- Email: [email protected]
Predictive factors of local-regional recurrences following parotid sparing intensity modulated or 3D conformal radiotherapy for head and neck cancer
Radiotherapy and Oncology 77 (2005) 32–38 www.thegreenjournal.com
Head and neck radiotherapy
Predictive factors of local-regional recurrences following parotid sparing intensity modulated or 3D conformal radiotherapy for head and neck cancer Mary Fenga, Siavash Jabbaria, Alexander Lina, Carol R. Bradfordb, Douglas B. Chepehab, Theodoros N. Teknosb, Francis P. Wordenc, Christina Tsiena, Matthew J. Schipperd, Gregory T. Wolfb, Laura A. Dawsona,1, Avraham Eisbrucha,* a
Department of Radiation Oncology, bDepartment of Otolaryngology-Head and Neck Surgery, Department of Medicine-Hematology/Oncology and dDepartment of Biostatistics, University of Michigan, Ann Arbor, MI, USA
c
Abstract Background and purpose: Predictive factors for local–regional (LR) failures after parotid-sparing, Intensity modulated (IMRT) or 3D conformal radiotherapy for head and neck (HN) cancers were assessed. Patients and methods: One hundred and fifty-eight patients with mostly stages III–IV HN squamous cell carcinoma underwent curative bilateral neck irradiation aimed at sparing the parotid glands. Patient, tumor, and treatment factors were analyzed as predictive factors for LR failure. Results: Twenty-three patients had LR recurrence (19 in-field and four marginal). No differences were found in the doses delivered to the PTVs of patients with or without in-field recurrences. In univariate analysis, tumor site was highly predictive for LR failure in both postoperative and definitive RT patients. In postoperative RT patients, pathologic tumor size, margin status, extracapsular extension (ECE) and number of lymph node metastases, were also significantly predictive. Multivariate analysis showed tumor site (oropharynx vs. other sites) to be a significant predictor in all patients, and involved margins and number of involved lymph nodes in postoperative patients. Conclusions: Clinical rather than dosimetric factors predicted for LR failures in this series, and were similar to those reported following standard RT. These factors may aid in the selection of patients for studies of treatment intensification using IMRT. q 2005 Elsevier Ireland Ltd. All rights reserved. Radiotherapy and Oncology 77 (2005) 32–38. Keywords: Head and neck cancer; IMRT; Predictive factors
Risk factors for local–regional (LR) recurrence have been well described in the literature for head and neck cancers treated with standard radiation (RT) alone, surgery alone, and multi-modality therapy. These include patient factors including age, performance status, and pre-treatment hemoglobin, tumor factors including stage, size, site, positive margins, lymphvascular invasion (LVI), and extracapsular extension (ECE) [36], and treatment factors including chemotherapy and altered fractionation [3,9,36]. Parotid-sparing intensity modulated RT (IMRT) or 3D RT aim to spare critical noninvolved organs by producing dose distributions that conform highly to the targets. With the 1
Current address: Department of Radiation Oncology, Princess Margaret Hospital, Toronto, Canada.
increasing use of IMRT, it is important to specifically investigate which prognostic factors apply to this treatment technique. The differences in the volume of irradiated tissue may cause new factors to gain prominence. We have assessed the impact of patient, tumor, and treatment factors on LR recurrence for nonnasopharyngeal head and neck cancers following IMRT.
Material and methods Patients and tumor characteristics Between February 1994 and August 2003, 160 patients with nonnasopharyngeal squamous cell carcinoma of mucosal origin in the head and neck received parotid-sparing, bilateral neck radiation at the Department of Radiation
0167-8140/$ - see front matter q 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.radonc.2005.07.008
M. Feng et al. / Radiotherapy and Oncology 77 (2005) 32–38
Oncology at the University of Michigan. Treatment was delivered with curative intent and was secondarily aimed at sparing the parotid glands. No patients had previous RT. Two of these patients did not complete their therapy due to medical problems and are excluded from analysis. Details of patient and tumor characteristics are provided in Table 1. Seventy-two patients received primary RT, while 86 received postoperative RT. The majority (94 patients) had oropharyngeal cancer, approximately equally divided between tonsil (51 patients) and base of tongue (43 patients) primary cancers. The second largest group had oral cavity cancer (36 patients), where the most common (18 of 36) was tongue cancer. The large majority (136 patients, 88%) had clinical or pathologic evidence of lymph node (LN) metastasis in the side of the neck ipsilateral to the primary tumor. Twenty-two patients did not have evidence of
Table 1 Patient and tumor characteristics for all patients. Percentages of the total number of patients in each category are listed Age Tumor site Oral cavity Oropharynx Larynx Hypopharynx Unknown primary Tumor stage (AJCC) I II III IVA IVB Recurrent Tumor histology Well-differentiated Moderately differentiated Poorly differentiated Other Not available Chemotherapy Hemoglobin Post-operative radiation Number of patients Size of primary tumor specimen (cm) Patients with positive margins Number lymph nodes containing metastasis Number of lymph node levels containing metastasis Percent of total lymph nodes containing metastasis ECE (C) ECE (K) Time from surgery to initiation of IMRT (days) Definitive radiation Number of patients GTV (cc) Largest neck node volume (cc)
57 (24-86) 36 (23%) 94 (59%) 12 (8%) 13 (8%) 3 (2%) 2 (1%) 9 (6%) 35 (22%) 93 (59%) 14 (9%) 5 (3%) 14 (9%) 70 (44%) 44 (28%) 8 (5%) 22 (14%) 66 (42%) 12.3 (8.6–15.9) 86 (54%) 2.7 (0–7) 7 (4%) 2 (0–10) 2 (0–4) 12 (0–100) 40 46 41 (11–83)
72 (46%) 43.7 (6–199) 13.7 (0.5–120)
For continuous variables, the median and range are listed.
33
LN metastasis, but they had oral cavity or oropharyngeal cancer judged to place them at risk for bilateral neck metastasis. Patients treated with surgery had resection of the primary tumor and modified radical dissection of the ipsilateral neck. No patient underwent dissection of the contralateral neck. Pathological evaluation of tumor margins and lymph node involvement included histochemistry using epithelial mucin stains and immunohistochemistry with cytokeratin. Almost all patients with oral cavity cancer received postoperative RT (35 of 37). In contrast, a higher number of patients with oropharyngeal cancer received primary compared with postoperative RT (56 and 40 patients, respectively). The majority of the patients with laryngeal (9 of 13) and hypopharyngeal (9 of 13) cancer received primary RT. In addition, 3 patients with unknown primary cancer received postoperative RT following neck dissection. For the analysis of predictive factors, factors specific for primary RT (GTV size) or for postoperative RT (pathologic tumor size, margins status, ECE, pathologically involved nodes) were analyzed separately for each of the patient groups (see statistical analysis).
Target definitions Our policies for the selection of the neck levels for each tumor site and stage, and for the delineation of the targets throughout the neck have been detailed elsewhere [14,15]. The targets include the gross tumor volumes (GTVs) and the clinical target volumes (CTVs) consisting of subclinical disease in the vicinity of the primary tumors, and neck nodal volumes at risk for subclinical disease. The delineation of the GTV was made according to both radiologic and clinical data: in addition to the radiological abnormalities, we relied on the results of indirect fiberoptic endoscopy in the clinic, and on the reports of direct endoscopy under anesthesia performed in all patients by a head and neck surgeon before RT planning. CTV delineation rules have been similar to those described by others [7,20,32,39]; the main distinction of our policies have been in the upper neck targets: on the side of the neck ipsilateral to the primary tumor, we defined level II CTV through the base of the skull. In the contralateral, grossly noninvolved, neck side, the most superior level II CTV was delineated at the axial planning CT scan in which the posterior belly of the digastric muscle was bisected by the jugular vein, ensuring that the subdigastric node was encompassed within the CTV, but excluding the junctional node from the CTV [14,15]. This was made in compliance with the anatomic observations of Rouviere, who described the subdigastric nodes as the upper-most node draining lymphatics from the nonnasopharyngeal upper aerodigestive tract [35]. Due to concerns about retrograde lymphatic flow in the ipsilateral neck, which had evidence of metastases in the large majority of cases, we have practiced a more conservative approach to the ipsilateral neck and outlined level II CTV further cranially to the base of skull. Initially, the retropharyngeal nodes were contoured through the top of C1 based on these observations. However, following two recurrences superior to the top of C1 and posterior to the upper nasopharynx, the retropharyngeal nodal CTVs were modified in 1999 to include the base of the skull in front of the clivus.
34
Recurrence predictors after HN IMRT
Therapy From 1994 until early 1996, parotid-sparing RT was delivered to 28 patients with conformal 3D techniques [17], and from 1996 through 2003, static multisegmental IMRT was delivered to 130 patients according to previously published methods [16,18]. All targets, including the CTVs in the low neck, were encompassed in the conformal/IMRT plans. The GTVs and the CTVs were expanded uniformly by 0.5 cm to yield their corresponding planning target volumes (PTVs). The median prescribed doses to the PTVs of the gross tumor, subclinical disease, and post-operative surgical beds, were 70.4 Gy (range: 66–76 Gy), 50.4 Gy (46–54 Gy), and 61.2 Gy (57.6–64 Gy), respectively. PTVs of postoperative tumor beds, where close margins or ECE were noted received prescribed doses of 64 Gy, and where these features were absent, they were prescribed 57.6–60 Gy. Daily fractions were 1.8–2.0 Gy. Cisplatin-based concurrent chemotherapy was delivered to 66 patients, the majority of whom (56 patients) were treated with primary radiation. In recent years, all patients who presented with neck lymphadenopathy measuring over 3 cm and who received primary chemo-RT underwent planned selective neck dissection 3 months after the completion of RT, according to an institutional protocol. The use of primary chemoradiation instead of surgery and postoperative RT became more prevalent in the latter years of the study as institutional policies shifted toward organ preservation approaches [13,37]. Clinical follow-up of all patients was performed by head and neck oncologic surgeons and radiation oncologists every 6–8 weeks during the first two years, every 3 months in the third year, and every 6 months, thereafter. Radiologic evaluation of the head and neck was performed at the time of clinical suspicion of recurrence.
patients, these factors included details from the pathological reports (largest dimension of primary tumor, number of positive lymph nodes, number of involved lymph node levels, percent positive lymph nodes, ECE, and findings of positive surgical margins). In the primary RT patients, the primary tumor size, represented by the volume of the GTV, and the volume of the largest LN, were derived from the treatment planning CT scans. Actuarial LR recurrence-free, disease-free, and overall survival curves were estimated using the Kaplan–Meier method. Patients failing in distant sites were not censored for LR recurrence-free survival. Log-rank tests were used to test for marginal differences in survival among the levels of the categorical predictors. Likelihood ratio tests from a proportional hazards model were used to test for significance of continuous predictors. Stepwise regression was used in a proportional hazards model to obtain a joint model for LR recurrence free survival. Individual significance testing in this joint model was done using Wald chi-squared statistics. All analysis was performed using the SAS system (SAS Institute, Cary, NC). Statistical significance was determined as P!0.05.
Results The median follow-up was 36.4 months, range 6–127 months. Of 117 patients still alive, 71 had at least 3 years of follow up. Twenty-three LR recurrences (15%) were observed at a median post-therapy interval of 13.3 months. The 2-year actuarial LR recurrence-free survival (LRRFS) was 85%. The 5-year LRRFS was 83%. The actuarial 2- and 5-year overall survival were 84 and 73%, respectively (Fig. 1).
Evaluation of tumor recurrence vs. dose Our methods to characterize recurrent tumors as in-field, marginal, or out-of-field were detailed elsewhere [11]. In brief, each patient’s recurrent tumor volume (Vr) was delineated on the CT or MRI scan at the time of recurrence, transferred to the planning CT scan via image registration, and a dose-volume-histogram (DVH) of Vr was calculated. The recurrence was defined as in-field if O95% of Vr had received the prescribed dose. Marginal and out-of-field recurrences were defined, where Vr had received between 25 and 95% and less than 25% of the prescribed dose, respectively. In cases of LR recurrences, the doses delivered to the rest of the PTV in which a recurrence occurred was assessed by subtracting the Vr from the PTV and calculating a DVH for the subtracted volume. The doses delivered to 1% of the volumes receiving the lowest or the highest dose in each target were reported as the minimal or the maximal target doses, respectively (rather than the doses delivered to single voxels receiving the lowest or highest dose).
Statistical analysis All the factors detailed in Table 1 were tested as potential clinical predictors of LR recurrences. Salvage of recurrences was not taken into account in the evaluation of recurrencefree survival or its predictors. In the postoperative RT
Fig. 1. Local–regional recurrence-free survival, Disease-free Survival, and Overall 1survival
M. Feng et al. / Radiotherapy and Oncology 77 (2005) 32–38
Patient and tumor characteristics in cases of LR recurrences Details of patients and tumors in cases, where LR recurrences were noted, and in cases without LR recurrences, are provided in Table 2. No AJCC stage I patients had LR recurrence. However, 1 of 9 (11%) stage II, 5 of 35 (14%) stage III, 15 of 93 (16%) stage IVA, 1 of 14 (7%) stage IVB, and 1 of 5 (20%) recurrent patients experienced LR recurrence. Patients with oral cavity cancer had the highest number of LR recurrences: they accounted for 48% of all recurrences although they only comprised 23% of all tumors. By site, 11 of 36 (31%) oral cavity, 6 of 94 (6%) oropharynx, 4 of 12 (33%) larynx, 2 of 13 (15%) hypopharynx, and 0 of 3 (0%) unknown
35
primary developed LR recurrence. Three of 7 (43%) patients with positive margins, 10 of 79 (13%) with negative margins, 10 of 39 (25%) with ECE and 4 of 46 (9%) without ECE developed LR recurrence. Postoperative patients with recurrence had generally larger primary tumors than those without recurrence (median of 3.6 vs. 2.5 cm, respectively). They also had more positive lymph nodes (median 4 vs. 2, respectively).
Evaluation of patient, tumor, and treatment factors Using univariate analysis, tumor site was highly predictive for LR failure in both primary and postoperative patients (PZ0.0005). Patients with oropharyngeal primaries had
Table 2 Patient and tumor characteristics for patients with and without LR recurrence
Age Tumor site Oral cavity Oropharynx Larynx Hypopharynx Unknown Primary Tumor stage (AJCC) I II III IVA IVB Recurrent Tumor histology Well-differentiated Moderately differentiated Poorly differentiated Not available Chemotherapy Hemoglobin Post-operative radiation Number of patients Size of primary tumor specimen (cm) Patients with positive margins Number lymph nodes containing metastasis Number of lymph node levels containing metastasis Percent of total lymph nodes containing metastasis ECE (C) ECE (K) Time from surgery to initiation of IMRT (days) Definitive radiation Number of patients GTV (cc) Largest neck node volume (cc)
Patients with
Patients without
P-value
LR recurrence (23)
LR recurrence (135)
55.7 (24–77)
56.3 (29–86)
11 (31%) 6 (6%) 4 (33%) 2 (15%) 0 (0%)
25 (69%) 88 (94%) 8 (67%) 11 (85%) 3 (100%)
0 (0%) 1 (11%) 5 (14%) 15 (16%) 1 (7%) 1 (20%)
2 (100%) 8 (89%) 30 (86%) 78 (84%) 13 (93%) 4 (80%)
3 (21%) 13 (19%) 4 (9%) 3 (14%) 8 (12%) 11.9 (9–14.8)
11 (79%) 57 (81%) 40 (91%) 19 (86%) 58 (88%) 12.4 (8.6–15.9)
NS NS
14 (16%) 3.6 (1–7)
72 (84%) 2.5 (0–6)
0.02
3 (43%) 4 (0–10)
4 (57%) 2 (0–8)
0.04a 0.006a
2 (0–4)
1 (0–3)
NS
16 (1–47)
9 (0–100)
NS
10 (25%) 4 (9%) 39.3 (28–63)
30 (75%) 42 (91%) 41.9 (11–83)
0.01 NS
9 (13%) 28.9 (10–53) 14.3 (1–42)
63 (87%) 46.2 (6–199) 13.2 (1–120)
NS NS
NS 0.0005a
NS
NS
For continuous variables, the median and range are listed. a Statistically significant in multivariate analysis.
36
Recurrence predictors after HN IMRT
Dosimetric evaluation of LR recurrences
Fig. 2. Local/regional disease-free survival by site (the figure does not include 3 patients with unknown primary tumors).
a 2-year actuarial LR recurrence-free survival (LRRFS) of 94%, significantly higher than patients with primary tumors in other sites (Fig. 2). In postoperative RT patients, the pathologic tumor size (PZ0.024), margin status (PZ0.046), presence or absence of ECE (PZ0.01), and the number of positive lymph nodes (PZ0.006) were also significant predictive features (Table 2). There was no difference in disease free or local–regional recurrence free survival between patients receiving definitive or post-operative therapy. The joint multivariate model showed tumor site (oropharynx vs. other all sites) to be a significant predictor of LR recurrences in all patients, and the presence of positive margins, and number of positive lymph nodes as significant predictors for LR recurrence in postoperative RT patients. The hazard for LR recurrence, including the other significant factors in the model, was increased by a factor of 1.44 for each additional pathologically involved lymph node. Two of 16 (12%) patients with pathological stage N0 had LR failure compared with 12 failures of 70 (16%) patients with pathological NC; the difference was not statistically significant due to the small number of N0 patients. The presence of positive margins increased the hazard for LR recurrence by 9.16. Factors not found to be predictive were age, pre-treatment hemoglobin, AJCC stage, and treatment with chemotherapy in all patients; the number of positive lymph node levels, percent positive lymph nodes, and the time from surgery to the initiation of radiation therapy in postoperative patients. In patients receiving primary RT, the volumes of the GTVs or the largest lymph node in each patient, as defined on the treatment planning CT scans, were not found to predict LR failure. Separate evaluation of patients with oropharyngeal and with larynx/hypopharynx primary tumors receiving definitive RT failed to identify the volumes of the GTVs as a significant predictive factor of LR recurrence in either group.
Of the 23 LR recurrences, 19 were in-field and 4 were marginal. Each of the marginal recurrences could be explained by a specific clinical decision regarding target delineation, detailed elsewhere [15]. In brief, these cases included marginal recurrences in retropharyngeal nodes cranial to C1 (initially, the cranial edges of the CTVs for the retropharyngeal nodes were delineated at the top of C1 according to Rouviere’s observations), a recurrence at level VI following therapy of oral cavity cancer with significant involvement of levels III–IV, and a marginal recurrence at an unexpected site in a patient irradiated for recurrence a few years after neck dissection, allowing the growth of new lymphatics whose position is unpredictable. No recurrences were observed in the contralateral high neck, where the level II CTV excluded the junctional nodes. No differences were found in the minimal, maximal or mean doses delivered to the PTVs of patients with and without recurrences (in both groups the meanCSD of the minimal PTV dose was 95C4%, the mean PTV dose 100C2%, and the maximal PTV dose 106C3%, of the prescribed dose). Additionally, in each of the 19 in-field recurrences there were no differences in the minimal doses delivered to Vr compared with the minimal doses delivered to the rest of the respective PTVs.
Discussion In this series, primary tumor site was the most significant predictive factor for LR recurrence. Patients with oropharyngeal primary tumors fared significantly better than all other primary sites, whether treated with primary or with postoperative RT. The relatively high success rate of primary irradiation for advanced oropharyngeal cancer has long been noted, offering similar rates of LR control as surgery and postoperative RT, but with less associated morbidity and better quality of life [34]. Recently, altered fractionated RT and concomitant chemotherapy has resulted in further improvements of the LR control rates of oropharyngeal cancer [5,21]. The Washington University experience with IMRT for oropharyngeal cancer has recently been reported, with 87% 4-year LR control rate in 74 patients [6]. Patient selection factors, length of follow-up, and other factors including differences in diagnostic and therapeutic procedures that are not related to RT technology, preclude meaningful direct comparisons among these series and ours. In comparison with oropharyngeal cancer, patients with oral cavity cancer in our series, almost all of whom received postoperative RT, fared less well. Their LR control rates were identical to those reported by Zelefsky et al. following standard postoperative RT [42]. These results highlight the aggressive nature of oral cavity cancer in many cases requiring bilateral neck postoperative RT. Other significant factors in the joint model were positive resection margins and the number of involved lymph nodes. Positive resection margins were quite rare in our series. Several large retrospective series of standard postoperative RT have demonstrated the impact of involved margins on LR control [2,8,10,22,27]. The number of positive lymph nodes has also been shown to impact poorly on locoregional control [1,12,28,33]. In our study, the hazard increased by 1.44 for
M. Feng et al. / Radiotherapy and Oncology 77 (2005) 32–38
each positive node. In addition, if patients had extracapsular extension in their positive nodes, they were at higher risk for LR. This is also consistent with prior studies for conventional postoperative RT [1,12,19,23,30]. Recent results of randomized studies suggest that treatment intensification via the addition of concurrent chemotherapy may improve the results of postoperative RT in patients with these adverse prognostic factors [3,9]. CT-based tumor size has previously been reported to bear prognostic importance in primary irradiation of laryngeal [26] and oropharyngeal [6,31] cancer, following both standard RT [26,31] and IMRT [6]. Mendenhall et al. found that size was not an independent factor in multivariate analysis, but that after stratification by primary site, volume was independently significant for supraglottic and glottic cancers, but not for oropharyngeal ones [29]. In our series, the pathologic primary tumor size in the post-operative RT patients was predictive of LR recurrences. However, the volumes of the GTVs in definitively irradiated patients were not found to be predictive. Specifically, we did not find it to be predictive in the oropharyngeal cancer patients who constituted the largest patient population site-wise. In our practice, the delineated GTVs extended in many cases beyond the obvious radiologic abnormalities. These extensions of the GTVs were not detected radiologically but were judged clinically to contain gross disease according to the descriptions of tumor extent made during direct endoscopy under anesthesia. Thus, the GTVs in our studies differed from those derived solely from imaging. We do not know whether or not a delineation of the GTVs based strictly on the radiological abnormalities, as analyzed in other series cited above, would emerge in our series as a significant predictor of LR recurrences. Dosimetric factors were not significant predictors of infield recurrences, all of which occurred within the CTVs, mainly due to the strict dose homogeneity criteria used in this series. Such homogeneity can be achieved in both forward planned and inverse-planned IMRT [38]. We cannot address, therefore, the relative importance of cold spots within the targets or what level of inhomogeneity may be allowed. Another confounding factor is the uncertainties related to the exact origin of the recurrent tumor. We used the doses delivered to the epicenter of the recurrence to define whether it was in-field, marginal or out-of-field. However, the accuracy of this determination depended in part on the timing of the radiologic evaluation relative to the growth of the recurrent tumor. Two recent prospective randomized trials, RTOG 95-01 and EORTC 22931, found that the addition of concurrent Cisplatin to postoperative RT for patients with high risk features including positive margins improved local control and may impact on overall survival [3,9]. Few of the postoperative RT patients in our series were treated with concurrent chemoradiation, while almost all of our primary RT patients received concurrent therapy; therefore, our series did not have the power to detect a potential significance of chemotherapy in LR recurrences. This series is very heterogeneous regarding tumor sites and therapy. Larger patient numbers at each site and therapy (definitive and post operative RT) may unveil additional prognostic
37
factors for LR failure, which this series has not been powered to detect. In conclusion, the large majority of the LR recurrences in our series occurred in-field. Similar observations have recently been reported in other series of HN IMRT or parotid-sparing 3DCRT [4,7,24,25]. In our series, dosimetric factors could not differentiate between the patients who had in-field recurrences and those who did not have LR failure, a byproduct of the strict target dose homogeneity criteria employed [16]. Thus, clinical, rather than dosimetric factors, predicted the risk of LR recurrences in this series. These risk factors were similar to those reported in the past following standard RT. Whether higher doses to targets at high risk of failure may reduce recurrence rates is a hypothesis that needs to be tested. Dose escalation studies that rely on the ability of IMRT to exclude noninvolved tissue from the high dose volumes have been proposed [40,41]. The clinical factors that significantly predicted LR failure in our series can serve as eligibility criteria for such dose escalation trials, as patients lacking these features had a low risk of LR failure.
Acknowledgements Supported by NIH grants CA59827 and CA78165, and the Duke Family Head and Neck Cancer Research Fund.
*
Corresponding author. Dr. Avraham Eisbruch, Department of Radiation Oncology, University of Michigan Hospital, Ann Arbor MI 48109-0010. E-mail address: [email protected]
Received 4 March 2005; received in revised form 30 June 2005; accepted 7 July 2005; available online 8 September 2005
References [1] Alvi A, Johnson JT. Extracapsular spread in the clinically negative neck: implications and outcome. Otolaryngol Head Neck Surg 1996;114:65–70. [2] Bastit L, Blot E, Debourdeau P, et al. Influence of the delay of adjuvant postoperative radiation therapy on relapse and survival in oropharyngeal and hypopharyngeal cancers. Int J Radiat Oncol Biol Phys 2001;49:139–46. [3] Bernier J, Domenge C, Ozsahin M, et al. Postoperative irradiation with or without concomitant chemotherapy for locally advanced head and neck cancer. NEJM 2004;350: 1945–52. [4] Bussels B, Maes A, Hermans R, et al. Recurrences after conformal parotid-sparing radiotherapy for head and neck cancer. Radiother Oncol 2004;72:119–27. [5] Calais G, Alfonsi M, Bardet E, et al. Randomized trial of radiation therapy versus concomitant chemotherapy and radiation therapy for advanced-stage oropharynx carcinoma. J Natl Cancer Inst 1999;91:2081–6. [6] Chao KS, Ozyigit G, Blanco AI, et al. Intensity-modulated radiation therapy for oropharyngeal carcinoma: impact of tumor volume. Int J Radiat Oncol Biol Phys 2004;59:43–50. [7] Chao KSC, Wippold FJ, Ozygit G, et al. Determination and delineation of nodal target volumes for head-and -neck cancer based on patterns of failure in patients receiving definitive and postoperative IMRT. Int J Radiat Oncol Biol Phys 2002;53: 1174–84.
38
Recurrence predictors after HN IMRT
[8] Cook JA, Jones AS, Phillips DE, et al. Implications of tumour in resection margins following surgical treatment of squamous cell carcinoma of the head and neck. Clin Otolaryngol 1993;18: 37–41. [9] Cooper JS, Pajak TF, Forastiere AA, et al. Postoperative concurrent radiotherapy and chemotherapy for high-risk squamous-cell carcinoma of the head and neck. NEJM 2004; 350:1937–44. [10] Cooper JS, Pajak TF, Forastiere A, et al. Precisely defining highrisk operable head and neck tumors based on RTOG 85-03 and 88-24: targets for postoperative radiochemotherapy? Head Neck 1998;20:588–94. [11] Dawson LA, Anzai Y, Marsh L, et al. Patterns of loco-regional recurrence following parotid-sparing conformal and segmental intensity-modulated radiotherapy for head and neck cancer. Int J Radiat Oncol Biol Phys 2000;46:1117–26. [12] de Carvalho MB. Quantative analysis of the extent of extracapsular invasion and its prognostic significance: a prospective study of 117 cases of carcinoma of the larynx and hypopharynx. Head Neck 1998;20:16–21. [13] Department of Veterans Affairs Laryngeal Cancer Study Group. Induction chemotherapy plus radiation compared with surgery plus radiation in patients with advanced laryngeal cancer. NEJM 1991;324:1685–90. [14] Eisbruch A, Foote RL, O’Sullivan B, et al. IMRT for head and neck cancer: emphasis on the selection and delineation of the targets. Semin Radiat Oncol 2002;12:238–49. [15] Eisbruch A, Marsh LH, Dawson LA, et al. Recurrences near base of skull after IMRT for head-and-neck cancer: implications for target delineation in high neck and for parotid gland sparing. Int J Radiat Oncol Biol Phys 2004;59:28–42. [16] Eisbruch A, Marsh LH, Martel MK, et al. Comprehensive irradiation of head and neck cancer using conformal multisegmental fields: assessment of target coverage and noninvolved tissue sparing. Int J Radiat Oncol Biol Phys 1998;41: 559–68. [17] Eisbruch A, Ship JA, Martel MK, et al. Parotid gland sparing in patients undergoing bilateral head and neck irradiation: techniques and early results. Int J Radiat Oncol Biol Phys 1996;36:469–80. [18] Fraass BA, Kessler ML, McShan DL, et al. Optimization and clinical use of multisegment intensity-modulated radiation therapy for high-dose conformal therapy. Semin Radiat Oncol 1999;9:60–77. [19] Greenberg JS, Flowler R, Gomez J, et al. Extent of extracapsular spread: a critical prognosticator in oral tongue cancer. Cancer 2003;97:1464–70. [20] Gregoire V, Coche E, Cosnard G, et al. Selection and delineation of lymph node target volumes in head and neck conformal radiotherapy. Proposal for standardizing terminology and procedure based on the surgical experience. Radiother Oncol 2000;56:135–50. [21] Horiot JC, Le Fur R, N’Guyen T, et al. Hyperfractionation versus conventional fractionation in oropharyngeal carcinoma: final analysis of a randomized trial of the EORTC cooperative group of radiotherapy. Radiother Oncol 1992;25:231–41. [22] Jacobs JR, Ahmad K, Schuller CR. Implications of positive surgical margins. Laryngoscope 1993;103:64–8. [23] Jose J, Coatesworth AP, Johnston C, MacLennan K. Cervical node metastases in squamous cell carcinoma of the upper aerodigestive tract: the significance of extracapsular spread and soft tissue deposits. Head Neck 2003;25:451–6. [24] Lee N, Xia P, Fischbein NJ, et al. Intensity-modulated radiation therapy for head-and-neck cancer: the UCSF experience focusing on target volume delineation. Int J Radiat Oncol Biol Phys 2003;57:49–60.
[25] Levendag P, Braaksma M, Coche E, et al. Rotterdam and Brussels CT-based neck nodal delineation compared with the surgical levels as defined by the American academy of otolaryngology—head and neck surgery. Int J Radiat Oncol Biol Phys 2004;58:113–23. [26] Mancuso AA, Mukherji SK, Schmalfus I, et al. Preradiotherapy computed tomography as a predictor of local control in supraglottic carcinoma. J Clin Oncol 1999;17:631–7. [27] Mantravadi RV, Haas RE, Liebner EJ, et al. Postoperative radiotherapy for persistent tumor at the surgical margin in had and neck cancers. Laryngoscope 1983;93:1337–40. [28] Mantravadi RV, Skolnik EM, Haas RE, et al. Patterns of cancer recurrence in the postoperatively irradiated neck. Arch Otolaryngol 1983;109:753–6. [29] Mendenhall WM, Morris CG, Amdur RJ, et al. Parameters that predict local control after definitive radiotherapy for squamous cell carcinoma of the head and neck. Head Neck 2003;25: 535–42. [30] Myers JN, Greenberg JS, Vo V, Roberts D. Extracapsular spread. A significant predictor of treatment failure in patients with squamous cell carcinoma of the tongue. Cancer 2001;92: 3030–6. [31] Nathu R, Mancuso AA, Zhu TC, et al. The impact of primary tumor volume on local control for oropharyngeal squamous cell carcinoma treated with radiotherapy. Head Neck 2000;22:1–5. [32] Nowak PJCM, Wijers OB, Lagerwaard FJ, Levendag PC. A threedimensional CT-based target definition for elective irradiation of the neck. Int J Radiat Oncol Biol Phys 1999;45:33–9. [33] Olsen KD, Daruso M, Foote RL, et al. Primary head and neck cancer. Histopathologic predictors of recurrence after neck dissection in patients with lymph node involvement. Arch Otolaryngol Head Neck Surg 1994;120:1370–4. [34] Parsons JT, Mendenhall WM, Stringer SP, et al. Squamous cell carcinoma of the oropharynx. Surgery, radiation therapy, or both. Cancer 2002;94:2967–80. [35] Rouviere H. Lymphatic systems of the head and neck Translated by MJ tobias. Ann Arbor, MI: Edwards Brothers; 1938. [36] Smith BD, Haffty BG. Prognostic factors in patients with head and neck cancer. In: Harrison LB, Sessions RB, Hong WK, editors. Head and neck cancer: a multidisciplinary approach. 2nd ed. Philadelphia: Lippincott Williams & Wilkins; 2004. p. 49–73. [37] Urba SG, Forastiere AA, Wolf GT, et al. Intensive induction chemotherapy and radiation for organ preservation in patients with advanced resectable head and neck carcinoma. JCO 1994; 12:946–53. [38] Vineberg KA, Eisbruch A, Coselmon MM, et al. Is uniform target dose possible in IMRT plans in the head and neck? Int J Radiat Oncol Biol Phys 2002;52:1159–72. [39] Wijers OB, Levendag PC, Tan T, et al. A simplified CT-based definition of the lymph node levels in the node negative neck. Radiother Oncol 1999;52:35–42. [40] Wu Q, Manning M, Schmidt-Ullrich R, Mohan R. The potential for sparing of parotids and escalation of biologically equivalent dose with intensity modulated radiation treatments of head and neck cancers: A treatment design study. Int J Radiat Oncol Biol Phys 2000;46:195–205. [41] Wu Q, Mohan R, Morris M, et al. Simultaneous integrated boost intensity-modulated radiotherapy for locally advanced headand-neck squamous cell carcinomas. I: dosimetric results. Int J Radait Oncol Biol Phys 2003;56:573–85. [42] Zelefsky MJ, Harrison LB, Fass DE, et al. Postoperative radiotherapy for oral cavity cancers: impact of anatomic subsite on treatment outcome. Head Neck 1990;12:470–5.
Head and neck radiotherapy
Predictive factors of local-regional recurrences following parotid sparing intensity modulated or 3D conformal radiotherapy for head and neck cancer Mary Fenga, Siavash Jabbaria, Alexander Lina, Carol R. Bradfordb, Douglas B. Chepehab, Theodoros N. Teknosb, Francis P. Wordenc, Christina Tsiena, Matthew J. Schipperd, Gregory T. Wolfb, Laura A. Dawsona,1, Avraham Eisbrucha,* a
Department of Radiation Oncology, bDepartment of Otolaryngology-Head and Neck Surgery, Department of Medicine-Hematology/Oncology and dDepartment of Biostatistics, University of Michigan, Ann Arbor, MI, USA
c
Abstract Background and purpose: Predictive factors for local–regional (LR) failures after parotid-sparing, Intensity modulated (IMRT) or 3D conformal radiotherapy for head and neck (HN) cancers were assessed. Patients and methods: One hundred and fifty-eight patients with mostly stages III–IV HN squamous cell carcinoma underwent curative bilateral neck irradiation aimed at sparing the parotid glands. Patient, tumor, and treatment factors were analyzed as predictive factors for LR failure. Results: Twenty-three patients had LR recurrence (19 in-field and four marginal). No differences were found in the doses delivered to the PTVs of patients with or without in-field recurrences. In univariate analysis, tumor site was highly predictive for LR failure in both postoperative and definitive RT patients. In postoperative RT patients, pathologic tumor size, margin status, extracapsular extension (ECE) and number of lymph node metastases, were also significantly predictive. Multivariate analysis showed tumor site (oropharynx vs. other sites) to be a significant predictor in all patients, and involved margins and number of involved lymph nodes in postoperative patients. Conclusions: Clinical rather than dosimetric factors predicted for LR failures in this series, and were similar to those reported following standard RT. These factors may aid in the selection of patients for studies of treatment intensification using IMRT. q 2005 Elsevier Ireland Ltd. All rights reserved. Radiotherapy and Oncology 77 (2005) 32–38. Keywords: Head and neck cancer; IMRT; Predictive factors
Risk factors for local–regional (LR) recurrence have been well described in the literature for head and neck cancers treated with standard radiation (RT) alone, surgery alone, and multi-modality therapy. These include patient factors including age, performance status, and pre-treatment hemoglobin, tumor factors including stage, size, site, positive margins, lymphvascular invasion (LVI), and extracapsular extension (ECE) [36], and treatment factors including chemotherapy and altered fractionation [3,9,36]. Parotid-sparing intensity modulated RT (IMRT) or 3D RT aim to spare critical noninvolved organs by producing dose distributions that conform highly to the targets. With the 1
Current address: Department of Radiation Oncology, Princess Margaret Hospital, Toronto, Canada.
increasing use of IMRT, it is important to specifically investigate which prognostic factors apply to this treatment technique. The differences in the volume of irradiated tissue may cause new factors to gain prominence. We have assessed the impact of patient, tumor, and treatment factors on LR recurrence for nonnasopharyngeal head and neck cancers following IMRT.
Material and methods Patients and tumor characteristics Between February 1994 and August 2003, 160 patients with nonnasopharyngeal squamous cell carcinoma of mucosal origin in the head and neck received parotid-sparing, bilateral neck radiation at the Department of Radiation
0167-8140/$ - see front matter q 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.radonc.2005.07.008
M. Feng et al. / Radiotherapy and Oncology 77 (2005) 32–38
Oncology at the University of Michigan. Treatment was delivered with curative intent and was secondarily aimed at sparing the parotid glands. No patients had previous RT. Two of these patients did not complete their therapy due to medical problems and are excluded from analysis. Details of patient and tumor characteristics are provided in Table 1. Seventy-two patients received primary RT, while 86 received postoperative RT. The majority (94 patients) had oropharyngeal cancer, approximately equally divided between tonsil (51 patients) and base of tongue (43 patients) primary cancers. The second largest group had oral cavity cancer (36 patients), where the most common (18 of 36) was tongue cancer. The large majority (136 patients, 88%) had clinical or pathologic evidence of lymph node (LN) metastasis in the side of the neck ipsilateral to the primary tumor. Twenty-two patients did not have evidence of
Table 1 Patient and tumor characteristics for all patients. Percentages of the total number of patients in each category are listed Age Tumor site Oral cavity Oropharynx Larynx Hypopharynx Unknown primary Tumor stage (AJCC) I II III IVA IVB Recurrent Tumor histology Well-differentiated Moderately differentiated Poorly differentiated Other Not available Chemotherapy Hemoglobin Post-operative radiation Number of patients Size of primary tumor specimen (cm) Patients with positive margins Number lymph nodes containing metastasis Number of lymph node levels containing metastasis Percent of total lymph nodes containing metastasis ECE (C) ECE (K) Time from surgery to initiation of IMRT (days) Definitive radiation Number of patients GTV (cc) Largest neck node volume (cc)
57 (24-86) 36 (23%) 94 (59%) 12 (8%) 13 (8%) 3 (2%) 2 (1%) 9 (6%) 35 (22%) 93 (59%) 14 (9%) 5 (3%) 14 (9%) 70 (44%) 44 (28%) 8 (5%) 22 (14%) 66 (42%) 12.3 (8.6–15.9) 86 (54%) 2.7 (0–7) 7 (4%) 2 (0–10) 2 (0–4) 12 (0–100) 40 46 41 (11–83)
72 (46%) 43.7 (6–199) 13.7 (0.5–120)
For continuous variables, the median and range are listed.
33
LN metastasis, but they had oral cavity or oropharyngeal cancer judged to place them at risk for bilateral neck metastasis. Patients treated with surgery had resection of the primary tumor and modified radical dissection of the ipsilateral neck. No patient underwent dissection of the contralateral neck. Pathological evaluation of tumor margins and lymph node involvement included histochemistry using epithelial mucin stains and immunohistochemistry with cytokeratin. Almost all patients with oral cavity cancer received postoperative RT (35 of 37). In contrast, a higher number of patients with oropharyngeal cancer received primary compared with postoperative RT (56 and 40 patients, respectively). The majority of the patients with laryngeal (9 of 13) and hypopharyngeal (9 of 13) cancer received primary RT. In addition, 3 patients with unknown primary cancer received postoperative RT following neck dissection. For the analysis of predictive factors, factors specific for primary RT (GTV size) or for postoperative RT (pathologic tumor size, margins status, ECE, pathologically involved nodes) were analyzed separately for each of the patient groups (see statistical analysis).
Target definitions Our policies for the selection of the neck levels for each tumor site and stage, and for the delineation of the targets throughout the neck have been detailed elsewhere [14,15]. The targets include the gross tumor volumes (GTVs) and the clinical target volumes (CTVs) consisting of subclinical disease in the vicinity of the primary tumors, and neck nodal volumes at risk for subclinical disease. The delineation of the GTV was made according to both radiologic and clinical data: in addition to the radiological abnormalities, we relied on the results of indirect fiberoptic endoscopy in the clinic, and on the reports of direct endoscopy under anesthesia performed in all patients by a head and neck surgeon before RT planning. CTV delineation rules have been similar to those described by others [7,20,32,39]; the main distinction of our policies have been in the upper neck targets: on the side of the neck ipsilateral to the primary tumor, we defined level II CTV through the base of the skull. In the contralateral, grossly noninvolved, neck side, the most superior level II CTV was delineated at the axial planning CT scan in which the posterior belly of the digastric muscle was bisected by the jugular vein, ensuring that the subdigastric node was encompassed within the CTV, but excluding the junctional node from the CTV [14,15]. This was made in compliance with the anatomic observations of Rouviere, who described the subdigastric nodes as the upper-most node draining lymphatics from the nonnasopharyngeal upper aerodigestive tract [35]. Due to concerns about retrograde lymphatic flow in the ipsilateral neck, which had evidence of metastases in the large majority of cases, we have practiced a more conservative approach to the ipsilateral neck and outlined level II CTV further cranially to the base of skull. Initially, the retropharyngeal nodes were contoured through the top of C1 based on these observations. However, following two recurrences superior to the top of C1 and posterior to the upper nasopharynx, the retropharyngeal nodal CTVs were modified in 1999 to include the base of the skull in front of the clivus.
34
Recurrence predictors after HN IMRT
Therapy From 1994 until early 1996, parotid-sparing RT was delivered to 28 patients with conformal 3D techniques [17], and from 1996 through 2003, static multisegmental IMRT was delivered to 130 patients according to previously published methods [16,18]. All targets, including the CTVs in the low neck, were encompassed in the conformal/IMRT plans. The GTVs and the CTVs were expanded uniformly by 0.5 cm to yield their corresponding planning target volumes (PTVs). The median prescribed doses to the PTVs of the gross tumor, subclinical disease, and post-operative surgical beds, were 70.4 Gy (range: 66–76 Gy), 50.4 Gy (46–54 Gy), and 61.2 Gy (57.6–64 Gy), respectively. PTVs of postoperative tumor beds, where close margins or ECE were noted received prescribed doses of 64 Gy, and where these features were absent, they were prescribed 57.6–60 Gy. Daily fractions were 1.8–2.0 Gy. Cisplatin-based concurrent chemotherapy was delivered to 66 patients, the majority of whom (56 patients) were treated with primary radiation. In recent years, all patients who presented with neck lymphadenopathy measuring over 3 cm and who received primary chemo-RT underwent planned selective neck dissection 3 months after the completion of RT, according to an institutional protocol. The use of primary chemoradiation instead of surgery and postoperative RT became more prevalent in the latter years of the study as institutional policies shifted toward organ preservation approaches [13,37]. Clinical follow-up of all patients was performed by head and neck oncologic surgeons and radiation oncologists every 6–8 weeks during the first two years, every 3 months in the third year, and every 6 months, thereafter. Radiologic evaluation of the head and neck was performed at the time of clinical suspicion of recurrence.
patients, these factors included details from the pathological reports (largest dimension of primary tumor, number of positive lymph nodes, number of involved lymph node levels, percent positive lymph nodes, ECE, and findings of positive surgical margins). In the primary RT patients, the primary tumor size, represented by the volume of the GTV, and the volume of the largest LN, were derived from the treatment planning CT scans. Actuarial LR recurrence-free, disease-free, and overall survival curves were estimated using the Kaplan–Meier method. Patients failing in distant sites were not censored for LR recurrence-free survival. Log-rank tests were used to test for marginal differences in survival among the levels of the categorical predictors. Likelihood ratio tests from a proportional hazards model were used to test for significance of continuous predictors. Stepwise regression was used in a proportional hazards model to obtain a joint model for LR recurrence free survival. Individual significance testing in this joint model was done using Wald chi-squared statistics. All analysis was performed using the SAS system (SAS Institute, Cary, NC). Statistical significance was determined as P!0.05.
Results The median follow-up was 36.4 months, range 6–127 months. Of 117 patients still alive, 71 had at least 3 years of follow up. Twenty-three LR recurrences (15%) were observed at a median post-therapy interval of 13.3 months. The 2-year actuarial LR recurrence-free survival (LRRFS) was 85%. The 5-year LRRFS was 83%. The actuarial 2- and 5-year overall survival were 84 and 73%, respectively (Fig. 1).
Evaluation of tumor recurrence vs. dose Our methods to characterize recurrent tumors as in-field, marginal, or out-of-field were detailed elsewhere [11]. In brief, each patient’s recurrent tumor volume (Vr) was delineated on the CT or MRI scan at the time of recurrence, transferred to the planning CT scan via image registration, and a dose-volume-histogram (DVH) of Vr was calculated. The recurrence was defined as in-field if O95% of Vr had received the prescribed dose. Marginal and out-of-field recurrences were defined, where Vr had received between 25 and 95% and less than 25% of the prescribed dose, respectively. In cases of LR recurrences, the doses delivered to the rest of the PTV in which a recurrence occurred was assessed by subtracting the Vr from the PTV and calculating a DVH for the subtracted volume. The doses delivered to 1% of the volumes receiving the lowest or the highest dose in each target were reported as the minimal or the maximal target doses, respectively (rather than the doses delivered to single voxels receiving the lowest or highest dose).
Statistical analysis All the factors detailed in Table 1 were tested as potential clinical predictors of LR recurrences. Salvage of recurrences was not taken into account in the evaluation of recurrencefree survival or its predictors. In the postoperative RT
Fig. 1. Local–regional recurrence-free survival, Disease-free Survival, and Overall 1survival
M. Feng et al. / Radiotherapy and Oncology 77 (2005) 32–38
Patient and tumor characteristics in cases of LR recurrences Details of patients and tumors in cases, where LR recurrences were noted, and in cases without LR recurrences, are provided in Table 2. No AJCC stage I patients had LR recurrence. However, 1 of 9 (11%) stage II, 5 of 35 (14%) stage III, 15 of 93 (16%) stage IVA, 1 of 14 (7%) stage IVB, and 1 of 5 (20%) recurrent patients experienced LR recurrence. Patients with oral cavity cancer had the highest number of LR recurrences: they accounted for 48% of all recurrences although they only comprised 23% of all tumors. By site, 11 of 36 (31%) oral cavity, 6 of 94 (6%) oropharynx, 4 of 12 (33%) larynx, 2 of 13 (15%) hypopharynx, and 0 of 3 (0%) unknown
35
primary developed LR recurrence. Three of 7 (43%) patients with positive margins, 10 of 79 (13%) with negative margins, 10 of 39 (25%) with ECE and 4 of 46 (9%) without ECE developed LR recurrence. Postoperative patients with recurrence had generally larger primary tumors than those without recurrence (median of 3.6 vs. 2.5 cm, respectively). They also had more positive lymph nodes (median 4 vs. 2, respectively).
Evaluation of patient, tumor, and treatment factors Using univariate analysis, tumor site was highly predictive for LR failure in both primary and postoperative patients (PZ0.0005). Patients with oropharyngeal primaries had
Table 2 Patient and tumor characteristics for patients with and without LR recurrence
Age Tumor site Oral cavity Oropharynx Larynx Hypopharynx Unknown Primary Tumor stage (AJCC) I II III IVA IVB Recurrent Tumor histology Well-differentiated Moderately differentiated Poorly differentiated Not available Chemotherapy Hemoglobin Post-operative radiation Number of patients Size of primary tumor specimen (cm) Patients with positive margins Number lymph nodes containing metastasis Number of lymph node levels containing metastasis Percent of total lymph nodes containing metastasis ECE (C) ECE (K) Time from surgery to initiation of IMRT (days) Definitive radiation Number of patients GTV (cc) Largest neck node volume (cc)
Patients with
Patients without
P-value
LR recurrence (23)
LR recurrence (135)
55.7 (24–77)
56.3 (29–86)
11 (31%) 6 (6%) 4 (33%) 2 (15%) 0 (0%)
25 (69%) 88 (94%) 8 (67%) 11 (85%) 3 (100%)
0 (0%) 1 (11%) 5 (14%) 15 (16%) 1 (7%) 1 (20%)
2 (100%) 8 (89%) 30 (86%) 78 (84%) 13 (93%) 4 (80%)
3 (21%) 13 (19%) 4 (9%) 3 (14%) 8 (12%) 11.9 (9–14.8)
11 (79%) 57 (81%) 40 (91%) 19 (86%) 58 (88%) 12.4 (8.6–15.9)
NS NS
14 (16%) 3.6 (1–7)
72 (84%) 2.5 (0–6)
0.02
3 (43%) 4 (0–10)
4 (57%) 2 (0–8)
0.04a 0.006a
2 (0–4)
1 (0–3)
NS
16 (1–47)
9 (0–100)
NS
10 (25%) 4 (9%) 39.3 (28–63)
30 (75%) 42 (91%) 41.9 (11–83)
0.01 NS
9 (13%) 28.9 (10–53) 14.3 (1–42)
63 (87%) 46.2 (6–199) 13.2 (1–120)
NS NS
NS 0.0005a
NS
NS
For continuous variables, the median and range are listed. a Statistically significant in multivariate analysis.
36
Recurrence predictors after HN IMRT
Dosimetric evaluation of LR recurrences
Fig. 2. Local/regional disease-free survival by site (the figure does not include 3 patients with unknown primary tumors).
a 2-year actuarial LR recurrence-free survival (LRRFS) of 94%, significantly higher than patients with primary tumors in other sites (Fig. 2). In postoperative RT patients, the pathologic tumor size (PZ0.024), margin status (PZ0.046), presence or absence of ECE (PZ0.01), and the number of positive lymph nodes (PZ0.006) were also significant predictive features (Table 2). There was no difference in disease free or local–regional recurrence free survival between patients receiving definitive or post-operative therapy. The joint multivariate model showed tumor site (oropharynx vs. other all sites) to be a significant predictor of LR recurrences in all patients, and the presence of positive margins, and number of positive lymph nodes as significant predictors for LR recurrence in postoperative RT patients. The hazard for LR recurrence, including the other significant factors in the model, was increased by a factor of 1.44 for each additional pathologically involved lymph node. Two of 16 (12%) patients with pathological stage N0 had LR failure compared with 12 failures of 70 (16%) patients with pathological NC; the difference was not statistically significant due to the small number of N0 patients. The presence of positive margins increased the hazard for LR recurrence by 9.16. Factors not found to be predictive were age, pre-treatment hemoglobin, AJCC stage, and treatment with chemotherapy in all patients; the number of positive lymph node levels, percent positive lymph nodes, and the time from surgery to the initiation of radiation therapy in postoperative patients. In patients receiving primary RT, the volumes of the GTVs or the largest lymph node in each patient, as defined on the treatment planning CT scans, were not found to predict LR failure. Separate evaluation of patients with oropharyngeal and with larynx/hypopharynx primary tumors receiving definitive RT failed to identify the volumes of the GTVs as a significant predictive factor of LR recurrence in either group.
Of the 23 LR recurrences, 19 were in-field and 4 were marginal. Each of the marginal recurrences could be explained by a specific clinical decision regarding target delineation, detailed elsewhere [15]. In brief, these cases included marginal recurrences in retropharyngeal nodes cranial to C1 (initially, the cranial edges of the CTVs for the retropharyngeal nodes were delineated at the top of C1 according to Rouviere’s observations), a recurrence at level VI following therapy of oral cavity cancer with significant involvement of levels III–IV, and a marginal recurrence at an unexpected site in a patient irradiated for recurrence a few years after neck dissection, allowing the growth of new lymphatics whose position is unpredictable. No recurrences were observed in the contralateral high neck, where the level II CTV excluded the junctional nodes. No differences were found in the minimal, maximal or mean doses delivered to the PTVs of patients with and without recurrences (in both groups the meanCSD of the minimal PTV dose was 95C4%, the mean PTV dose 100C2%, and the maximal PTV dose 106C3%, of the prescribed dose). Additionally, in each of the 19 in-field recurrences there were no differences in the minimal doses delivered to Vr compared with the minimal doses delivered to the rest of the respective PTVs.
Discussion In this series, primary tumor site was the most significant predictive factor for LR recurrence. Patients with oropharyngeal primary tumors fared significantly better than all other primary sites, whether treated with primary or with postoperative RT. The relatively high success rate of primary irradiation for advanced oropharyngeal cancer has long been noted, offering similar rates of LR control as surgery and postoperative RT, but with less associated morbidity and better quality of life [34]. Recently, altered fractionated RT and concomitant chemotherapy has resulted in further improvements of the LR control rates of oropharyngeal cancer [5,21]. The Washington University experience with IMRT for oropharyngeal cancer has recently been reported, with 87% 4-year LR control rate in 74 patients [6]. Patient selection factors, length of follow-up, and other factors including differences in diagnostic and therapeutic procedures that are not related to RT technology, preclude meaningful direct comparisons among these series and ours. In comparison with oropharyngeal cancer, patients with oral cavity cancer in our series, almost all of whom received postoperative RT, fared less well. Their LR control rates were identical to those reported by Zelefsky et al. following standard postoperative RT [42]. These results highlight the aggressive nature of oral cavity cancer in many cases requiring bilateral neck postoperative RT. Other significant factors in the joint model were positive resection margins and the number of involved lymph nodes. Positive resection margins were quite rare in our series. Several large retrospective series of standard postoperative RT have demonstrated the impact of involved margins on LR control [2,8,10,22,27]. The number of positive lymph nodes has also been shown to impact poorly on locoregional control [1,12,28,33]. In our study, the hazard increased by 1.44 for
M. Feng et al. / Radiotherapy and Oncology 77 (2005) 32–38
each positive node. In addition, if patients had extracapsular extension in their positive nodes, they were at higher risk for LR. This is also consistent with prior studies for conventional postoperative RT [1,12,19,23,30]. Recent results of randomized studies suggest that treatment intensification via the addition of concurrent chemotherapy may improve the results of postoperative RT in patients with these adverse prognostic factors [3,9]. CT-based tumor size has previously been reported to bear prognostic importance in primary irradiation of laryngeal [26] and oropharyngeal [6,31] cancer, following both standard RT [26,31] and IMRT [6]. Mendenhall et al. found that size was not an independent factor in multivariate analysis, but that after stratification by primary site, volume was independently significant for supraglottic and glottic cancers, but not for oropharyngeal ones [29]. In our series, the pathologic primary tumor size in the post-operative RT patients was predictive of LR recurrences. However, the volumes of the GTVs in definitively irradiated patients were not found to be predictive. Specifically, we did not find it to be predictive in the oropharyngeal cancer patients who constituted the largest patient population site-wise. In our practice, the delineated GTVs extended in many cases beyond the obvious radiologic abnormalities. These extensions of the GTVs were not detected radiologically but were judged clinically to contain gross disease according to the descriptions of tumor extent made during direct endoscopy under anesthesia. Thus, the GTVs in our studies differed from those derived solely from imaging. We do not know whether or not a delineation of the GTVs based strictly on the radiological abnormalities, as analyzed in other series cited above, would emerge in our series as a significant predictor of LR recurrences. Dosimetric factors were not significant predictors of infield recurrences, all of which occurred within the CTVs, mainly due to the strict dose homogeneity criteria used in this series. Such homogeneity can be achieved in both forward planned and inverse-planned IMRT [38]. We cannot address, therefore, the relative importance of cold spots within the targets or what level of inhomogeneity may be allowed. Another confounding factor is the uncertainties related to the exact origin of the recurrent tumor. We used the doses delivered to the epicenter of the recurrence to define whether it was in-field, marginal or out-of-field. However, the accuracy of this determination depended in part on the timing of the radiologic evaluation relative to the growth of the recurrent tumor. Two recent prospective randomized trials, RTOG 95-01 and EORTC 22931, found that the addition of concurrent Cisplatin to postoperative RT for patients with high risk features including positive margins improved local control and may impact on overall survival [3,9]. Few of the postoperative RT patients in our series were treated with concurrent chemoradiation, while almost all of our primary RT patients received concurrent therapy; therefore, our series did not have the power to detect a potential significance of chemotherapy in LR recurrences. This series is very heterogeneous regarding tumor sites and therapy. Larger patient numbers at each site and therapy (definitive and post operative RT) may unveil additional prognostic
37
factors for LR failure, which this series has not been powered to detect. In conclusion, the large majority of the LR recurrences in our series occurred in-field. Similar observations have recently been reported in other series of HN IMRT or parotid-sparing 3DCRT [4,7,24,25]. In our series, dosimetric factors could not differentiate between the patients who had in-field recurrences and those who did not have LR failure, a byproduct of the strict target dose homogeneity criteria employed [16]. Thus, clinical, rather than dosimetric factors, predicted the risk of LR recurrences in this series. These risk factors were similar to those reported in the past following standard RT. Whether higher doses to targets at high risk of failure may reduce recurrence rates is a hypothesis that needs to be tested. Dose escalation studies that rely on the ability of IMRT to exclude noninvolved tissue from the high dose volumes have been proposed [40,41]. The clinical factors that significantly predicted LR failure in our series can serve as eligibility criteria for such dose escalation trials, as patients lacking these features had a low risk of LR failure.
Acknowledgements Supported by NIH grants CA59827 and CA78165, and the Duke Family Head and Neck Cancer Research Fund.
*
Corresponding author. Dr. Avraham Eisbruch, Department of Radiation Oncology, University of Michigan Hospital, Ann Arbor MI 48109-0010. E-mail address: [email protected]
Received 4 March 2005; received in revised form 30 June 2005; accepted 7 July 2005; available online 8 September 2005
References [1] Alvi A, Johnson JT. Extracapsular spread in the clinically negative neck: implications and outcome. Otolaryngol Head Neck Surg 1996;114:65–70. [2] Bastit L, Blot E, Debourdeau P, et al. Influence of the delay of adjuvant postoperative radiation therapy on relapse and survival in oropharyngeal and hypopharyngeal cancers. Int J Radiat Oncol Biol Phys 2001;49:139–46. [3] Bernier J, Domenge C, Ozsahin M, et al. Postoperative irradiation with or without concomitant chemotherapy for locally advanced head and neck cancer. NEJM 2004;350: 1945–52. [4] Bussels B, Maes A, Hermans R, et al. Recurrences after conformal parotid-sparing radiotherapy for head and neck cancer. Radiother Oncol 2004;72:119–27. [5] Calais G, Alfonsi M, Bardet E, et al. Randomized trial of radiation therapy versus concomitant chemotherapy and radiation therapy for advanced-stage oropharynx carcinoma. J Natl Cancer Inst 1999;91:2081–6. [6] Chao KS, Ozyigit G, Blanco AI, et al. Intensity-modulated radiation therapy for oropharyngeal carcinoma: impact of tumor volume. Int J Radiat Oncol Biol Phys 2004;59:43–50. [7] Chao KSC, Wippold FJ, Ozygit G, et al. Determination and delineation of nodal target volumes for head-and -neck cancer based on patterns of failure in patients receiving definitive and postoperative IMRT. Int J Radiat Oncol Biol Phys 2002;53: 1174–84.
38
Recurrence predictors after HN IMRT
[8] Cook JA, Jones AS, Phillips DE, et al. Implications of tumour in resection margins following surgical treatment of squamous cell carcinoma of the head and neck. Clin Otolaryngol 1993;18: 37–41. [9] Cooper JS, Pajak TF, Forastiere AA, et al. Postoperative concurrent radiotherapy and chemotherapy for high-risk squamous-cell carcinoma of the head and neck. NEJM 2004; 350:1937–44. [10] Cooper JS, Pajak TF, Forastiere A, et al. Precisely defining highrisk operable head and neck tumors based on RTOG 85-03 and 88-24: targets for postoperative radiochemotherapy? Head Neck 1998;20:588–94. [11] Dawson LA, Anzai Y, Marsh L, et al. Patterns of loco-regional recurrence following parotid-sparing conformal and segmental intensity-modulated radiotherapy for head and neck cancer. Int J Radiat Oncol Biol Phys 2000;46:1117–26. [12] de Carvalho MB. Quantative analysis of the extent of extracapsular invasion and its prognostic significance: a prospective study of 117 cases of carcinoma of the larynx and hypopharynx. Head Neck 1998;20:16–21. [13] Department of Veterans Affairs Laryngeal Cancer Study Group. Induction chemotherapy plus radiation compared with surgery plus radiation in patients with advanced laryngeal cancer. NEJM 1991;324:1685–90. [14] Eisbruch A, Foote RL, O’Sullivan B, et al. IMRT for head and neck cancer: emphasis on the selection and delineation of the targets. Semin Radiat Oncol 2002;12:238–49. [15] Eisbruch A, Marsh LH, Dawson LA, et al. Recurrences near base of skull after IMRT for head-and-neck cancer: implications for target delineation in high neck and for parotid gland sparing. Int J Radiat Oncol Biol Phys 2004;59:28–42. [16] Eisbruch A, Marsh LH, Martel MK, et al. Comprehensive irradiation of head and neck cancer using conformal multisegmental fields: assessment of target coverage and noninvolved tissue sparing. Int J Radiat Oncol Biol Phys 1998;41: 559–68. [17] Eisbruch A, Ship JA, Martel MK, et al. Parotid gland sparing in patients undergoing bilateral head and neck irradiation: techniques and early results. Int J Radiat Oncol Biol Phys 1996;36:469–80. [18] Fraass BA, Kessler ML, McShan DL, et al. Optimization and clinical use of multisegment intensity-modulated radiation therapy for high-dose conformal therapy. Semin Radiat Oncol 1999;9:60–77. [19] Greenberg JS, Flowler R, Gomez J, et al. Extent of extracapsular spread: a critical prognosticator in oral tongue cancer. Cancer 2003;97:1464–70. [20] Gregoire V, Coche E, Cosnard G, et al. Selection and delineation of lymph node target volumes in head and neck conformal radiotherapy. Proposal for standardizing terminology and procedure based on the surgical experience. Radiother Oncol 2000;56:135–50. [21] Horiot JC, Le Fur R, N’Guyen T, et al. Hyperfractionation versus conventional fractionation in oropharyngeal carcinoma: final analysis of a randomized trial of the EORTC cooperative group of radiotherapy. Radiother Oncol 1992;25:231–41. [22] Jacobs JR, Ahmad K, Schuller CR. Implications of positive surgical margins. Laryngoscope 1993;103:64–8. [23] Jose J, Coatesworth AP, Johnston C, MacLennan K. Cervical node metastases in squamous cell carcinoma of the upper aerodigestive tract: the significance of extracapsular spread and soft tissue deposits. Head Neck 2003;25:451–6. [24] Lee N, Xia P, Fischbein NJ, et al. Intensity-modulated radiation therapy for head-and-neck cancer: the UCSF experience focusing on target volume delineation. Int J Radiat Oncol Biol Phys 2003;57:49–60.
[25] Levendag P, Braaksma M, Coche E, et al. Rotterdam and Brussels CT-based neck nodal delineation compared with the surgical levels as defined by the American academy of otolaryngology—head and neck surgery. Int J Radiat Oncol Biol Phys 2004;58:113–23. [26] Mancuso AA, Mukherji SK, Schmalfus I, et al. Preradiotherapy computed tomography as a predictor of local control in supraglottic carcinoma. J Clin Oncol 1999;17:631–7. [27] Mantravadi RV, Haas RE, Liebner EJ, et al. Postoperative radiotherapy for persistent tumor at the surgical margin in had and neck cancers. Laryngoscope 1983;93:1337–40. [28] Mantravadi RV, Skolnik EM, Haas RE, et al. Patterns of cancer recurrence in the postoperatively irradiated neck. Arch Otolaryngol 1983;109:753–6. [29] Mendenhall WM, Morris CG, Amdur RJ, et al. Parameters that predict local control after definitive radiotherapy for squamous cell carcinoma of the head and neck. Head Neck 2003;25: 535–42. [30] Myers JN, Greenberg JS, Vo V, Roberts D. Extracapsular spread. A significant predictor of treatment failure in patients with squamous cell carcinoma of the tongue. Cancer 2001;92: 3030–6. [31] Nathu R, Mancuso AA, Zhu TC, et al. The impact of primary tumor volume on local control for oropharyngeal squamous cell carcinoma treated with radiotherapy. Head Neck 2000;22:1–5. [32] Nowak PJCM, Wijers OB, Lagerwaard FJ, Levendag PC. A threedimensional CT-based target definition for elective irradiation of the neck. Int J Radiat Oncol Biol Phys 1999;45:33–9. [33] Olsen KD, Daruso M, Foote RL, et al. Primary head and neck cancer. Histopathologic predictors of recurrence after neck dissection in patients with lymph node involvement. Arch Otolaryngol Head Neck Surg 1994;120:1370–4. [34] Parsons JT, Mendenhall WM, Stringer SP, et al. Squamous cell carcinoma of the oropharynx. Surgery, radiation therapy, or both. Cancer 2002;94:2967–80. [35] Rouviere H. Lymphatic systems of the head and neck Translated by MJ tobias. Ann Arbor, MI: Edwards Brothers; 1938. [36] Smith BD, Haffty BG. Prognostic factors in patients with head and neck cancer. In: Harrison LB, Sessions RB, Hong WK, editors. Head and neck cancer: a multidisciplinary approach. 2nd ed. Philadelphia: Lippincott Williams & Wilkins; 2004. p. 49–73. [37] Urba SG, Forastiere AA, Wolf GT, et al. Intensive induction chemotherapy and radiation for organ preservation in patients with advanced resectable head and neck carcinoma. JCO 1994; 12:946–53. [38] Vineberg KA, Eisbruch A, Coselmon MM, et al. Is uniform target dose possible in IMRT plans in the head and neck? Int J Radiat Oncol Biol Phys 2002;52:1159–72. [39] Wijers OB, Levendag PC, Tan T, et al. A simplified CT-based definition of the lymph node levels in the node negative neck. Radiother Oncol 1999;52:35–42. [40] Wu Q, Manning M, Schmidt-Ullrich R, Mohan R. The potential for sparing of parotids and escalation of biologically equivalent dose with intensity modulated radiation treatments of head and neck cancers: A treatment design study. Int J Radiat Oncol Biol Phys 2000;46:195–205. [41] Wu Q, Mohan R, Morris M, et al. Simultaneous integrated boost intensity-modulated radiotherapy for locally advanced headand-neck squamous cell carcinomas. I: dosimetric results. Int J Radait Oncol Biol Phys 2003;56:573–85. [42] Zelefsky MJ, Harrison LB, Fass DE, et al. Postoperative radiotherapy for oral cavity cancers: impact of anatomic subsite on treatment outcome. Head Neck 1990;12:470–5.