Presented at the 43rd Annual Meeting of the American Society of Clinical Oncology, Chicago, Illinois, June 1-5, 2007.
The current study was conducted to determine the long-term outcomes of patients with squamous cell carcinoma of the larynx and pharynx who were treated with platin-based exclusive chemotherapy (EC) after they achieved a complete clinical response (CCR) to induction chemotherapy.
One hundred forty-two who achieved a CCR after platin-based induction chemotherapy were treated exclusively with additional chemotherapy, and 98.6% were followed for a minimum of 3 years or until death. Thirty-five patients had >10 years of follow-up.
The survival rates at 1 year and 5 years were 95.8% and 61.2%, respectively. The main causes of death were metachronous second primary tumors (n = 27) and intercurrent disease (n = 21). Death related to EC was not encountered, and only 2 patients (1.4%) had grade 4 toxicity. In multivariate analysis, primary tumor arising outside the glottic larynx (P = .0001) and a Charlson comorbidity index >1 (P = .0001) were associated with a statistically significant reduction in survival. The 1-year and 5-year Kaplan-Meier local control estimates were 76.1% and 50.7%, respectively. Salvage treatment resulted in an observed final local control rate of 93% that varied from 97.2% in patients who had glottic cancer to 88.7% in patients who had tumor originating from other sites (P = .097). Combined chemotherapy with cisplatin and 5-fluorouracil (PF) allowed for the successful modulation of local therapy in 54.9% of patients.
Since its introduction by Decker et al1 and Jacobs et al,2 the use of platin-based induction chemotherapy has evolved from experimental therapy to an integral part of the comprehensive care3-6 of patients with invasive squamous cell carcinoma (SCC) of the head and neck. Beginning in 1981, we began to use induction chemotherapy before conservation laryngeal surgery.7, 8 Since then, our experience as well as that of others3-5, 9-14 has demonstrated that the use of platin-based induction chemotherapy before conventional treatment results in a 20% to 40% rate of complete clinical response (CCR). On the basis of these encouraging results, a novel approach based on the use of platin-based chemotherapy alone termed “exclusive chemotherapy” (EC) was pioneered at our institution. There were no specific criteria that were used initially to select patients for this approach. At first, certain patients who had no residual cancer after a CCR refused surgery or radiation but would consent to further chemotherapy. Nonetheless, using this empiric approach, first, we studied a series of 21 patients with glottic cancer.7 The initial results achieved were confirmed in a larger cohort of 67 patients with selected SCC of the larynx and hypopharynx who had achieved a CCR.8
In this final report, we review all patients who received EC at our department from 1981 to 2004, resulting in a much larger cohort of 142 patients with a longer duration of follow-up. We critically appraise this experience to identify the limits and benefits of this experimental approach and, in particular, to determine the feasibility of extending this approach more broadly to patients with head and neck cancer (HNC). We hypothesized that certain patients with HNC would achieve long-term, durable remission with chemotherapy alone. For those patients with residual disease after EC, we studied whether or not the systematic use of additional chemotherapy after CCR would allow us to down-modulate the intensity of local therapy. Herein, we present the results of this novel experience in terms of overall survival (OS), local recurrence-free survival, causes of death, and chemotherapy-related toxicity.
MATERIALS AND METHODS
From 1981 to 2004, 2271 patients with a previously untreated, moderately to well differentiated, invasive SCC of the larynx or pharynx (excluding the nasopharynx and velum) were managed with an induction therapy regimen at our institution. Of these patients, 23.9% (545 of 2271) achieved a CCR. A CCR after induction chemotherapy was defined as the absence of tumor on control endoscopy as well as by corroborating radiographic studies after induction chemotherapy. Of the 545 complete clinical responders, 26.1% (142 of 545) were managed with EC and are analyzed in the current retrospective series.
In this inception cohort of 142 patients, the tumor originated from the glottic larynx in 71 patients, from the nonglottic endolarynx in 9 patients, from the epilarynx15, 16 or the hypopharynx in 43 patients, and from the oropharynx in 19 patients. The epilarynx is defined as mucosa along the “margins” of the larynx as it transitions into the surrounding pharynx. The epilarynx is divided into the anterior epilarynx (the suprahyoid epiglottis), the lateral epilarynx (aryepiglottic folds), and the posterior epilarynx (arytenoids).15, 16 Table 1 documents the TNM staging according to the 2002 International Union Against Cancer staging classification system.17 Metachronous second primary tumors were defined according to the criteria first proposed by Warren and Gates18 and by Nikolaou et al19; thus, each newly diagnosed tumor was located least 2 cm from the treated field margin of the previous tumor.
Table 1. The International Union Against Cancer TNM Staging Classification
Internal review board protocols were not used routinely in France when this approach was first begun; therefore, the decision to use EC was made after a multidisciplinary discussion in which the roles and benefits of standard therapy (surgery, radiation, chemoradiation) as well as the dangers of chemotherapy were presented to the patient. Table 2 documents the demographic and comorbidity status according to the Charlson20 index-weighted score (excluding the index cancer).
Before starting the induction chemotherapy regimen, 93 patients (65.5%) were considered amenable to radiation therapy or to a partial resection sparing the physiologic functions of phonation, respiration, and deglutition. These patients were defined as Group A. Among the remaining 49 patients, 43 (30.3%) were amenable only to a surgical resection that did not spare the physiologic functions of the tumor-laden organ (mandibular resection for 4 patients with oropharyngeal cancer or permanent stoma and loss of voice for 42 patients with laryngeal or hypopharyngeal cancer). These patients were categorized as Group B. The last 6 patients with CCR (4.2%) were considered unresectable before the induction chemotherapy regimen and were defined as Group C.
The number of induction chemotherapy cycles varied from 2 to 3 (depending on the toxicity and the response). The number of cycles and the agents that patients received were not standardized. The treatment was tailored individually to each patient's tumor classification and performance status. Overall, the total number of courses used for EC varied from 1 to 17 (median, 5 courses). The median course duration was 5 days. Our main and current regimen consisted of cisplatin (25 mg/m2 daily) and 5-fluorouracil (1 g/m2 daily) delivered through an implanted venous (subclavian vein) access device (Baxter HealthCare Corporation, Deerfield Ill). Over time, other agents, as depicted in Table 3, also were used. The total dosage of the drugs delivered varied for each patient. Neither taxanes nor long-term maintenance therapies, such as 13-cis-retinoic acid, were used. The dosages were adjusted to tolerance. Antiemetics and oral hyperhydration (2 L daily) were used routinely during the treatment course. Induction chemotherapy was administered with a 15-day to 21-day hiatus between courses (the ultimate interval was determined by toxicity). Clinical examinations, blood counts, and serum chemistry performed 15 days after each course allowed for the analysis of cumulative toxicity (Table 4) according to Eastern Cooperative Oncology Group criteria.21 Follow-up was calculated as the time from a patient's first appointment in our department for the index SCC until the date of either last contact or death. The median follow-up was 5 years and 3 months. Ninety-nine patients were followed for a mean duration of 3 years or until death. Thirty-four patients had >10 years of follow-up, and 1 patient was followed for 19 years.
Table 3. Total Dosages of the Drugs Delivered
No. of Patients Who Received the Drug
Total Mean Dose Delivered (Range), mg
Bleomycin sulfate, mg
Table 4. Toxicity According to the Eastern Cooperative Oncologic Group Criteria*
For the current study, the modulation of local therapy was considered successful after EC when 1) patients who had unresectable tumors could receive conservation or nonconservation treatment, 2) patients who required a total laryngectomy or composite resection could undergo a conservation surgical approach or radiation therapy, or 3) a patient who initially was amenable to a conservation surgical approach or radiation therapy achieved local control with EC alone. Unsuccessful modulation of local therapy was defined as a patient who was amenable to conservation therapy before EC but underwent total laryngectomy or composite transmandibular resection or a patient who had any resectable tumor before EC and became unresectable after EC.
The variables under analysis are listed in Tables 5 and 6. The data were summarized using standard descriptive statistics and frequency tabulations. Associations between categorical variables were assessed by using cross-tabulation, chi-square tests, or Fisher exact tests, and either Wilcoxon rank-sum tests or Kruskal-Wallis tests were used to assess the differences between patients' clinical characteristics with respect to continuous variables.22 The distribution of OS was defined from the initiation of EC to death; the time to local recurrence was defined from the initiation of EC to local recurrence; and the time to lymph node recurrence was defined from the initiation of EC to lymph node recurrence. The distribution of OS, local recurrence-free survival, and lymph node recurrence-free survival was estimated by using the Kaplan-Meier method. The log-rank test was used to test the differences in survival between groups. Regression analyses of survival data based on the Cox proportional hazards model were conducted on survival data. The assumption of proportional hazards was examined for each variable.23 No violations were detected. P values <.05 were considered statistically significant. All tests were 2-sided. All analyses were conducted in SAS statistical software (version 9.1; SAS Institute Inc, Cary, NC) and S-PLUS (Insightful Corporation, Seattle, Wash).
Table 5. Univariate Statistical Analysis of Overall Survival*
The 1-year, 3-year, and 5-year Kaplan-Meier actuarial survival estimates were 95.8%, 72.8%, and 61.2%, respectively. Over the course of this longitudinal study, 77 deaths (54.2%) were recorded. The causes of death were metachronous second primary tumors (27 patients), intercurrent disease (21 patients), local recurrence (10 patients), distant metastasis (8 patients), and lymph node recurrence (4 patients). In the remaining 7 patients, the cause of death was unknown, but there was no evidence of the primary cancer at the time of death.
Table 5 demonstrates the clinical parameters associated with diminished survival. Patients who had tumors arising from the glottic larynx had improved OS compared with patients who had tumors arising in the epilarynx and hypopharynx (P < .0001) (Fig. 1). There was a significant statistical decrease in survival for patients who were not staged initially as N0 (P = .0032), and this was especially true when patients who were staged as N0 were compared with patients who were staged as N2 and N3 (P = .0051 and P = .0135, respectively) (Fig. 2). Medical comorbidity played a significant role in OS; compared with patients who had a Charlson comorbidity index of 0 or 1, patients who had a Charlson comorbidity index ≥2 had worse OS (P < .0001) (Fig. 3).
Tumor site and medical comorbidity remained significant in the final multivariate model. Patients who had primary tumors arising from the glottic larynx had better OS (P < .0001; hazards ratio [HR], 0.2.97 [95% confidence interval (95% CI), 0.175-0.505]) compared with patients who had primary tumors arising from the epilarynx and hypopharynx. No significant differences were detected for patients who had tumors arising from the oropharynx or supraglottic larynx compared with patients who had primary tumor arising from the epilarynx and hypopharynx. The development of a metachronous second primary tumor was associated with reduced survival (P = .001). Comparing patients based on the Charlson comorbidity index, patients who had an index ≥2 had worse OS than patients who had an index of 0 or 1 (P = .0001; HR, 2.573 [95% CI, 1.597-4.144]).
Incidence and Variables Influencing Local Recurrence
The 1-year, 3-year, and 5-year Kaplan-Meier actuarial local control estimates were 76.1%, 53.7%, and 50.7%, respectively. Sixty-five patients developed local recurrence.
In univariate analysis, as depicted in Table 6, several variables were associated statistically with local recurrence-free survival. Group B had a significantly greater risk of local recurrence than Group A (P = .0024). More courses of therapy (P = .0072) (Fig. 4) and high total dose of 5-fluorouracil (≥37 mg/m2; P = .0014) (Fig. 5) were associated statistically with increased rates of local recurrence. The median number of cycles of chemotherapy associated with local control was 5, and patients who received >5 cycles had an increased risk of local recurrence in univariate analysis (P = .0072). In multivariate analysis, only high-dose 5-fluorouracil (≥37 mg/m2) had a significant effect on local recurrence (P = .0014; HR, 2.470 [95% CI, 1.420-4.296]).
Consequences of Local Recurrence
In the current series, salvage treatment was received by 54 of 65 patients (83.1%) with local recurrence. Of the remaining 11 patients, palliative treatment was offered to 10 patients, and the treatment modality for 1 patient was unknown. The overall final (observed) control rate for patients after salvage (and post-EC) was 93%. The final local control rate after salvage therapy varied from 97.2% in patients who had glottic cancer to 88.7% in patients who had tumors originating from other locations (Fisher exact test; P = .097). Lymph node recurrence was more likely to occur in patients who had local recurrence than in patients who achieved local control (P = .0003, HR, 4.611 [95% CI, 2.025-10.499]), and the 1-year, 3-year, and 5-year Kaplan-Meier actuarial lymph node control estimate remained at 100% in patients without local recurrence compared with 91.8%, 79.7%, and 74.7%, respectively in patients with local recurrence. Patients with local recurrence were more likely to have progression to distant metastasis (P = .041; HR, 4.866 [95% CI, 1.065-22.235]).
The Effect of Exclusive Chemotherapy and Down-Modulating the Intensity of Locoregional Therapy
Table 7 summarizes the impact of EC in modulating the extent of local and regional (LR) therapy. EC resulted in a successful reduction of the extent of LR therapy in 78 of 142 patients (54.9%). This extent of reduction in the treatment varied according to the group classification based on the conventional treatment available before the start of induction therapy. Fifty-seven of 93 patients (61%) who were considered amenable to a conservative approach (Group A) achieved local control after EC without the need for radiation and/or conservation surgery. Thirty-one of 43 patients (72%) who initially were amenable to a nonconservative surgical approach (Group B) achieved local control (either with EC or with a salvage conservation approach). Of the 6 patients who presented initially with unresectable disease (Group C), 4 patients had local control with EC, and 1 patient had local control with salvage conservation surgery. Conversely, 11 patients (7.7%) required more intensive LR therapy (2 patients) or became unresectable (9 patients).
Table 7. Positive and Negative Effects of the Modulation Performed Using Exclusive, Platin-based Chemotherapy Alone
Definitive Local Control After EC
Local Failure After EC: Methods of Salvage and Definitive Result
Table 4 documents toxicity associated with the administration of EC. Overall, 30 patients (21.1%) experienced grade 3 toxicity, and 2 patients had grade 4 toxicity (1.4%). Eight patients (5.6%) reported neurologic sequelae, including peripheral neuropathy in 5 patients and intermittent dysesthesias in 3 patients. Eleven patients (7.7%) had central venous catheter-related complications. Toxicity-related and/or chemotherapy-related complications did not lead to death.
Since 1981, at the University of Paris V, we have systemically studied the use of platin-based induction chemotherapy for patients with cancers of the larynx and pharynx. For selected patients who achieved a CCR after platin-based induction chemotherapy, neither radiation nor conservation laryngopharyngeal surgery was performed. Instead, these patients went on to receive additional cycles of chemotherapy and were observed. Many were cured. These small pilot studies7, 8 suggested the feasibility of using EC to achieve durable, long-term locoregional control without diminishing OS.7 Since then, 2 other groups have explored this approach and have documented similar results in patients classified with N0/N1 disease.24, 25
We present the results from 142 patients with SCC of the larynx and pharynx who received EC during the years from 1981 through 2004. This large cohort represents 26.8% of the patients who achieved a CCR after a platin-based induction chemotherapy regimen during the same period at our institution.
Although all of the patients we studied had SCC originating from the larynx or pharynx and were treated with a platin/fluorouracil-based regimen, the study population was diverse in terms of tumor site, tumor and lymph node classification, medical comorbidity, chemotherapy regimen, and duration of treatment. However, despite this heterogeneity, we were able to analyze a large number of patients over a long duration of follow-up; 98.6% of these patients were followed for a mean duration of 3 years, and 35 patients for >10 years. The size of the study and this lengthy duration of follow-up provide a unique opportunity to determine clinical parameters that are correlated significantly with local control and/or survival after EC.
In the current series, the rates of survival achieved at 1 year, 3 years, and 5 years were 94.5%, 71.5%, and 60.2%, respectively. The main cause of death was metachronous second primary tumors and intercurrent disease, which were much more common than local recurrence, distant metastasis, and/or lymph node recurrence. The definitive use of surgery and radiation therapy was tailored for patients who had residual disease after maximally cytoreductive therapy. Such an approach may diminish acute and long-term swallowing, airway, and speech dysfunction, which are observed commonly with concurrent chemoradiation.26 Finally, as in previous studies from our department7, 8 and from other institutions24, 25 the toxicity of EC was tolerable and did not result in death (Table 4).
With this large cohort, we were able to identify (Table 5) several clinical variables that were associated with diminished survival. Patients who had tumors arising from the glottic larynx had improved OS compared with patients who had tumors arising in other sites (P < .0001). There was a significant decrease in survival noted in patients who did not initially have N0 disease (P = .01), and this trend was especially true when patients who had N0 disease were compared with the few patients who had N3 disease (P < .001). There also was a significant decline in survival (P = .0007) noted when the Charlson comorbidity index20 increased from 0 to 5. The development of a metachronous second primary tumor was associated with reduced survival (P = .001). Take together, these data suggest that the best patients for EC are those who have tumors originating from the glottic larynx, N0 lymph node status, and low medical comorbidity status. This conclusion drawn from our univariate analysis is reinforced by the multivariate analysis, which demonstrated that primary tumor arising outside the glottic larynx and a Charlson comorbidity index20 score >1 were associated with a statistically significant reduction in survival. (Table 5) The lymph node status did not appear in the multivariate analysis because of the finding that very few patients did not have N0 disease. Also, if this approach is used in patients who have nonglottic tumors classified as N0, then we believe that the neck should be treated up-front, because lymph node failure (P = .002) and the subsequent development of distant metastasis (P = .004) were associated significantly with diminished survival.
Two clinical parameters were associated with diminished local recurrence-free survival (Table 6): patients who received a total dose of 5-fluorouracil ≥37 mg/m2 and patients who received >5 cycles. This is a major finding when we compare the results achieved in the current report with the results from our previous reports,7, 8 in which multivariate analysis was not performed. These 2 findings suggest that, in certain individuals, there is a threshold dose past which further systemic therapy is likely not effective, both in terms of the dose of 5-fluorouracil given and in terms of extending the number of cycles past 5. These limitations of EC should be considered carefully in the design of future trials examining this approach.
Overall, 60 patients developed a local recurrence after EC in the current series. By using the Kaplan-Meier27 actuarial life-table method, actuarial local control estimates of 73.5% at 1 year, 53.3% at 3 years, and 50.3% at 5 years were achieved in our series, and no local recurrences were noted after the fifth year. For each group (Groups A-C), local control after EC alone was inferior to what has been described in the literature and accepted as the standard of care. However, when local recurrences developed, salvage treatment was possible in 83.3% of patients, resulting in an overall observed 85.6% local control rate. However, salvage treatment after EC was quite effective in many patients and was very much within the standard of care—especially for patients who had tumors of the glottic larynx. In fact, we observed that the final local control rate for patients who had glottic cancer (96%) was significantly better (P = .0001) compared with the rate for patients who had tumors originating from other locations (71%). This finding suggests that further studies using EC must use extremely rigorous and close follow-up before this approach is implemented widely.
Indeed, despite these promising results, the use of EC should not be considered as a standard of care for the treatment of patients with HNC. Multi-institutional, prospective studies of EC must be performed before such an approach is used widely or advocated. A single-institution experience in a large referral center could be performed in which close multidisciplinary follow-up is possible. Close observation by the physician for local or regional recurrence and patients' commitment to this regimen is absolutely essential for this approach. Further prospective evaluation in the setting of a clinical trial must be done.
Furthermore, not all patients with HNC are or should be amenable to this EC approach. We wish to highlight the potential advantages of EC, which advocates a purely neoadjuvant approach. With advances in genomics, proteomics, and systems biology, we believe that biomarkers can be identified that predict these important outcomes, as has been done in other tobacco-related cancers, such as lung cancer.28 Future prospective and multicenter trials of EC can be used to select these biomarkers and, thus, to identify which patients are best (or least) suited to receive chemotherapy alone as definitive treatment.
Some may argue that such an approach, in the face of highly effective radiation therapy with or without chemotherapy, is not indicated. However, even in the era of epidermal growth factor receptor-targeted therapy,26, 29 the mounting toxicity of current therapeutic regimens26, 30, 31 suggests that we should continue to explore alternative approaches for this challenging disease. For selected patients with laryngopharyngeal carcinoma, EC may be worth further consideration, especially for those patients who have tumors of the glottic larynx.
In conclusion, the current data confirmed that EC can achieve long-term remission and locoregional control without surgery or radiation in selected patients who have laryngopharyngeal SCC. Even for those patients who initially responded to EC and later developed recurrent disease, standard salvage therapy still was feasible in the majority. In fact, for this whole cohort, OS survival was comparable to results reported with more standard treatment. Finally, EC has permitted us to successfully down-modulate the extent of LR therapy while achieving OS benefits with a profile of treatment-related toxicity.
We are grateful to the following otorhinolaryngologists/head and neck surgeons: Henri Laccourreye, Daniel Brasnu, Jean Lacau St. Guily, Madeleine Ménard, François Janot, Régis Cauchois, and Ollivier Laccourreye; and to medical oncologists Vincent Bassot and Dan Atlan for allowing us to analyze the medical charts of their patients.
Conflict of Interest Disclosures
Supported in part by the Fulbright US Scholars Grant Program (Dr. Holsinger), administered by the Commission Franco-Américaine d'Échanges Universitaires et Culturels, Paris, France; by the Council for International Exchange of Scholars, Washington, DC; by the University of Texas Faculty Development Project; and by the PROGRES 2000 Association, Paris, France. Further research was supported by a support grant from The University of Texas M. D. Anderson Cancer Center (CA16672) by Specialized Program of Research Excellence in Head and Neck Cancer grant P50 CA97007 from the National Cancer Institute, and by the “Clinician Investigator Program in Translational Research” (grant K12 CA88084 to Dr. Holsinger).