Kaplan-Meier curves showing survival following reirradiation in surgical patients (n = 38) and in nonsurgical patients (n = 49).
Customize your JAMA Network experience by selecting one or more topics from the list below.
Iseli TA, Iseli CE, Rosenthal EL, et al. Postoperative Reirradiation for Mucosal Head and Neck Squamous Cell Carcinomas. Arch Otolaryngol Head Neck Surg. 2009;135(11):1158–1164. doi:10.1001/archoto.2009.161
To compare toxic effects and functional outcomes of reirradiation with and without salvage surgery for nonnasopharyngeal mucosal head and neck squamous cell carcinoma.
Academic tertiary referral hospital.
Between December 1992 and March 2007, a total of 87 patients underwent reirradiation (64 for cure and 23 for palliation).
Patients underwent reirradiation with (n = 38) or without salvage surgery (n = 49). After January 2000 there was increased use of concurrent platinum-based chemotherapy (80% vs 5%) and intensity-modulated radiation therapy (82% vs 0%).
Main Outcome Measures
Early and late toxic effects of treatment by Radiation Therapy Oncology Group criteria, tracheostomy retention, gastrostomy tube dependence, and survival.
The median follow-up among patients alive at last contact was 5.0 years. Compared with reirradiation without surgery, postoperative reirradiation was associated with increased early grade 3 to grade 5 toxic effects (50% [19 of 38] vs 29% [14 of 49], P = .04) and with longer median survival (17.3 vs 8.9 months, P < .001). Free-flap reconstruction decreased early toxic effects in the surgical cohort by 16% (from 60% [9 of 15] to 43% [10 of 23], P = .32). Gastrostomy tube dependence (P = .05) and tracheostomy retention (P = .04) have increased since 2000. The median survival for curative patients was 12.5 months. The estimated 2-year survival was 25%, and the estimated 5-year survival was 8%.
Reirradiation represents the only chance for cure in patients with unresectable disease. After surgery, reirradiation is performed in patients at high risk of locoregional recurrence and may increase acute toxic effects. However, free-flap reconstruction may reduce toxic effects. Functional outcomes have declined since 2000 likely because of the addition of concurrent platinum-based chemotherapy. Future research may define the subpopulation of postoperative patients for whom survival benefits most outweigh reirradiation toxic effects.
Treatment of mucosal head and neck squamous cell carcinoma (HNSCC) within an irradiated field remains a significant clinical challenge. Wherever possible, salvage surgery is preferred and has demonstrated disease-free survival beyond 5 years in 39% but with significant morbidity.1 For those at high risk of recurrence following salvage surgery or with unresectable disease, reirradiation may offer improved disease-free survival.2
However, reirradiation has significant toxic effects and influence on functional outcomes, and these may be exacerbated following salvage surgery. In one of the largest single-institution experiences, Salama et al from the University of Chicago Pritzker School of Medicine, Chicago, Illinois, reviewed 115 cases of reirradiation and concluded that “owing to the toxicity associated . . . treatment of these patients should be limited to investigational protocols.”3(p382) Among surgical candidates, French and Belgian centers randomized 130 patients undergoing salvage surgery for poor-prognosis HNSCC to reirradiation or observation and demonstrated a significant improvement in disease-free survival, with an associated increase in toxic effects.4 To mitigate this, free-flap reconstruction has been proposed to reduce the synergistic toxic effects of salvage surgery and reirradiation and to improve functional outcomes.5
In patients with unresectable disease, investigators have used evolving treatment paradigms in the anticipation of improving cure while limiting toxic effects and optimizing functional outcomes. Radiation Therapy Oncology Group (RTOG) 9610 reported on 79 unresectable HNSCCs and found that concurrent fluorouracil and hydroxyurea with reirradiation is feasible, with acceptable acute and late effects.6 However, the median survival was 8.5 months, and 4% survived to 5 years. Preliminary data from RTOG 9911, which used concurrent cisplatin and paclitaxel with reirradiation for unresectable cases, indicate an improved median survival of 12.1 months, at the expense of increased toxic effects.7 Advanced radiation therapy techniques such as intensity-modulated radiation therapy (IMRT) may counterbalance the effects of more toxic chemotherapy.8
The main objective of this study was to present modern outcomes of reirradiation in postoperative and unresectable settings in a clinical tertiary referral setting. Furthermore, the study tested the hypotheses that salvage surgery before reirradiation may increase toxic effects and worsen functional outcomes for patients with HNSCC and that the use of free-flap reconstruction may ameliorate these effects.
Institutional review board approval was obtained. The following patients were identified from the databases in the Department of Radiation Oncology, University of Alabama at Birmingham: all patients undergoing reirradiation for histologically proved recurrent or second primary HNSCC in an area previously irradiated to greater than 45 Gy between December 1992 and March 2007 without known distant metastases. Nonsquamous cell carcinoma (n = 8) and nasopharyngeal (n = 5) and skin (n = 6) primary cancers were excluded.
Patients without contraindications received salvage surgery first if complete macroscopic resection was deemed possible. Free-flap reconstruction was used if the resultant defect could not be optimally closed primarily. Reirradiation was generally recommended for high-functioning patients with poor prognostic features (eg, involved pathologic margins and extracapsular nodal disease), longer than 6 months' disease-free interval between courses of radiation therapy, and no severe late sequelae from the first course of radiation therapy. Most patients were treated in institutional or multi-institutional phase I or II clinical trials that used split-course altered fractionation radiation therapy with concurrent chemotherapy. Since January 2000, there has been an increase in the use of concurrent platinum-based chemotherapy (80% vs 5% before 2000) and IMRT (82% vs 0% before 2000).
Abstracted via retrospective medical record review were demographic characteristics, initial and secondary tumor characteristics and treatment, surgical and pathologic records, treatment and disease-related toxic effects, and gastrostomy (G-tube) and tracheostomy requirement. Neck-only and dermal metastases were coded as recurrent tumors. Patients receiving a total laryngectomy (n = 26) were excluded from the tracheostomy analysis. Toxic effects were scored according to RTOG and European Organization for Research and Treatment of Cancer criteria.9 Reirradiation toxic effects were coded as early if occurring within 3 months of completion of treatment and as late if occurring after 3 months. Five patients who died within 90 days of treatment were excluded from the late toxic effects analysis. In the event of uncertainty about adverse effects vs disease effects, toxic effects were attributed to disease progression. Survival was estimated using the Kaplan-Meier method from the start of reirradiation to the date of death or was censored at the last follow-up.
Statistical analyses were performed. Commercially available software (SPSS version 13.0; SPSS Inc, Chicago, Illinois) was used.
Between December 1992 and March 2007 a total of 87 patients underwent reirradiation for biopsy-proved primary site recurrence (50 [58%]), second primary (23 [26%]), neck-only recurrence (12 [14%]), or dermal metastatic HNSCC (2 [2%]). Dermal metastases are not considered an indication for reirradiation, but the clinical presentation in both patients was neck recurrence in a previously dissected field. There were 73 men (84%) and 14 women (16%). At the time of analysis, 82 patients (94%) were deceased, and 5 patients (6%) were alive, with a median follow-up of 5.0 years among survivors. The mean age at second presentation was 61.2 years (age range, 39.5-88.1 years). Thirty-eight patients (44%) received salvage surgery before reirradiation, and 49 patients (56%) received reirradiation alone because of unresectable disease or contraindications to surgery (Table 1).
Surgical salvage patients (50% [19 of 38]) were more likely to experience grade 3 to grade 5 early toxic effects than patients receiving reirradiation alone (29% [14 of 49], P = .04) (Table 2). Early grade 5 toxic effects occurred in 5 patients (6%): 2 patients had septic neutropenia, 2 patients experienced strokes during treatment, and 1 patient died following an episode of aspiration pneumonia.
Seventy-three patients (84%) completed more than 58 Gy of their intended reirradiation course. The remaining 14 patients (16%) stopped reirradiation treatment early because of toxic effects (11 [13%]) or disease progression (3 [3%]) and included 6 surgical patients (7%) and 8 nonsurgical patients (9%). Patients who stopped reirradiation early or received a cumulative dose of 120 Gy or less were more likely to have experienced grade 3 to grade 5 toxic effects. Early toxic effects were similar regardless of palliative or curative intent and before and after 2000 (when platinum-based chemotherapy and IMRT were increasingly used).
Twenty-six patients (30%) had free-flap reconstruction before reirradiation, 23 as part of surgical salvage for the reirradiated tumor and 3 after surgery following the first course of radiation therapy. Free-flap reconstruction decreased early toxic effects by 16% (from 60% [9 of 15] to 43% [10 of 23]) compared with surgery without free-flap reconstruction, although the numbers were too low for statistical significance (P = .32).
Carotid rupture occurred in 5 patients (6%) at a median of 7.2 months after treatment, resulting in 1 late death (1%), 1 major stroke (1%), and 3 ligations (3%) without significant neurologic effects. Three carotid ruptures (1 with free-flap reconstruction and 2 without) followed surgery and reirradiation. Two carotid ruptures followed reirradiation without salvage surgery, although both patients had undergone neck dissections as part of earlier cancer treatment.
There was a trend to increased late toxic effects following postoperative reirradiation compared with reirradiation alone (27% [12 of 44] vs 45% [17 of 38], P = .10) (Table 2). Free-flap reconstruction did not reduce the incidence of late toxic effects and was associated with a higher rate than surgery without free-flap reconstruction (33% [5 of 15] vs 52% [12 of 23], P = .25). Reirradiation doses exceeding 58 Gy were associated with a higher rate of late toxic effects (25% [9 of 36] vs 43% [20 of 46], P = .08). Patients younger than 60 years had a higher rate of late toxic effects (20% [9 of 44] vs 53% [20 of 38], P = .002), which may have been related to more platinum chemotherapy and higher reirradiation doses or to longer survival in this group.
Overall, 61 patients (70%) were G-tube dependent at the last follow-up. Neither salvage surgery nor free-flap reconstruction significantly affected the rate of G-tube requirement (Table 3). Overall, there was increased G-tube dependence after 2000 (80% [35 of 44] vs 60% [26 of 43], P = .05). In small numbers, neck-only recurrence was associated with a lower rate of G-tube dependence than recurrences at the primary site (76% [32 of 42] for oral or oropharynx and 74% [20 of 27] for larynx or hypopharynx vs 50% [9 of 18], P = .11).
Twenty-six patients (30%) underwent laryngectomy, reflecting 42% of the postoperative reirradiation group and 20% of the unresectable reirradiation group. Overall, 20 of the 61 patients (33%) required a tracheostomy at the last follow-up. Patients having salvage surgery were more likely to be without a tracheostomy at the last follow-up (9% [2 of 22] vs 46% [18 of 39], P = .003) (Table 3). Free-flap reconstruction was associated with a lower final rate of tracheostomy (38% [18 of 48] vs 15% [2 of 13], P = .13). Tracheostomy retention has increased since 2000 (45% vs 20%, P = .04) with greater use of platinum-based chemotherapy.
The median overall survival (OS) from the start of treatment for patients treated with curative intent (n = 64) was 12.5 months (Table 4). For patients treated with curative intent, the estimated 2-year OS was 25%, and the estimated 5-year OS was 8%. The median survival for patients treated with palliative intent (n = 23) was 9.5 months, with an estimated 2-year OS of 4% and an estimated 5-year OS of 0%.
Patients receiving surgery and postoperative reirradiation had longer median survival than the unresectable group (17.3 vs 8.9 months, P < .001), although few survived past 2 years (Figure). Patients receiving a reirradiation dose exceeding 58 Gy and a cumulative dose exceeding 120 Gy had longer median survival. In addition, there was a trend to improved survival among patients younger than 60 years (13.6 vs 10.4 months, P = .10) (Table 4).
At the last follow-up (range, 3.7-8.0 years after reirradiation), 4 patients (5%) were alive without disease, and 1 patient (1%) was alive with disease 24 months after reirradiation. Overall, 3 patients were disease free more than 5 years after reirradiation (Table 5).
Reirradiation is primarily considered for HNSCC as the only potential for cure in patients with unresectable disease or after surgery when the chance of locoregional recurrence is high.2 Similar to recent prospective trial data,4,6,7 our clinical institutional experience with reirradiation showed a guarded prognosis (8% 5-year survival), with significant early (38%) and late (35%) toxic effects and poor functional outcomes (70% G-tube dependence). Our data support the hypothesis that patients with resectable disease have a better prognosis (median survival, 17.3 vs 8.9 months; P < .001) but with increased acute toxic effects following reirradiation (P = .04). Free-flap reconstruction seemed to reduce acute reirradiation toxic effects (P = .32) but did not ameliorate late toxic effects. Based on a limited data set, long-term survivors following reirradiation are most likely to be young and have a long disease-free interval and second primary rather than recurrent tumors.
This study supports the hypothesis that reirradiation in the postoperative period is associated with increased early toxic effects compared with reirradiation alone (P = .04). Other series have not directly compared these 2 distinct groups. Janot et al4 found increased toxic effects with reirradiation compared with observation following surgery. However, there were fewer toxic effects in their series than in the present series (28% early toxic effects and 11% late toxic effects). This may be because of the use of fluorouracil-based chemo-reirradiation rather than platinum-based chemotherapy and frequent free-flap reconstruction. The RTOG 9610 noted rates of grade 3 to grade 5 acute toxic effects (63.3%) similar to ours with fluorouracil-based chemo-reirradiation without surgery and increased late effects (22.4%).6 Since 2000, our institution has shifted to the use of platinum-based chemotherapy, which may account for incidences of grade 3 to grade 5 acute toxic effects similar to those reported by the RTOG 9911 (77.8% early toxic effects and 37.4% late toxic effects).7
With cure as the primary goal, most clinicians have moved toward platinum-based chemotherapy, despite increased toxic effects.7 The IMRT modality is also increasingly used with the objectives of limiting grade 4 or grade 5 toxic effects to 20% and improving functional outcomes.8,10 Our study found an increase in G-tube dependence (P = .05) and tracheostomy retention (P = .04) since 2000. This may be because of increased use of platinum-based chemotherapy or a change in institutional practice because of perceived swallowing outcomes in this patient population. Our results do not support the findings by Goldstein et al11 of a higher rate of grade 3 or grade 4 toxic effects among patients treated with curative intent than among patients treated with palliative intent (57.1% vs 23.1%). Although the toxic effects profile may change with more aggressive chemotherapy and IMRT, the incidence remains high and may continue to increase.
Our data support the hypothesis that free-flap reconstruction may reduce the synergistic toxic effects of surgery and reirradiation (60% [9 of 15] vs 43% [10 of 23], P = .32). Although free-flap reconstruction is preferred at our institution, it is likely that the introduction of previously unirradiated vascularized tissue obtained from locoregional sites will similarly reduce early toxic effects. Suh et al5 proposed that free-flap reconstruction before reirradiation may prevent carotid rupture. Carotid rupture was seen more frequently in the absence of free-flap reconstruction; despite free-flap reconstruction, it occurred in 1 case. Carotid rupture in our series (1 death among 5 ruptures) was not associated with as high a mortality rate as noted in other reports. Other investigators report a similar rate of carotid rupture (5%) but with higher mortality (83%).3 Rupture may occur when the carotid experiences prolonged exposure to air or saliva. Withrow et al12 showed that vascularized tissue will improve fistula healing in an irradiated field. Overall, the beneficial healing effects of unirradiated vascularized tissue seem to outweigh the potential detrimental effect of increasing the volume of treatment for reirradiation. Consequently, this series supports the use of free-flap reconstruction in patients considered for postoperative reirradiation.
Reirradiation in the postoperative setting has significant effects on swallowing (70% [61 of 87] required G-tube feeding in our series) and on airway functioning (33% [20 of 61] required tracheostomy). Tracheostomy retention was more likely in the unresectable group (P = .003). This finding partly reflects that tracheostomy is preferred to laryngectomy in the case of a nonfunctioning larynx with unresectable disease and partly reflects a high rate of laryngeal dysfunction after reirradiation (46% [18 of 39] in our series). The rate of G-tube dependence in the present series is similar to other findings.3 However, Janot et al4 found that the rate dropped for survivors after 2 years (2-year G-tube rate, 28%). Persistent need for G-tube feeding has been associated with patient age, hypopharyngeal and tongue base sites, the use of adjuvant chemotherapy, and altered fractionation.13 The IMRT modality has been shown to improve posttreatment dysphagia,10 although our data do not support this. It may be that the increased use of adjuvant platinum-based chemotherapy has overshadowed the beneficial effects of IMRT.
Postoperative chemoradiotherapy is associated with significant treatment-related mortality (7% in our series). This is consistent with another series that found early deaths to be caused by chemotherapy-induced septic neutropenia and late mortality to have a greater relationship to surgery and reirradiation effects (edema, mucosal necrosis, and carotid rupture).4 Mortality rates among patients with unresectable disease are similar whether fluorouracil-based chemoradiotherapy (7.6% mortality in RTOG 96106) or platinum-based chemoradiotherapy (8% mortality in RTOG 99117) is used.
Given the guarded prognosis and significant toxic effects, this series supports the selective use of reirradiation in younger patients (<60 years [P = .10]) with a long disease-free interval, ideally more than 5 years after finishing their original radiation therapy. Recent evidence has correlated survival following reirradiation with surgical salvage, comorbid medical conditions, recurrent tumor stage and bulk, and reirradiation dose.14 Younger patients with few late sequelae of their first treatment are more likely to tolerate the prescribed course of reirradiation (survival increased if >58 Gy was achieved [P = .02]). Postoperative candidates for reirradiation should be counseled about significant toxic effects, poor functional outcomes, and the likely benefit to local control, with limited survival advantage. Given the high rates of early toxic effects and mortality, we advocate that reirradiation should be performed at centers that can manage difficult wound issues and catastrophic late events, including carotid rupture. At our institution, reirradiation has a limited role in the palliative setting.
To our knowledge, this is the first study to compare toxic effects and functional outcomes when reirradiation is used in the postoperative setting. Increased toxic treatment may be acceptable in postoperative patients because they have a better prognosis (median survival, 17.3 vs 8.9 months; P < .001), but this must be balanced by improvements in tumor control or survival. The best prospective trial of postoperative reirradiation to date, by Janot et al,4 demonstrates the difficulty in translating improved local control to survival. This may be because of the high rate of distant metastases in these patients (≤40% in some reirradiation series) or the eventual use of chemoradiotherapy in the observation arm (≤25%).8 Our study highlights the need for further research to select patients who may benefit from postoperative reirradiation.
Reirradiation is considered for HNSCC in the following 2 settings: (1) when it represents the only proven chance for cure in patients with unresectable disease and (2) after surgery when the chance of local recurrence is high (involved surgical margins or extracapsular spread in lymph nodes).This series found that postoperative reirradiation is associated with higher acute toxic effects but a better prognosis than reirradiation alone. Free-flap reconstruction may reduce acute reirradiation toxic effects. Since 2000, there has been an increase in G-tube and tracheostomy retention likely because of the increased use of platinum-based chemotherapy and despite the use of IMRT. Further research is required to define the subpopulation of postoperative patients whose survival benefits most from reirradiation.
Correspondence: William R. Carroll, MD, Division of Otolaryngology–Head and Neck Surgery, Department of Surgery, University of Alabama at Birmingham, BDB 563, 1530 Third Ave S, Birmingham, AL 35294-0012 (William.email@example.com).
Submitted for Publication: April 21, 2009; final revision received May 31, 2009; accepted June 10, 2009.
Author Contributions: Drs T. A. Iseli, Rosenthal, and Magnuson had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: T. A. Iseli, Rosenthal, Magnuson, and Carroll. Acquisition of data: T. A. Iseli, Caudell, and Smith. Analysis and interpretation of data: T. A. Iseli, C. E. Iseli, Caudell, Spencer, and Carroll. Drafting of the manuscript: T. A. Iseli, C. E. Iseli, and Rosenthal. Critical revision of the manuscript for important intellectual content: C. E. Iseli, Rosenthal, Caudell, Spencer, Magnuson, Smith, and Carroll. Statistical analysis: C. E. Iseli and Caudell. Administrative, technical, and material support: T. A. Iseli and Rosenthal. Study supervision: T. A. Iseli, Rosenthal, Magnuson, and Carroll.
Financial Disclosure: None reported.
Previous Presentation: This study was presented as an abstract at the American Head and Neck Society 2009 Annual Meeting; May 31, 2009; Phoenix, Arizona.