Presented as a Candidate's Thesis to the American Laryngological, Rhinological and Otological Society, Inc.
Objectives/Hypotheses: Salvage surgery is widely viewed as a “double-edged sword.” It is the best option for many patients with recurrent cancer of the upper aerodigestive tract, especially when original therapy included irradiation, yet it may provide only modest benefit at high personal cost to the patient. The stakes are high because alternatives are of limited value. The primary objective of this study was to fully assess the value of salvage surgical procedures in the treatment of local and regional recurrence. The following hypotheses were developed to focus the study design and data analysis. 1) The efficacy of salvage surgery correlates recurrent stage, recurrent site, and time to presalvage recurrence. 2) The economic and noneconomic costs of salvage surgery increase with higher recurrent stage. 3) Information relating the value of salvage surgery to recurrent stage and recurrent site will be useful to these patients and the physicians who treat them.
Study Design: Two complimentary methods of investigation were used: a meta-analysis of the published literature and a prospective observational study of patients undergoing salvage surgery for recurrent cancer of the upper aerodigestive tract.
Methods: The meta-analysis combined 32 published reports to obtain an estimate of average treatment effect for salvage surgery with regard to survival, disease-free survival, surgical complications, and operative mortality. The prospective observational study included detailed data in 109 patients who underwent salvage surgery. In addition to parameters studied in the meta-analysis, we obtained baseline and interval quality of life data (Functional Living Index for Cancer [FLIC] scores), baseline and interval performance status evaluations (Performance Status Scale for Head and Neck Cancer Patients [PSS head and neck scores]), length of hospital stay, and hospital and physician charges, and related this data primarily to recurrent stage, recurrent site, and time to presalvage recurrence.
Results: The weighted average of 5-year survival in the meta-analysis was 39% in 1,080 patients from 28 different institutions. In the prospective study, median disease-free survival was 17.9 months in 109 patients, and this correlated strongly with recurrent stage, weakly with recurrent site, and not at all with time to presalvage recurrence. Noneconomic costs for patients and economic costs correlated with recurrent stage, but not with site. Baseline FLIC and PSS head and neck scores correlated with recurrent stage, but not with site. After salvage surgery the percentage of patients reaching or exceeding baseline was 51% for FLIC scores, and this differed significantly with recurrent stage. Postoperative interval “success” in PSS head and neck subscale scores for diet and eating in public also correlated with recurrent stage.
Conclusions: Overall, the expected efficacy for salvage surgery in patients with recurrent head and neck cancer was surprisingly good, but success was limited and costs were great in stage III and, especially, in stage IV recurrences. A strong correlation of efficacy and noneconomic costs with recurrent stage allowed the creation of expectation profiles that may be useful to patients. Additional systematic clinical research is needed to improve results. In the end, the decision to undergo salvage surgery should be a personal choice made by the patient after honest and compassionate discussion with his or her surgeon.
The best opportunity to cure patients who present with a squamous cell carcinoma (SCC) of the upper aerodigestive tract (UADT) is through the delivery of appropriate therapy at the time of the original diagnosis. The results achieved in the treatment of patients who have had recurrence after definitive cancer therapy are generally inferior to the results of therapy in previously untreated patients. Whereas initial treatment is successful (no evidence of recurrent disease at the time of last follow-up) in approximately 50% of newly diagnosed patients, 1,2 Gilbert and Kagan 3 combined the results of 26 studies published between 1967 and 1972 to report an average “salvage” rate (term not clearly defined) of only 16% for treatment of local/regional failures.
Recurrence of cancer at the primary site is relatively common compared with recurrence in the neck and recurrence at distant sites. Despite all of the advances of modern multidisciplinary care, a recurrent tumor at the primary site occurs in approximately 20% to 30% of patients and remains the most common cause of failure in patients treated for head and neck cancer. 1,2 Regional recurrence in the neck occurs in 10% to 15% of patients and is the next most common cause of disease-related death in these patients. 1,2 Death from distant metastases is relatively uncommon when local and regional disease is controlled, occurring in less than 1% of patients treated for cancer of the oral cavity 2 and in approximately 10% of patients treated for oropharyngeal cancer. 1
A variety of tumor-related and patient-related factors may influence the expected prognosis of patients who seek care for a recurrent SCC of the oral cavity, pharynx, and larynx. The site and stage of the recurrent cancer are likely to be important. The time elapsed between the completion of the initial definitive therapy and the diagnosis of recurrence (time to first recurrence) may represent indirect evidence of tumor virulence and/or host resistance. The general health and activity level of the patient, along with the presence or absence of specific comorbidities, may impact both host resistance and the capability of the patient to undergo and then complete therapy. Using a Cox proportional hazards model to analyze data from a study of 515 patients with recurrent disease after definitive radiation therapy, Stell 4 found that the time to first recurrence was the single most important factor affecting survival (P < .0001). Other significant factors influencing survival after recurrence included recurrent site (P < .005), general condition as measured by the Eastern Cooperative Oncology Group (ECOG) performance status scale (P < .05), and recurrent stage (P < .05).
Most head and neck oncologists agree that a salvage surgical procedure provides the best chance of long-term disease control and possible cure for operable patients with resectable, recurrent cancers. The extensive surgery that is often required, however, may be associated with short-term suffering from the surgery and related complications, permanent loss of function, readily visible deformity, high economic cost, and even death. Thus one must ask, when do the ends justify the means? Individual patients present a wide array of variables, making the answer to this question one of the most difficult judgments in the practice of head and neck oncology. Surprisingly, little good information is available in the current medical literature to provide a foundation for making informed decisions.
Determination of the value of a given medical treatment or procedure includes considerations of both quality (e.g., survival duration, quality of life [QOL]) and cost (e.g., complications, mortality rate, economic costs). Survival statistics and surgical complication rates, as important as they are, do not provide a full picture of the impact of surgical treatment on patients with cancer. Today the QOL experienced by the patient, as well as the performance status he or she achieves, can be assessed at intervals after treatment and compared with a baseline evaluation. Quality of life is generally measured by patient questionnaires that are validated in a cohort of patients for stability across raters, sensitivity to differences, and patient acceptability. Most patients in the United States think only secondarily, if at all, about the economic cost of therapy. The more relevant cost for them is related to the pain, suffering, and loss of function they incur. The term “noneconomic cost to patients” is used to refer to this issue hereafter. Given the difficult but well accepted fact that the total economic resources available to society are limited, economic costs are also relevant to a complete description of treatment-related outcomes.
The recent emphasis on outcomes research has been called “the third revolution in medical care” by Arnold Relman, 5 the influential editor of the New England Journal of Medicine. Proponents predict widespread benefits, including better information for patients and physicians, valid guidelines for medical practice, and better decisions by health care purchasers. Physicians and surgeons are concerned that expensive treatments of uncertain value will be disallowed by third-party payers unless the value of such procedures is demonstrated through modern outcome studies of acceptable validity.
Salvage surgery is widely viewed as a “double-edged sword.” It is the best option for many patients with recurrent cancer, especially when original therapy included irradiation, yet it may also provide only modest benefit to the patient at a high noneconomic cost. The stakes are high for individual patients who have had recurrence after definitive therapy, because options are limited. A more complete understanding of the utility of salvage surgery will also have an important role in treatment planning for newly diagnosed cancers, because many patients and treating physicians prefer to try other, less invasive modalities first, if surgical salvage is viewed as effective in the event of treatment failure. The primary objective of this study is to more fully assess the value of salvage surgical procedures in the treatment of patients with locally and regionally recurrent cancer of the oral cavity, pharynx, and larynx.
Specific aims in support of the study objective include the following.
1. Elucidation of recurrent site and recurrent stage prognoses that are specific to patients who undergo salvage surgery.
2. Determination of the effect of time to presalvage surgery recurrence (as measured from the completion of original therapy) on prognosis.
3. Assessment of the impact of salvage surgery on QOL.
4. Estimation of the economic cost of salvage surgery.
5. Assessment of the expected rates of complications and mortality associated with current surgical technique.
6. Development of recommendations for further study.
The following hypotheses were developed to focus the study design and data analysis.
1. The expected efficacy of salvage surgery correlates with recurrent stage (better in lower stage), time to presalvage recurrence (better if a longer time), and recurrent site (larynx is superior to oral cavity, which is superior to pharynx).
2. The social and economic cost of salvage surgery increases with higher recurrent stage.
3. Information relating the efficacy and noneconomic cost of salvage surgery to recurrent stage and recurrent site will be useful to patients and the physicians who treat them.
MATERIALS AND METHODS
The primary objective of study was addressed by two complementary methods of investigation: a meta-analysis of the published medical literature and a prospective observational study of consenting patients undergoing salvage surgery for recurrent SCC of the UADT.
A brief description of meta-analysis is in order, as it is one of the methods chosen to address the objective of this study. The term meta-analysis was coined by Glass et al. 6 in 1976 to define “… the statistical analysis of a large collection of results from individual literature, for the purpose of integrating the findings.” The method originated in the social sciences and has been applied to medicine with a seemingly higher standard of excellence. 7 It is, if you will, a more structured form of “review article,” which seeks the following.
1. To display the results of all relevant studies in a consistent and objective manner.
2. To resolve apparent conflicts in the results of pertinent studies.
3. To obtain an estimate of average treatment effect over all relevant studies.
4. To determine whether new, definitive clinical studies are needed and estimate the size of treatment effects that might be anticipated.
Although some authorities believe that “meta-analyses in medicine should almost always be limited to randomized clinical trials”, 7 not all issues requiring study lend themselves to this form of investigation. Despite significant limitations in the published literature pertaining to the subject of this thesis, the potential advantages described above make a compelling case for a meta-analysis.
A MEDLINE search of the English language literature was conducted for studies appearing between January 1, 1980, and May 1, 1998, and dealing with recurrent SCC of the UADT or head and neck. More than 1,000 reports were identified, and review of the abstracts established 417 studies that reported significant experience with treatment of patients with recurrent cancer. The majority of pertinent studies reviewed experience with the results of chemotherapy (n = 261). A smaller number of studies reviewed experience with surgical salvage (n = 48), radiation therapy (n = 58), biotherapy (n = 22), or combinations (n = 28).
Criteria for inclusion of published reports dealing with the surgical treatment of resectable, recurrent SCC of the UADT were established in advance. The minimal data reported in each study had to include 3-year and/or 5-year survival statistics taken from the date of salvage surgery or the 2-year disease-free survival rate following salvage surgery. The incidence of complications and the operative mortality associated with salvage surgery were recorded when available. Additional data linking the overall survival or 2-year disease-free survival data with the site of recurrence, recurrent stage, and time to recurrence were recorded when available.
Forty-eight publications reporting experience with surgical salvage for recurrent squamous cell cancer at the primary site were reviewed in detail. Of these, 32 reports met minimal criteria and were included in the meta-analysis. 8–39 Ten studies were excluded because they did not contain the information necessary to determine a survival figure, three studies were excluded because the data reported was derived from a duplicate or overlapping group of patients studied in other eligible reports, and three studies were excluded because all patients were treated according to a specific combined therapy strategy (one with surgery plus intraoperative radiotherapy, one with surgery plus brachytherapy, and one with surgery plus chemotherapy). Data were recorded on preprinted sheets and entered on a Microsoft Excel spreadsheet for review and analysis (see Appendix).
Among the eligible studies, there were no randomized clinical trials and no prospective observational studies. No studies included data from validated QOL questionnaires. Fifteen of the eligible studies focused solely on experience with salvage surgery for recurrent disease, while the other 17 reported a broader experience covering the treatment of patients throughout the course of their disease, including both initial treatment and surgical salvage for recurrences observed. Sixteen studies were from institutions within the United States, and 16 reported experience from other countries. The average values reported overall and for each group were weighted for the number of patients contributing to each average value.
Prospective Observational Study
Eligible patients selecting salvage surgery for the treatment of recurrent SCC of the UADT were invited to participate in the study, and informed consent was obtained. The staff of the Department of Otolaryngology was encouraged to offer the study to all eligible patients, but the majority were enrolled by the author, who was wholly or partially responsible for the care of 89 of 109 patients (82%). The author offered the opportunity to participate to all eligible patients. The setting was an urban academic medical center.
One hundred nine patients were entered between January 1991 and November 1997. The average age of the patients was 66.7 years. There were 26 female and 83 male patients. The ethnic classification included 5 blacks, 31 Hispanics, and 73 white non-Hispanics. The distribution of patients by site and stage is shown in Table I. The site of recurrence was the oral cavity in 21 patients (19%), the pharynx in 31 patients (28%), the larynx in 46 patients (42%), and the neck in 11 patients (10%). Recurrent stage was I in 14 patients (13%), II in 26 patients (24%), III in 29 patients (27%), and IV in 40 patients (37%). Fifty patients (46%) had been treated previously by surgery, 96 patients (88%) had received irradiation previously, and 18 patients (17%) had received chemotherapy previously. Time to presalvage recurrence was less than 6 months in 34 patients (31%), 6 to 12 months in 24 patients (22%), 12 to 24 months in 18 patients (17%), and more than 24 months in 33 patients (30%).
Table Table 1.. Distribution of 109 Patients in the Prospective Observational Study by Stage and Site.
Eligibility criteria included the following. 1) Biopsy-proven recurrent SCC of the oral cavity, pharynx, larynx, or neck was present. 2) The tumor was judged to be resectable after complete clinical evaluation. 3) The patient was considered operable.
General principles of salvage surgical procedure.
The recurrent primary tumor was resected with a 1.5- to 2.0-cm margin of surrounding normal tissue. The ipsilateral neck was generally cleared with a selective neck dissection when there were no palpable nodes. When neck metastases were present, the type of neck dissection depended on the extent of disease and the findings at the time of surgery. Bilateral neck dissections were strongly considered for patients with recurrent supraglottic cancer and for patients with midline recurrences at other sites.
Reconstruction was accomplished at the time of the salvage surgery. Priorities for selecting an individualized reconstructive plan included safety, function, and appearance—generally in that order. Unirradiated tissue was most often used for reconstruction after composite resection of the primary tumor and lymph nodes in patients whose original definitive therapy included a full course of ionizing radiation. Myocutaneous flaps (usually the pectoralis major) were used in 33 patients (30%) and free flaps with microvascular anastomosis were used in 4 patients (4%).
Rehabilitation and nutritional support.
Most patients were seen preoperatively by a speech pathologist for a discussion of the planned rehabilitation program. Speech and swallowing therapy was routinely provided following the surgical procedure. A percutaneous endoscopic gastrostomy was performed before salvage surgery when a period of prolonged postoperative dysphagia was expected.
Reirradiation or chemotherapy or both were frequently offered to the patient when the pathology report revealed perineural spread, vascular invasion, or lymphatic invasion. Nineteen patients (17%) received chemotherapy, and 15 patients (14%) received additional radiation therapy.
Evaluation of quality of life and performance.
The Functional Living Index for Cancer (FLIC) was developed and validated in 837 Canadian cancer patients by Schipper et al. 40 in the early 1980s. It consists of 22 questions, which cover a broad range of issues relevant to daily life such as pain, anxiety, confidence, depression, nausea, strength, sociability, impact of illness, and level of satisfaction. Each question is completed in Likert format on a scale of 10 to 70. Most patients complete the FLIC in less than 10 minutes. A single, overall score is derived by averaging the scores of all 22 questions (maximum score, 70). The FLIC is a validated “measure of the overall functional quality of a cancer patient's day-to-day life.”40
The Performance Status Scale for Head and Neck Cancer Patients (PSS head and neck) was designed by List et al. 41 to “measure the unique disabilities of head and neck cancer patients in areas of eating and speaking.” It was validated in a group of 181 patients with head and neck cancer by demonstration of its stability across raters and its sensitivity to functional differences. The PSS head and neck consists of three subscales: eating in public, intelligibility of speech, and normalcy of diet, each rated from 0 (most abnormal) to 100 (most normal) by a clinician talking with the patient for just a few minutes. Results for PSS head and neck are reported as three separately scaled scores.
Data were collected prospectively and entered on a preprinted data sheet. The data collected included the following. 1) Date of birth, sex, race, and ethnicity of the patient. 2) Site and stage of the original SCC. 3) Description of original definitive therapy. 4) Time to presalvage recurrence. 5) Site and stage of recurrent (presalvage) primary tumor. 6) Date, CPT codes, and charges for salvage surgical procedure. 7) Complications and hospital length of stay for salvage surgery. 8) Additional therapy given for the recurrence. 9) FLIC and PSS head and neck results completed before surgery and at 3, 6, 12, 18, and 24 months after surgery. 10) Date and site of second recurrence, if any. 11) Date and cause of death.
The value of a medical procedure is based on the ratio of its efficacy (the procedural correlate of quality) to its cost. To assess efficacy, the author chose median cancer-specific survival duration, 2-year disease-free survival rate, and QOL as quantified by the FLIC and the PSS head and neck. To analyze the QOL data, the author elected to determine the percentage of patients who reach or exceed the pretreatment baseline (percentage of patients who improved over baseline), calculate the average time in months taken to reach the pretreatment baseline (average time to improvement), and calculate the average percentage of improvement over baseline. Cost was represented by surgical complications, surgical mortality, hospital length of stay (a reflection of the difficulty of the postoperative recovery period for the patient and an indicator of the magnitude of hospital charges), and professional fees for the surgical procedure.
Data were entered on a customized Microsoft Excel spreadsheet for review and analysis. Statistical analyses were performed using programs from SAS Institute, Inc. 42 For descriptive purposes, the relative frequencies, means, and corresponding standard deviations were calculated for key variables. The product-limit (Kaplan-Meier) method was used for elementary survival analyses, and the Cox proportional hazards model was used for analysis of covariate-adjusted survival. 43
For analysis of survival, “death” was defined in one of two ways: first, as cancer-related deaths (deaths from unrelated causes were censored) and, second, as all deaths, regardless of cause (total mortality included deaths from cancer and deaths unrelated to cancer). The first approach to analysis gives a conservative estimate of the chance of avoiding death from cancer. The second approach to analyzing survival takes into account total mortality. Operative mortality, defined as any death within 30 days of salvage surgery, was always considered to be cancer related. There were 58 cancer-related deaths and 7 unrelated deaths (65/109 [60%]). Patients alive at the date of last known follow-up were censored at that point (they contributed useful information up to, but not after, that point).
The term “time to presalvage recurrence” is defined as the interval of time to the diagnosis of a presalvage recurrence, as measured from the date of last treatment for the “original” cancer. The term “recurrence-free interval” refers to the time from salvage surgery until another recurrence, if any, was observed. Patients who died of cancer without having a reported date of recurrence subsequent to the salvage surgery were, nevertheless, considered to represent a recurrent event. An estimated recurrence time was attributed to these patients (19 patients) by using the median recurrence time observed in patients who had similar survival time after salvage surgery and who had died after a recurrence.
Quality of life data were available at baseline in 107 of 109 patients. Eighty-four patients had at least one follow-up QOL score. The 23 patients who did not have a least one follow-up score were assumed to have failed to improve. This assumption is supported by data indicating that this group were more likely to have undergone surgery for advanced recurrent cancer and were more likely to die of their cancer. For analysis of FLIC, we defined a “successful outcome” as one in which recovery after salvage surgery exceeded or equaled baseline. For the analysis of QOL based on PSS head and neck results we defined a “successful outcome” as follows: 1) for patients with a baseline score below 75, a recovery after salvage surgery that exceeded baseline and 2) for patients with a baseline score of 75 or greater, a recovery after salvage surgery that exceeded or equaled baseline.
Thirty-two eligible studies reported experience with surgical salvage in a total of 1,633 patients. Of the 1,633 patients, 1,624 (99%) had been previously treated with a full course of radiation therapy. Six hundred seventy-seven patients (41%) had surgery for recurrent cancer of the larynx, 532 (32%) for recurrent cancer of the pharynx, 389 (24%) for recurrent cancer of the oral cavity, and 35 (2%) for other reasons.
Survival and disease-free survival.
A summary of 3-year survival, 5-year survival, and 2-year disease-free survival rates for all patients represented in the publications, and rates by site of presalvage recurrence is shown in Table II. Methods for the calculation of 3-year and 5-year survival varied among studies, and no attempt was made to adjust for this in the calculation of the weighted averages.
Table Table 2.. Weighted Averages of Survival End-Points From Publications in the Meta-analysis.
SALVAGE SURGERY FOR RECURRENT STAGE I/II LARYNGEAL CANCER . Ten of the 32 studies reported experience for salvage surgery in 199 patients with recurrent stage I/II cancer of the larynx. The 5-year survival rates available in three studies (68 patients) produced a weighted (for the number of patients) average of 83.4% (range, 80%–86%). The percentage of patients who were alive and disease-free at 2 years was available in eight studies (156 patients). The weighted average of 2-year disease-free survival in this group was 83.6% (range, 65%–100%). This group was analyzed separately because of its uniquely excellent prognosis, which is consistently demonstrated in the reports from different institutions. These statistics were then added back into the overall and site-specific laryngeal cancer data (as described in the paragraphs that follow).
OVERALL EXPERIENCE WITH SALVAGE SURGERY . Of the 22 remaining studies, all stages were combined in 19, while 3 studies dealt solely with stage III/IV. Three-year survival rates were given in 6 studies, representing a total of 319 patients. The weighted average of 3-year survival in these 6 studies was 37% (range, 28%–80%). The weighted average of 3-year survival at each site was, for the pharynx, 34% (range, 28%–80%) in 263 patients, and for the larynx, 53% (range, 33%–60%) in 56 patients. Five-year survival data were given in 12 of these studies, representing a total of 1,012 patients. The weighted average of the 5-year survival statistics in these 12 studies was 36.4% (range, 23%–55%). Adding back the unique patients with stage I/stage II laryngeal cancer, the weighted average of 5-year survival is 39.4% (range, 23%–86%) in 1,080 patients. Site-specific 5-year survival data were available in 607 of these 1,012 patients. The weighted average of 5-year survival at each site was as follows: oral cavity, 43.4% (range, 35%–55%) in 116 patients; pharynx, 26% (range, 23%–35%) in 266 patients; and larynx, 36.7% (range, 30%–39%) in 225 patients (167 cases, stage III/IV). Adding back the early-stage laryngeal cancer recurrences represented in the paragraph above, the weighted average of 5-year survival in 293 patients with all stages of laryngeal cancer was 47.5% (range, 30%–86%).
Two-year disease-free survival rates were provided in 11 of the 22 reports representing 343 patients. The weighted average was 36.3% (range, 0%–59%). Adding back the early-stage laryngeal cancer patients, the overall weighted average for 2-year disease-free survival was 51.1% (range, 0%–100%) in 499 patients. Site-specific 2-year disease-free survival data were available in all 343 patients. The weighted average of 2-year disease-free survival was as follows: oral cavity, 36.2% (range, 23%–45%) in 239 patients; pharynx, 24.6% (range, 0%–28%) in 57 patients; and larynx, 51% (range, 20%–59%) in 47 patients. Adding back the early-stage laryngeal cancer recurrences, the 2-year disease-free survival rate in 203 patients with recurrent laryngeal cancer was 76.1% (range, 50%–100%).
Surgical complications and operative mortality.
Operative mortality was reported in 18 studies representing a total of 718 patients. The weighted average was 5.2% (range, 0%–18%). The authors of 16 studies discussed surgical complications. The definitions of major and total complications across reports were reasonably consistent, and the following statistics were combined to give a rough idea of the personal cost experienced by these patients. The weighted average of total complications in seven studies, including 344 patients, was 39% (range, 10%–88%). Major complications were discussed in eight studies (388 patients). The weighted average was 27% (range, 5%–48%).
Association of survival with stage and time to presalvage surgery recurrence.
Surprisingly, no report presented a statistical comparison of survival rates or 2-year “no evidence of disease” (NED) rates for different stages at the time of salvage surgery, or for differences in the time interval between the original definitive therapy and the diagnosis of recurrence (time to recurrence). Of the 18 studies presenting results for patients with laryngeal cancer, 10 dealt solely with stage I/II, 3 dealt solely with stage III/IV, and 5 covered all stages. A 5-year survival rate was available in two of the stage III/IV studies involving 225 patients, producing a weighted average of 36.7% (range, 30%–39%). This compares to a 5-year survival weighted average of 83.4% (range, 80%–86%) in 68 patients with stage I/II recurrent laryngeal cancer. In all reports of experience with salvage surgery in recurrent oral cavity and recurrent pharyngeal cancer, survival statistics included all stages, making it impossible to establish a similar comparison.
Prospective Observational Study
Survival calculated by cancer-related deaths.
Taking cancer-related deaths as the end point, Table III shows that median cancer-specific survival time after salvage surgery in all 109 patients was 21.5 months. Median survival for men (21.7 mo) was significantly (P = .030) greater than that for women (12.2 mo), a result that cannot be attributed to differences in age, recurrent (presalvage) stage, recurrent (presalvage) site, or the other covariates shown in Table III (previous irradiation, previous chemotherapy, previous surgery).
Table Table 3.. Median Survival Times (Months) After Salvage Surgery.
Recurrent (presalvage) stage was a significant predictor of survival, but recurrent site or time to presalvage recurrence did not predict survival. Median survival by cancer site was 18.3 months for the oral cavity, 13.7 months for the pharynx, 30.8 months for the larynx, and 9.3 months for the neck (P = .419 [Table III ]). As one would expect, survival declined progressively and significantly with increasing presalvage stage (P = .0006): median survival was in excess of 24.3 months for stage I disease, in excess of 23.8 months for stage II disease, and was 21.7 and 9.3 months for stages III and IV disease, respectively. Increasing time to presalvage recurrence did not have a significant impact on the duration of disease-free survival after salvage surgery: the median survival by recurrence interval was 18.3 months for recurrence interval of less than 6 months, 24.3 months for recurrence interval of 6 to 12 months, in excess of 29.1 months for recurrence interval of 12 to 24 months, and 21.5 months for recurrence interval of more than 24 months (P = .844 [covariate adjusted test of trend]). Curves for cancer-specific mortality by recurrent stage, recurrent site, and time to presalvage recurrence are shown in Figures 1 through 3.
In contrast to the results observed for the analysis by recurrent (presalvage) stage, the stage of the original cancer bore no relationship to survival following salvage surgery. On the other hand, the results observed for the analysis by original site of SCC of the UADT revealed significant differences in median cancer-specific survival: oral cavity, 10.4 months; pharynx, 14.9 months; and larynx, 30.8 months (P = .043, after adjustment for recurrent [presalvage] stage, age at recurrence, and sex). The preceding analyses of original site and stage are based on 93 patients, the original TNM stage being unavailable in 16 patients.
Survival was related to whether a patient had received chemotherapy at some point before salvage surgery, but not to whether they had received previous irradiation or surgery. Patients who had previously received chemotherapy had a median survival of 8.8 months as compared with 26.9 months for those who had not received chemotherapy (P = .0007). The difference observed is not compatible with chance, nor can it be attributed to differences in age, disease stage, or the other covariates shown in Table III (sex, presalvage site, recurrence interval, recurrence site, previous irradiation, previous surgery). Patients who had previously received irradiation had a median survival of 21.7 months as compared with 8.0 months for those who had not, a difference compatible with chance (P = .582) after adjustment for covariates. For patients who had previously undergone surgery, median survival was 19.1 months as compared with 26.9 months for patients who had not (P = .499).
Table IV shows that 2-year disease-free survival after salvage surgery was 44% in 109 patients (estimate based on Kaplan-Meier analysis). The 2-year disease-free survival rate declined significantly with increasing stage of presalvage recurrence: 73% for stage I, 67% or less for stage II, 33% or less for stage III, and 22% for stage IV (P = .0005 [log-rank test]). The 2-year disease-free survival rate also differed by site of presalvage recurrence, but not at the traditional level of statistical significance (P = .0645): 28% or less for the pharynx, 47% or less for the oral cavity, 58% for the larynx, and 25% for the neck. The 2-year disease-free survival rate bore no relationship to time to presalvage recurrence: 39% or fewer of patients who had recurrence within 6 months were alive and free of disease 2 years after salvage surgery, as compared with 47% or less of patients who had recurrence within 6 to 12 months, 64% or less who had recurrence within 12 to 24 months, and 38% or less of patients who had recurrence after 2 years (P = .589 [log-rank test]).
Table Table 4.. Two-Year Disease-Free Survival After Salvage Surgery.
The stage of the original cancer was of no prognostic value for determining disease-free survival after salvage surgery. The median disease-free interval differed by original site, but not significantly: oral cavity, 4.8 months; pharynx, 7.8 months; and larynx, 24.5 months (P = .064).
Table III shows that median disease-free survival in all 109 patients was 17.9 months. Median disease-free survival for men (22.1 mo) was greater than that for women (7.3 mo). In contrast to the result for cancer-related survival, however, the male-female difference in disease-free survival was not statistically significant (P = .098), which is principally due to small sample size and the fact that two operative deaths in men were “censored” in the analysis of disease-free survival.
Recurrent (presalvage) stage was a significant predictor of disease-free survival, but presalvage site or time to presalvage recurrence did not predict disease-free survival. Median disease-free survival by cancer site (Table III) was 9.2 months for the oral cavity, 7.8 months for the pharynx, in excess of 24.5 months for larynx, and 7.2 months for the neck (P = .625). There was a significant trend (P = .0004) toward decreasing disease-free survival with increasing presalvage stage: median disease-free survival was in excess of 22.1 months for stage I disease, was in excess of 11.5 months for stage II disease, and was 14.4 and 5.5 months for stage III and IV disease, respectively (Table III). Time to presalvage recurrence did not have a significant impact on disease free survival: the median disease-free survival by recurrence interval was 8.7 months for time to presalvage recurrence of less than 6 months, 22.1 months for 6 to 12 months, in excess of 10.4 months for 12 to 24 months, and 10.9 months for more than 24 months (P = .344 [covariate adjusted test of trend]).
Disease-free survival was related to whether a patient had received chemotherapy at some time before salvage surgery, but not to whether a patient had received irradiation or undergone surgery. Patients who had previously received chemotherapy had a median disease-free survival of 5.0 months as compared with 24.5 months for those who had not received chemotherapy (P = .001). The difference observed is not compatible with chance, nor can it be attributed to differences in age, disease stage, or the other covariates shown in Table III (sex, presalvage site, recurrence interval, recurrence site, original irradiation, original surgery). Patients who had previously received irradiation had a median disease-free survival of 15.1 months as compared with 4.8 months for those who had not, a difference compatible with chance (P = .376) after adjustment for covariates. For patients who had previously undergone surgery, median disease-free survival was 10.4 months as compared with 24.5 months for patients who had not (P = .531).
Since only 7 of the 65 deaths observed in this study were not cancer related, the analysis of cancer-related mortality presented above reflects fairly accurately what occurred for total mortality. Nevertheless, a few salient points with respect to all deaths observed on study are summarized below.
Median survival was 21.0 months. Median survival for men (21.7 mo) was significantly (P = .023) greater than that for women (12.2 mo), a result that cannot be attributed to differences in age, recurrent stage, recurrent site, previous irradiation, previous chemotherapy, or previous surgery. Recurrent (presalvage) stage was a significant predictor of survival, but not of presalvage site or time to presalvage recurrence.
Quality of life data.
GENERAL QOL AS MEASURED BY THE FLIC . FLIC scores, a measure of overall QOL, were available in 107 patients at baseline, for which the average score was 47 ± 9 on a scale of 10 to 70. The FLIC scores at baseline (Table V) did not correlate with the site of presalvage recurrence, but did show a significant decline with increasing stage of presalvage recurrence: the average FLIC was 51 ± 8 for stage I, 48 ± 8 for stage II, 48 ± 10 for stage III, and 43 ± 7 for stage IV (P = .040). After surgery, the percentage of patients showing improvement in FLIC (Table VI) was 51%, and this varied significantly by stage of presalvage recurrence: 64% for stage I, 65% for stage II, 41% for stage III, and 39% for stage IV (P < .03). There were no significant differences, however, by cancer site in the percentage of patients showing improvement over time. For the patients who did improve, the average increase in FLIC score over baseline was 16% (range, 0%–66%). The extent of improvement did not vary with recurrent (presalvage) stage or site. The mean time to reach improvement over baseline, for the group of patients who achieved this, was 4.8 months.
Table Table 5.. Baseline Quality of Life Scores ± Standard Deviation by Site and Stage.
Table Table 6.. Percent of Patients With a Postsalvage Successful Outcome* on Quality of Life, by Presalvage Site and Stage.
SITE-SPECIFIC PERFORMANCE STATUS AS MEASURED BY PERFORMANCE STATUS SCALE FOR HEAD AND NECK CANCER PATIENTS . PSS head and neck scores were available in 107 patients at baseline, and 84 patients had at least one follow up score. Data are presented according to the three subscales.
PERFORMANCE STATUS SCALE FOR HEAD AND NECK CANCER PATIENTS: NORMALCY OF DIET . For all patients, the average baseline score for normalcy of diet (“diet”) was 67 ± 36 (SD). The baseline dietary scores (Table V) varied significantly by site of presalvage recurrence (P = .005): 64 ± 33 for the oral cavity, 54 ± 35 for the pharynx, 74 ± 35 for the larynx, and 75 ± 36 for the neck. As one might expect, the baseline score for diet decreased significantly as stage of presalvage recurrence increased (P < .001): the average diet score was 94 ± 14 for stage I, 65 ± 33 for stage II, 77 ± 28 for stage III, and 49 ± 40 for stage IV
After salvage surgery, 47% of patients (50/107) had a successful outcome with respect to diet (Table VI). This percentage varied significantly by site of presalvage recurrence (P < .01): 25% for the oral cavity, 32% for the pharynx, 73% for the larynx, and 18% for the neck. In addition, the percentage of patients with a successful outcome declined significantly with increasing presalvage stage (P < .01): 71% for stage I, 62% for stage II, 45% for stage III, and 29% for stage IV For the 50 patients who had a successful outcome on diet, the improvement on average was 30% above their baseline score, and mean time to “success” was 4.4 months.
PERFORMANCE STATUS SCALE FOR HEAD AND NECK CANCER PATIENTS: UNDERSTANDABILITY OF SPEECH . For all patients, the average baseline score for understandability of speech (“speech”) was 83 ± 21 (SD). The average scores varied by presalvage site (Table VI), but not significantly, as follows: 88 ± 24 for the oral cavity, 84 ± 20 for the pharynx, 82 ± 23 for the larynx, and 77 ± 13 for the neck. There was also no statistically significant relationship between average score for speech at baseline and presalvage stage: 82 ± 15 for stage I, 91 ± 12 for stage II, 84 ± 24 for stage III, and 76 ± 25 for stage IV
After surgery, 41% of patients (44/107) had a successful outcome with respect to speech (Table VI). The percentage varied by presalvage site, but not significantly (P = .20): 30% for the oral cavity, 32% for the pharynx, 53% for the larynx, and 36% for the neck. The percentage of patients with a successful outcome on speech did decline significantly with increasing presalvage stage (P = .001): 86% for stage I, 58% for stage II, 34% for stage III, and 18% for stage IV. For the 44 patients who had a successful outcome for speech, the improvement averaged 2% above their baseline score, and the mean time to “success” was 5.5 months. The low percentage of improvement is due to the fact that many patients had baseline scores of 100 and returned to 100, each such patient contributing a 0% improvement to the group.
PERFORMANCE STATUS SCALE FOR HEAD AND NECK CANCER PATIENTS: PUBLIC EATING BEHAVIOR . For all patients, the average public eating behavior (“public eating”) score at baseline was 66 ± 40 (SD). The average varied by presalvage site, but not significantly, as follows: 63 ± 40 for the oral cavity, 56 ± 40 for the pharynx, 78 ± 36 for the larynx, and 57 ± 50 for the neck. The average public eating score at baseline declined significantly with presalvage stage (P < .001): 95 ± 14 for stage I, 69 ± 41 for stage II, 78 ± 36 for stage III, and 46 ± 40 for stage IV
After surgery, 50% of patients (54/107) had a successful outcome for public eating (Table VI). The percentage varied significantly by presalvage site (P < .01): 35% for the oral cavity, 35% for the pharynx, 73% for the larynx, and 27% for the neck. There was also a significant trend of declining percentage “success” with increasing presalvage stage (P = .001): 86% for stage I, 58% for stage II, 48% for stage III, and 34% for stage IV. For the 54 patients who had a successful outcome for public eating, the average improvement was 17% above their baseline value, and the mean time to “success” was 4.8 months.
Operative mortality and complications.
Two patients died within 30 days of salvage surgery for an operative mortality of 1.8%. Both of these patients were classified as having recurrent stage IV disease before salvage surgery.
Significant complications occurred in a total of 22 patients (20.2%) and included pneumonia (2 patients), heart problems (3 patients), hemorrhage (1 patient), wound infection (5 patients), flap necrosis (2 patients), and fistula formation (6 patients). Sixteen patients had one complication, 5 patients had two complications, and 1 patient had three complications. The frequency of complications correlated with presalvage stage of recurrence (P = .060): 0% for stage I, 12% for stage II, 29% for stage III, and 28% for stage IV.
Length of hospital stay for salvage surgery.
The average length of the hospital stay (LOS) for salvage surgery was 11.5 days in 108 patients. For stage I, LOS was 6 days; for stage II, 7.9 days; for stage III, 13.3 days; and for stage IV, 14.4 days. Relating LOS to site, LOS for oral cavity was 12.6 days; for pharynx, 13.3 days; for larynx, 10.5 days; and for neck, 8.4 days.
Hospital and physician charges for salvage surgery.
The average total charge per hospital day on the surgical service was $5,700. Therefore the average hospital charge for the salvage surgery admission in all patients was $65,550. Average hospital charges for the salvage surgery admission by stage were $34,200 for stage I, $45,030 for stage II, $75,810 for stage III, and $82,080 for stage IV. Average hospital charge by site was $71,820 for oral cavity, $75,810 for pharynx, $59,850 for larynx, and $47,880 for neck. The average physician charge for the salvage surgery was $10,093 per patient in all patients and correlated with recurrent stage, but not with site. The average physician charge was $7,485 for stage I, $7,397 for stage II, $11,382 for stage III, and $11,922 for stage IV.
The initial diagnosis of cancer in the head and neck region is devastating to patients. In addition to the discomfort of current symptoms, patients sense that this disease may have a significant impact on their life span, their ability to interact with others, and their identity. 44 A recent study suggested that survival, pain, and activity level are of even greater concern to these patients at the time of diagnosis than are appearance and critical functions such as eating and speech. 45 A diagnosis of recurrence redoubles these concerns and reduces the patient's faith in the value of treatment. Compassionate communication of accurate and meaningful information about the likelihood of survival, the expected QOL, and the noneconomic cost of the treatment is of paramount importance. Such information has not been readily available in published reports to date.
The data from the meta-analysis and the prospective outcomes study reported in this thesis are complementary and reasonably consistent. The data from the meta-analysis is limited in its breadth and depth, but represents experience with a large number of patients treated at 32 different institutions. The prospective observational study adds relatively high-quality data covering a broad range of issues, but from a small number of patients treated in a single institution.
It is not possible to fully characterize the patients in the meta-analysis with regard to age, sex, ethnicity, or stage distribution, but we do know that nearly all of them had received a full course of radiation therapy at some time before the recurrence of their cancer and the subsequent salvage surgery. The patients in the prospective clinical study are mostly men (76%) with an average age of 67 years, advanced recurrent disease (27%, stage III; 37%, stage IV), and extensive prior treatment (88%, previous irradiation; 46%, previous surgery; 17%, previous chemotherapy).
The 5-year survival of 39.4% in 1,080 patients of the meta-analysis represents our best estimate of long-term overall survival in a group of patients who undergo salvage surgery for recurrent disease. When one thinks about this group of patients as a whole, salvage surgery conveys a reasonably good chance of long-term survival. However, survival data were not available by stage or time to presalvage recurrence, demonstrating one of the limitations of the meta-analysis. Five-year survival tended to be better in recurrent cancer of the larynx and oral cavity (larynx, 47.5% in 293 patients; oral cavity, 43.4% in 116 patients) than in pharynx (26% in 266 patients), but the author does not believe that these numbers are strictly comparable and is unwilling to draw firm conclusions from this apparent difference.
Five-year survival data are not yet available for the prospective observational study because of the relatively short duration of follow-up after salvage surgery, but median cancer-specific survival for the entire group was 21.5 months. Recurrent stage was a highly significant predictor of the duration of cancer-specific survival, but the site of recurrence was of only marginal significance. Median cancer-specific survival was just 9.3 months for patients with a stage IV recurrence, but was considerably better for stage III (21.7 mo) and has not yet been reached for stage I or II (>2 y for both). Median disease-free survival after salvage surgery in patients with a stage IV recurrence is just 5.5 months, a finding that demonstrates the poor outlook for the majority of these individuals. Neither time to presalvage recurrence nor the stage of the original cancer predicted duration of survival after salvage surgery in this study. Survival did not vary with original stage, but did differ significantly with original site.
Sex was found to have an unexpected effect on cancer-specific survival, with men living longer than women. The author has no explanation for this finding. Patients who had received chemotherapy at some time before salvage surgery also did poorly with regard to survival when compared with those who had never received chemotherapy. Again, there is no clear explanation, although one might suspect that patients selected for chemotherapy could have displayed various reasons to suspect poor prognosis that were not controlled for in this study design. It should be noted that neither of these associations represents study postulates; thus these post hoc observations may be due to chance or uncontrolled confounding.
Two-Year Disease-Free Survival
Ninety percent of patients who are going to have recurrence after treatment of SCC of the UADT do so within 2 years. 1,2 Thus disease control at 2 years is an important benchmark for the efficacy of treatment. Considering the number of patients in the prospective study who underwent salvage surgery for advanced, recurrent disease, it is rather surprising how well these patients did as a whole. The overall 2-year disease-free survival rate of 44% for the prospective study is consistent with the weighted average of 51% calculated from the meta-analysis. Recurrent stage and site were both significant predictors of the percentage of patients who were alive and disease free at 2 years, although the correlation with recurrent site was of marginal significance. Interestingly, and perhaps importantly, there was less separation between the efficacy of salvage surgery in stage III patients (33% alive and free of disease) and stage IV patients (22% alive and free of disease), when looked at in this way, as compared with the difference in median survival duration. Apparently, a fairly similar percentage of patients with stage III and stage IV disease live for 2 years after salvage surgery without having recurrence, but the stage IV patients who fail do so much more quickly. Much of the apparent difference in 2-year disease-free survival by site (58% for patients with recurrent laryngeal cancer >47% for patients with recurrent oral cavity cancer >28% for patients with recurrent pharyngeal cancer >25% for patients with recurrent cancer of the neck) is actually due to differences in stage distribution for the four sites, as seen in Table I.
Site-related and stage-related disease control data are reasonably similar when they are available for both methods of study (Table IV), suggesting that the more detailed findings of the prospective study may be useful to form expectations for patients treated at other institutions as well.
Operative Mortality Rate and Surgical Complications
Although the data from the two methods of study are not strictly comparable, the lower mortality rate in the prospective study, as opposed to the meta-analysis (1.8% vs. 5.2%) is probably due to growing experience with more reliable reconstructive techniques and is likely to be a more accurate reflection of the risk that patients face today. The surgical procedures of the prospective study were all performed after January 1991, while some of the operations reported in the meta-analysis date back to the early 1970s. Both operative mortalities in the prospective study occurred following surgery for advanced disease.
Surgical complications are an important problem, reported in 20% of the patients in the prospective study and in 39% of 388 patients in the meta-analysis. As expected, the frequency of complications in the prospective study is greater following the more extensive surgery required for stage III (29% frequency) and stage IV (28% frequency) recurrent SCC of the UADT. One should also remember that patients in both the meta-analysis and the prospective observational study were selected, at least partially, for their capability to withstand surgery. Wound complications seem to be a manageable problem, despite the fact that the vast majority of patients represented in both study methods had received a full course of radiation therapy (99% in the meta-analysis and 88% in the prospective study) at some time before the salvage surgery. Wound complications occurred in just 10% of patients in the prospective study. The availability of myocutaneous flaps, as well as free flaps with microvascular anastomosis, provides the surgeon with a wide array of opportunities to bring unirradiated tissue into the surgical field for reconstruction. Prior irradiation is no longer a barrier to the use of surgical therapy.
Quality of Life
The QOL data are more difficult to interpret. Despite a great deal of ongoing study, good quality of life has not been clearly defined, and there is no single instrument or battery of instruments that is generally accepted as the best for evaluating patients. Moreover, there is presently no standard way of evaluating and communicating this data.
Most investigators use at least two validated questionnaires, one to evaluate the status of general life issues that are important to all individuals and a second to focus on a variable number of specific domains that are of particular significance to the group and topic of study. The author selected the FLIC and the PSS head and neck from many available measures. The FLIC was chosen to measure overall QOL because of its ease of administration and its coverage of general issues that are pertinent to this group of patients, such as pain, activity, and social interaction. The PSS head and neck was selected as a measure of performance because of its focus on the most important functional problems faced by these patients and its ease of administration.
The author chose to analyze the data in a relatively simple way, and the chosen method is important enough to the interpretation of the study to be reviewed. Using baseline scores obtained before salvage surgery, averages and standard deviations were calculated for the group as a whole and for each subgroup. Each subgroup was then tested for significant differences between it and other subgroups within the same classification (e.g., stage with stage and site with site). To analyze the interval data, the author chose to compare each patient's follow-up data to his or her own score at baseline. For the FLIC, an individual outcome was considered successful if any follow-up score equaled or exceeded baseline. If scores later worsened, perhaps due to a second recurrence, the outcome was still considered a success. The percentage of patients who achieved success was then determined for each subgroup and for the group as a whole. The author reasoned that patients want to know their chance of getting better after surgery, but they do not expect to remain better if the cancer recurs again. Interval PSS head and neck subscale scores were analyzed similarly, but success required improvement over baseline when the baseline score was less than 75 and return to baseline or better when the original score was equal to or greater than 75.
The average FLIC score for the group at baseline indicates that the recurrent cancer had a relatively consistent, adverse effect on overall QOL. Average FLIC score at baseline did decline significantly, though modestly, with increasing presalvage stage, but did not correlate with the site of presalvage recurrence. The expected variability of the FLIC score supports the validity of the instrument in this group of patients and suggests that the effect of the cancer on overall QOL, or at least, on the patient's perception of that quality, is more dependent on the extent of disease than it is on the location of the recurrent cancer.
The averages of the baseline scores for the three subscales of PSS head and neck are indicative of suboptimal eating and speech function, which was probably due to the recurrent cancer. In addition, the standard deviations suggest a large amount of variation within the group as a whole. Scores for normalcy of diet (diet) and the ability to eat in public (public eating) are lower than the score for the understandability of speech (speech), but this difference means nothing, since the scales are not standardized. The baseline scores for PSS head and neck public eating and PSS head and neck diet correlate with recurrent stage, and PSS head and neck diet varies significantly by site of presalvage recurrence (patients with recurrences in the larynx and the neck performed better). Baseline scores for PSS head and neck speech were high at baseline and did not vary significantly by stage or site. This may be due to greater reserve in the physiology of speech function, and/or to greater resilience of patients regarding their speech when compared with their eating.
Follow-up QOL data were not available in 23 of 107 patients who completed questionnaires before salvage surgery. The cause of the lack of follow-up was not completely clear, but the author made the conservative assumption that these patients did poorly with regard to QOL and included them in the denominator of the fraction of patients who reached or exceeded baseline QOL, even though they had no opportunity of inclusion in the numerator. Thus the percentage of patients reported as achieving “success” in QOL relative to baseline represents a conservative estimate.
As with the survival and disease control statistics, the patients in this study did surprisingly well with regard to their interval FLIC scores after salvage surgery when compared with the scores at baseline. The majority of patients (51%) achieved their baseline status or improved on it within an average of 4.8 months, and the average increase over baseline was 17%. The likelihood of these successful outcomes correlated with recurrent (presalvage) stage, but not with recurrent site. Patients operated on for stage IV recurrences had only a 39% chance of achieving or exceeding their baseline FLIC score after surgery. Similarly, follow-up scores in many of the patients achieved or exceeded baseline for PSS head and neck diet (47%) and PSS head and neck public eating (50%) in about the same timeframe, and the likelihood of improvement declined significantly with increasing recurrent stage and correlated with recurrent site (larynx better than oral cavity and pharynx). For PSS head and neck speech, scores returned to baseline or better less frequently (41%), but this may be due, at least partially, to higher baseline scores. The likelihood of a successful outcome on PSS head and neck speech after salvage surgery was strongly correlated with lower recurrent (presalvage) stage, but not with the site of presalvage recurrence.
Patient Expectation Profiles for Salvage Surgery
Do the results of salvage surgery justify its use for individual patients? The patient must decide following an informed discussion with a knowledgeable surgeon who is experienced in the multidisciplinary care of patients with SCC of the UADT. A supportive family or an advocate/friend can be of great assistance to the patient in understanding and interpreting this complex information, but in the end, many patients appreciate and accept the surgeon's advice. By synthesizing the data of the prospective observational study and the meta-analysis, it is possible to construct expectation profiles that may be of assistance to patients and surgeons. In these profiles, information on survival, surgical complications, and QOL expectations depend mainly on recurrent stage, since that is the most significant predictor for each outcome. Additional factors, such as the site of recurrence and the presence of other medical problems, should be considered as a part of the judgment involved in making recommendations to individual patients. Review of the pertinent information at a multidisciplinary head and neck tumor conference may also help the surgeon with this complex judgment.
Patients with stages I and II recurrent cancer.
Following salvage surgery these patients have a 70% chance of living for at least 2 years without recurrence of their cancer and a 60% to 85% chance of achieving a successful outcome with respect to their QOL. Significant surgical complications occur in about 6% of patients, and death related to the surgical procedure is rare.
Patients with stage III recurrent cancer.
Approximately one third of these patients live for at least 2 years without recurrence of their cancer. Half of the patients live for more than 14 months before their cancer recurs again, and for more than 22 months until death. However, half of the patients have recurrence before 14 months and die before 22 months. Forty percent of patients have a successful outcome in their QOL at some point after salvage surgery, but this may be temporary. Significant complications occur in about 30% of patients after salvage surgery, and the average hospital stay is a little less than 2 weeks. Death that is due to the surgery is unlikely, occurring in less than 2% of patients.
Patients with stage IV recurrent cancer.
Less than one fourth of these patients live for at least 2 years without recurrence of their cancer. Half of the patients live for more than 5 months before their cancer recurs again, and for more than 9 months until death. However, half of the patients have recurrence before 5 months and die before 9 months. Approximately 30% of patients have a successful outcome in their QOL, and this may be temporary. Significant complications occur in about 30% of patients after salvage surgery, and the average hospital stay is a little more than two weeks. Death due to the surgery is unlikely, occurring in less than 2% of patients.
Other Treatment Options
The role of surgical salvage in the management of patients with recurrent SCC of the UADT must be understood within the context of the efficacy of other options. Indeed, the process of obtaining informed consent for any therapeutic procedure requires a full discussion of the pros and cons of all alternatives. Alternatives for treating patients with recurrent SCC of the UADT include 1) reirradiation, if the initial therapy included a full course of radiation therapy; 2) chemotherapy; 3) biotherapy, including immunotherapy and gene therapy; 4) combinations thereof; and 5) hospice or supportive care, or both.
The efficacy of reirradiation in patients who have had recurrence after definitive therapy, which included a full course of ionizing radiation, may be adversely affected by local scarring and fibrosis, as well as difficulty in achieving adequate dosage because of the proximity of vital structures which are already near tissue tolerance. In addition, the recurrence of the cancer suggests a degree of radiation resistance if the initial therapy was properly planned and delivered. Moreover, complication rates are higher because of excess dose to adjacent nontumor tissue. While skillful retreatment planning and the use of brachytherapy may ameliorate some of the above problems, cure rates are low and complications are common. Levendag et al. 46 reported achieving local control in 29% of 55 patients treated with external-beam reirradiation and in 50% of 18 patients treated with a combination of external-beam reirradiation and interstitial radiation therapy. Actuarial survival at 5 years was 20% in both groups. Severe side effects were experienced by 28% of patients, but no treatment-related deaths occurred. Similarly, Stevens et al. 47 reported a local control rate of 27% and 5-year actuarial survival of 17% in 85 patients receiving high-dose reirradiation for recurrent head and neck cancer. Wang 48 emphasized the importance of patient selection and offered the opinion that “the indiscriminate use of high-dose reirradiation often makes a bad matter worse.”
Currently available drugs have not yet been associated with a significant potential for cure or long-term disease control. Combination chemotherapy regimens that include cisplatin and 5-fluorouracil produce high rates of objective response in previously untreated patients, but are less often effective in patients with recurrent disease. Objective responses occur in approximately one third of previously treated patients, are generally partial, and are of short duration, with a median survival of 4 to 6 months. 49–52 In a study of 102 patients with recurrent or metastatic SCC, Recondo et al. 53 found that age less than 60 years, good performance status, absence of prior therapy, and absence of local/regional recurrence were all significant predictors of response to therapy. Median survival was 9 months for responders and 5 months for nonresponders. The degree of palliation achieved in this group of patients is also important. Constenla et al. 54 recently reported experience with symptom relief by combination chemotherapy in 52 patients with evaluable disease. Objective responses were seen in 35% of patients. Symptoms improved in 67% of patients who had an objective response and in 33% of nonresponders. Pain improved in 11 of 28 patients in whom it was present, and dysphagia improved in 6 of 18 patients. The recent addition of paclitaxel to combination chemotherapy regimens shows some promise for improving response rates, but a positive impact on survival has not yet been demonstrated. 55,56
Treatment with biological agents, alone or in combination with chemotherapy, remains experimental. Responses have been seen with a variety of agents, including locally injected interleukin-2, 57 locally injected OK-432 (a multicytokine inducer), 58 and locally injected gene therapy with wild-type p53, 59 but reliable treatment regimens have not yet been established.
Combinations of nonsurgical therapy.
Combinations of nonsurgical therapies have also been tried, most with only modest success. However, Tan et al. 60 recently described impressive results with aggressive concurrent chemoradiotherapy in nine patients with unresectable, recurrent SCC of the UADT. Seven patients had received prior radiation therapy and seven patients had prior surgery. At the time of the report, five of nine patients remained free of disease at 41 to more than 50 months.
Hospice-style supportive care may be the best choice for some patients. All physicians take a vow to “first, do no harm.” The wise physician counsels his or her patient about the shortcomings of aggressive therapy and recommends supportive care when this advice is appropriate.
Better definition of the value of surgical salvage is of significance for society as a whole. SCC of the UADT is one a relatively common solid tumor. The American Cancer Society has estimated that there were 41,650 new cases diagnosed in the United States in 1997. 61 At least one third of all patients have recurrence at the primary site or in the neck following initial therapy. 1–3 The cost of medical care has become increasingly important to our society as its share of the gross national product has grown and American corporations have felt the effects of global competition.
The actual economic cost of medical treatment is not easily determined. A percentage of the charges submitted to third-party payers by hospitals and physicians represents a reasonable approximation of the economic cost to society. Using the total of average hospital and professional charges, the economic cost of salvage surgery in all 13,883 patients who recur annually approximates $10,500,000. Does the efficacy of this treatment justify its cost to society? This study does not provide the answer, but it does focus the question. Is it appropriate for society to allocate $20,842 ($7,485 professional charge + $34,200 hospital charge × 50% payment rate) for a patient with a stage I or II recurrence to have a 70% chance of long-term (>2 y) disease control and a 70% chance of improved overall QOL? The high level of efficacy and low cost (noneconomic and economic) of salvage surgery in patients with stage I/II recurrence speaks for itself. On the other hand, is it reasonable for society to allocate $47,000 for a patient with a stage IV recurrence to have a 23% chance of long-term (>2 y) disease control and a 30% chance of improved overall QOL? In this case there might be a conflict between the point of view of the patient and that of society, but most authorities would agree that the physician's role is to help each individual patient and his or her family make an informed decision after considering all options, including hospice care.
This study defines the efficacy and the costs of salvage surgery for recurrent head and neck cancer. It does not address the question of the best therapy or test a new therapeutic strategy. The efficacy of salvage surgery for patients with stage III and IV recurrent cancer is modest, and the noneconomic cost for a significant percentage of patients is great. We would certainly like to do better.
Head and neck oncologists have put a great deal of effort into the search for better multidisciplinary treatment strategies for newly diagnosed patients presenting with advanced disease, but there have been essentially no systematic studies of treatment combinations for patients with recurrent disease. Many of the patients in the meta-analysis and in the prospective observational study reported in this thesis received reirradiation and/or chemotherapy, in addition to salvage surgery, but there was no attempt to define the efficacy of the additional therapy. Huang et al. 62 reported an intriguing experience with a combination of salvage surgery and multidrug chemotherapy in 1983. Eight of 10 patients were alive and free of disease with follow-up of 20 to 65 months. Phase I experience with intraoperative radiation therapy in 47 patients 63 and with brachytherapy in 35 patients 64 has also been reported. No definitive trials have been conducted.
A randomized clinical trial could be designed to compare the results of salvage surgery with the results of a nonsurgical option (e.g., chemotherapy plus reirradiation) or to test the value of combining other therapies (e.g., adjuvant chemotherapy and/or reirradiation) with salvage surgery as compared with salvage surgery alone. The author believes that either approach would be ethical and that the issue is of sufficient importance to justify the effort. Either trial would be best conducted by one or more of the multi-institutional groups, and trial design would be best addressed by the multi disciplinary head and neck committees of those groups.
Do the ends justify the means? The results of salvage surgery clearly justify its use in approximately 70% of the patients who are treated for an early-stage (stage I/II) recurrence and in 25% to 30% of the patients who have surgery for a late-stage (stage III/IV) resectable recurrence. But what about the many patients who are subjected to the hardships of salvage surgery only to see no real or lasting benefits? We would like to be able to better identify the patients who are most likely to benefit, and we would especially like to improve the overall results by increasing the benefits and reducing the noneconomic cost to patients.
The expected efficacy of salvage surgery is strongly correlated with the stage of the recurrent cancer, but only weakly correlated with the site of the recurrent cancer, and not at all correlated with time to presalvage recurrence. Survival and disease-free survival correlate best with the recurrent stage, and improvement in QOL correlates with both recurrent stage and with the site of recurrence. The cost of salvage surgery for the patients in terms of complications, surgical mortality, length of hospital stay, and reduction in QOL also correlates best with recurrent stage.
Patients and families need useful information regarding the realistic expectations for salvage surgery and for alternative therapies, to make the best possible decision on how they wish to proceed. The patient expectation profiles for salvage surgery that are described above may be useful in that regard.
Recurrent SCC of the UADT afflicts more than 13,500 patients each year, and we have not yet done enough to study this problem systematically. We should design randomized clinical trials for patients with advanced stage, resectable recurrences to test the efficacy of nonsurgical combination therapies and to develop improved treatment strategies by adding adjuvant therapy to surgical salvage.
The author is especially grateful to Ms. Kara Hamilton and to Dr. James Schlesselman of the Biostatistics Division of the Sylvester Comprehensive Cancer Center for assistance with study design and data analysis.
This study could not have been conducted without the assistance of the head and neck oncology nurses in the Department of Otolaryngology at the Sylvester Comprehensive Cancer Center and Jackson Memorial Hospital. The author is grateful to Ms. Donna Benjamin, Ms. Belkis Rodriguez, and Ms. Zenaida Recendiz, all of whom assisted with the patient care and data management. Ms. Penny Fisher provided constructive criticism of the manuscript.
Dr. Marcie List of the University of Chicago (Chicago, IL) and Dr. Jay Picarillo of Washington University (St. Louis, MO) reviewed the manuscript and provided input that improved the manuscript significantly.
Ms. Angie Monnar spent many hours assembling and reassembling the manuscript.
Special thanks to Dr. Ryan Chandler for his guidance throughout the author's development as a head and neck oncologist, and to Dr. Thomas Balkany for his steadfast support during the 7 years of this project. Dr. Francisco Civantos, Dr. Don Weed, Dr. David Arnold, and colleagues in the Division of Head and Neck Oncology of the Department of Otolaryngology all provided valuable advice.