and Arm B patients, respectively (P = 0.04). Distant metastasis free intervals (Fig. 3) were 75% versus 84% in Arm A versus Arm B patients, respectively, again favoring Arm B (P = 0.09).

Figure 1.

Kaplan–Meier projected overall survival.

Figure 2.

Kaplan–Meier projected likelihood of a recurrence free interval.

Figure 3.

Kaplan–Meier projected likelihood of a distant metastasis free interval.

The possibility of primary site preservation can be measured in several ways. Figure 4 depicts the likelihood of local disease control without the need for surgical resection. At 5 years this was 45% for Arm A versus 77% for Arm B (P < 0.001). Perhaps more important is the estimation of overall survival with primary site preservation (Fig. 5). At 5 years this was 34% versus 42% for Arm A versus Arm B, respectively (P = 0.004). The 5-year overall survival with primary site preservation was significantly worse for Arm A than for Arm B patients with laryngeal primary tumors (16% vs. 29%; P = 0.03) and hypopharyngeal primary tumors (0% vs. 14%; P = 0.008), but not for those patients with oropharyngeal primary tumors (63% vs. 64%; P = 0.86).

Figure 4.

Kaplan–Meier projected likelihood of local disease control without surgical resection.

Figure 5.

Kaplan–Meier projected overall survival with primary site preservation.

Both univariate and multivariate analyses were performed on these data examining the effect of treatment group, primary site, stage, and T and N classification on overall survival and overall survival with primary site preservation. The only variable of importance in predicting for overall survival was the primary site. Patients with oropharyngeal and laryngeal primary sites did significantly better than those with hypopharyngeal primary tumors, regardless of the treatment arm. However, when evaluating the overall survival with primary site preservation, both primary tumor site (oropharynx and larynx) and treatment group (Arm B) were predictive of a better result. Patients treated with radiation therapy alone had a relative risk of 2.1 (P = 0.006; 95% confidence interval, 1.24–3.50) of either dying or failing at the primary site when compared with those treated with chemoradiotherapy (Table 3).

Table 3. Cox Regression Analysis for Overall Survival with Primary Site Preservation
VariableParameter estimateStandard errorWald test statisticP valueRisk ratio95% CI
  1. 95% CI: 95% confidence interval; OC: oral cavity; HP: hypopharynx; OP: oropharynx.

Univariate analysis
Treatment0.5080.2534.0440.041.662(1.01–2.73)
OC vs. HP−1.6280.7594.6020.045.092(1.15–22.53)
Larynx vs. HP−0.6570.3293.9910.051.928(1.01–3.67)
OP vs. HP−1.6730.35522.161< 0.0015.327(2.65–10.69)
Multivariable analysis
Treatment0.7320.2657.6520.0062.080(1.24–3.50)
OC vs. HP−1.3920.7683.2880.074.023(0.89–18.12)
Larynx vs. HP−0.7260.3314.8180.032.066(1.08–3.95)
OP vs. HP−1.8690.36725.859< 0.0016.479(3.15–13.31)

It must be remembered that patients were not stratified by primary site, and any data analysis by primary site must be considered suspect. However, patients were stratified by lymph node status (N0-1 vs. N2-3), a variable that proved to be of no significance to any endpoint analyzed.

Surgical resection of residual or recurrent primary site disease was pursued vigorously in all appropriate patients. Table 4 details our success at primary site control. Local failure was statistically more likely among the patients treated with radiation therapy alone. This included a 34% (vs. 6%) incidence rate of disease persistence after the completion of radiation therapy and an additional 30% (vs. 18%) who experienced a delayed local recurrence. Local surgical salvage was considered successful if local control subsequently was maintained, even if the patient died of distant metastases or of unrelated reasons. It is important that surgical salvage of the patients who failed locally proved successful in 63% (17 of 27 patients) of those treated with radiation therapy and 73% (8 of 11 patients) of those given chemoradiotherapy. This successful surgical salvage rate suggests that a treatment philosophy based on initial nonoperative management, followed by surgery for those patients with residual or recurrent disease, is reasonable. Successful surgical salvage also served to obscure any possible survival benefit resulting from the addition of chemotherapy to the treatment regimen.

Table 4. Disease Control at the Primary Site and Surgical Salvage
 Arm A ( n = 50)Arm B ( n = 50) 
Persistence after nonsurgical treatment17/50 (34%)3/50 (6%)P < 0.001
Local recurrence after initial complete response10/33 (30%)8/45 (18%)P = 0.20
Total local failures27/50 (54%)11/50 (22%)P < 0.001
Successful local salvage17/27 (63%)8/11 (73%)P = 0.57

Table 5 details the status at last follow-up of all patients entered on this study. Equivalent numbers of patients on both treatment arms were alive at last follow-up and disease free. However, it is important to note that of the 58 patients on this study who had died by the time of last follow-up, 27 (47%) died of unrelated causes while free of active malignancy.

Table 5. Patient Status
 Arm A (n = 50)Arm B (n = 50) 
Alive
 Disease free2021
 Active disease10
Dead
 Disease free1116P = 0.26
 Active disease1813P = 0.28

DISCUSSION

The current follow-up report serves to confirm our initial observations. Overall survival was not impacted by the addition of concurrent chemotherapy to definitive radiation therapy. However, disease clearance, recurrence free interval, and primary site preservation were improved significantly by the chemotherapy. Although a trend existed favoring the chemoradiotherapy arm when estimating the likelihood of distant metastases, statistical significance was not been reached, reflecting both the limited number of patients failing due to hematogenous disease and the small size of our study.

These results demonstrate the importance of assessing multiple endpoints in any evaluation of the role of chemotherapy for patients with this tumor. A simple look at survival will prove deceptive because survival is impacted by multiple other factors in this patient population, including second malignancies and death from comorbid illness.1 In theory, such unrelated deaths should occur with equal frequency, regardless of treatment arm, resulting in the need for a larger patient cohort to demonstrate a survival difference. However, a survival trend still should be identifiable, even if statistical significance is not reached. No such trend was observed in our data.

Instead, we must examine the role of surgical salvage. In assessing the results of the current trial, one must recognize that it is not just a comparison of radiation therapy with concurrent chemoradiotherapy, but a comparison of radiation therapy with surgical salvage and concurrent chemoradiotherapy with surgical salvage. Fifty-four percent of those patients treated with radiation therapy alone failed at the primary site. Sixty-three percent of these local failures were salvaged successfully with surgery. One might view these data as reflective of the failure of radiation therapy alone to achieve primary site preservation. It also represents the success of radiation therapy with surgical salvage as an overall treatment approach and may help explain why, unlike in other recent chemoradiotherapy trials, we were unable to demonstrate a survival benefit.

Three randomized studies in similar patients, in which a survival advantage was found after adding aggressive combination chemotherapy to concurrent radiation therapy, reported 3-year survival rates after radiation therapy alone of 24–34%.4–6 However, the role of surgical salvage after local failure was not well defined in these publications. Our 3-year overall survival rate for radiation therapy-only patients was 60%, with a 5year survival rate of 48%, results that reflect both our aggressive approach to surgical salvage and the exclusion of unresectable patients from this trial. However, we suggest that any clinical trials of nonoperative treatment for patients with head and neck carcinoma should fully detail any role of surgery in patient management. Studies of laryngeal preservation from the Veterans Affairs Laryngeal Cancer Study Group15 and the European Organization for Research and Treatment of Cancer16 are examples of how surgical resection can be incorporated into a treatment protocol after failure of nonoperative management. Although surgical salvage will obscure the identification of an overall survival benefit, it optimizes patient management and is more consistent with clinical practice.

It also must be pointed out that the need for surgical salvage was significantly less in those patients treated on the chemoradiotherapy arm, and that survival with primary site preservation for these patients was significantly better. This improvement in primary site preservation thus becomes our major achievement after the addition of concurrent chemotherapy to definitive radiation therapy and represents a valid treatment endpoint.

Although no patient died during treatment, significant toxicity was encountered with this chemoradiotherapeutic schedule. Utilization of these kinds of multimodality approaches requires a commitment from the responsible clinicians to close patient monitoring and aggressive intervention for mucositis, myelosuppression, and weight loss. Even among our patients treated with radiation therapy alone, 32% required placement of a feeding tube in an effort to avoid any treatment breaks. This allowed us to maintain radiation therapy dose intensity and to maximize our results. However, the additional toxicity produced by the chemotherapy is significant and can be justified only if a significant benefit also can be demonstrated. The improvement in primary site preservation identified after chemoradiotherapy is just such a benefit.

It also must be pointed out that this study did not address the role of primary surgical resection in the management of these patients. Although initial surgery precludes the possibility of primary site preservation, current reconstruction and rehabilitation techniques may produce equivalent or superior organ function conservation.17 Indeed, a large primary site tumor treated with aggressive chemotherapy and radiation therapy still may result in significant functional impairment, and negate the value of any organ preservation achieved. It also is unclear whether primary surgical and primary nonsurgical treatment approaches produce equivalent survival in this patient population. List et al. recently reminded us of our patients' charge.18 Their work strongly suggests that when confronted with potential treatment outcomes, patients choose cure or survival prolongation as the most important, even when compared with the preservation of a natural voice or of swallowing. Additional investigation of these aggressive nonoperative treatments is indicated but comparisons must to be made with primary surgical approaches.

The addition of concurrent chemotherapy to definitive radiation therapy can improve the likelihood of disease clearance, a recurrence free interval, and primary site preservation. However, overall survival is unaffected in part due to our aggressive approach to surgical salvage after local treatment failure. The multiple competing causes for death in this patient population, including the likelihood of second malignancy, must be factored into any assessment of a potential survival benefit, and alternative endpoints should be examined in any future studies.

Ancillary