Patterns of failure after postoperative intensity-modulated radiotherapy for locally advanced and recurrent head and neck cancer

Abstract

Objective

To evaluate the feasibility of postoperative intensity-modulated radiotherapy for head and neck cancer by investigating the patterns of failure after this therapy.

Methods

A retrospective chart review was performed.

Results

Between March 2006 and December 2013, 122 consecutive patients with head and neck squamous cell carcinoma were treated by surgery followed by postoperative intensity-modulated radiotherapy. In regard to the site of the primary tumor, 59 (48%) patients had cancer of the oral cavity, 31 (26%) patients had cancer of the hypopharynx, 14 (11%) patients had cancer of the oropharynx, 10 (8%) patients had cancer of the larynx and 8 (7%) patients had cancer of unknown primary. The median follow-up period of the surviving patients was 54 months (range, 25–115). Concurrent chemotherapy was administered in 76 patients (62%). The median prescribed radiation dose was 66 Gy. The 3-year overall survival, progression-free survival, distant metastasis free survival and loco-regional control rates were 59%, 48%, 52.4% and 71%, respectively. Of the 122 patients, 32 developed loco-regional recurrence as the initial recurrence, including in-field recurrence in 26 patients, marginal recurrence in five patients and out-field recurrence in seven patients. Of the five patients with marginal recurrence, four have had two or more surgeries before the intensity-modulated radiotherapy and three had oral cavity cancer. Severe adverse events were not frequent, occurring at a frequency of <5%, except for mucositis. No severe toxicities associated with the flap reconstruction were observed either.

Conclusion

Postoperative intensity-modulated radiotherapy appears to be effective and feasible for patients with head and neck squamous cell carcinoma.

postoperative radiotherapy, intensity-modulated radiotherapy, patterns of failure

Introduction

Intensity-modulated radiotherapy (IMRT) is now widely accepted as a useful treatment modality for patients with head and neck squamous cell carcinoma (HNSCC), since it can contribute to improving the quality of life of the treated patients by reducing the unintended dose to normal structures such as the oral cavity and parotid glands (1–6). However, IMRT has a very sharp dose gradient between the target and the surrounding normal tissues (7–9), which could lead to marginal recurrence, especially when IMRT is used in postoperative settings where prior surgical manipulation might have dispersed microscopic disease widely over the surgical bed (10–12). In addition, definition of the target tends to be more difficult in the adjuvant postoperative setting as compared to that in the definitive radiotherapy setting, because of the lack of gross visible tumor tissue and architectural distortion by surgery (13). Moreover, the dose inhomogeneity arising from the technological nature of IMRT poses another concern; focal hot and/or cold spots may cause unexpected effects on the treatment outcomes (14,15).

Under this circumstance, we developed criteria to define the Clinical Target Volume (CTV) in the adjuvant postoperative IMRT setting taking anatomical compartments of the neck intro consideration, and have been treating patients since the advent of IMRT at our institution. In this retrospective study, we analyzed the treatment outcomes of patients with HNSCC treated by surgery followed by adjuvant postoperative IMRT with or without chemotherapy, especially focusing on the patterns of failure.

Materials and methods

Study design and institutional review board approval

We performed a retrospective chart review of the patients who were treated by surgery followed by adjuvant postoperative IMRT. This study was conducted with the approval of the institutional review board.

Patients

Classification of the primary tumor sites and staging was performed according to the American Joint Committee on Cancer (AJCC) Staging Manual, seventh edition. All patients underwent computed tomography (CT) and/or magnetic resonance imaging (MRI), and endoscopy of the head and neck as pretreatment evaluations before undergoing surgery. Positron emission tomography (PET) was performed where necessary.

Surgical procedure and assessment of the risk of recurrence

Basically, the surgical procedures consisted of dissection of the primary lesion and neck dissections with or without reconstruction. Neck dissections (either ipsilateral or bilateral) were routinely performed in the presence of clinical and/or radiological evidence of lymphadenopathy.

Postoperative radiotherapy was offered to patients with histopathological risk factors, such as a close or positive margin at the site of resection of the primary disease, extracapsular extension (ECE) of a nodal metastasis, multiple nodal metastases and/or lymphovascular invasion. After surgery, postoperative risk stratification was performed based on the intraoperative findings and findings of the resected specimens on histopathology. Patients with a histopathologically positive surgical margin (PSM) and/or ECE of a nodal metastasis were defined as being at a high risk for loco-regional recurrence, in accordance with the definitions described in previous reports (16,17). Concurrent chemotherapy, consisting of cisplatin (CDDP) at the dose of 80 mg/m² on Days 1, 22, 43, weekly CDDP at the dose of 20 mg/m², weekly carboplatin (CBDCA) at a dose corresponding to an area under the curve of 2 mg/mL/min, or CDDP 80 mg/m² on Days 1, 28, 55 and 5-fluorouracil 800 mg/m² on Days 1–5, 28–32, 55–59, was administered to patients with high-risk features. Other patients received radiotherapy alone.

Contouring of the clinical target volume

All patients underwent simulation by CT (3-mm slice thickness) with or without contrast enhancement. For immobilization, a thermoplastic mask was utilized in all cases. The obtained CT images were then sent to a radiotherapy planning system.

Lymph node stations in the neck, including levels I–V, and the supraclavicular and retropharyngeal nodes were designated according to the DAHANCA, EORTC, GORTEC, NCIC and RTOG consensus guidelines (18). Anatomical compartments were determined according to the criteria. The potential sites for PSM and ECE were identified by the surgeon, or identified on the basis of the intraoperative findings and histopathological findings of the resected specimens.

Figure 1 illustrates the concept of target delineation in this study. ‘High-risk Clinical Target Volume (CTV)’: first, tumor beds adjacent to the sites of the PSM and/or ECE were defined as the virtual gross tumor volume (vGTV). Then, the vGTV was uniformly expanded by 1 cm (vGTV+1). In patients with a PSM at the pharyngeal mucosa, a 2-cm margin was added along with the mucosa (PSM+2) to contain potential mucosal spreading (Fig. 1A). The entire anatomical region including vGTV+1 and PSM+2 was defined as the high-risk CTV. The anastomoses, including those of the reconstruction flaps, and the postsurgical scar tissues adjacent to the vGTV were also included (Fig. 1B).

Figure 1.

Schematic example of the anatomical boundary-based target definitions. (A) The virtual gross tumor volume (vGTV, solid black area) was defined at the site where microscopic disease resulted from positive surgical margin (PSM) and/or extracapsular extension (ECE) are plausible (black arrowhead). Then, the vGTV was uniformly expanded by 1 cm (dotted black line). In cases where widespread mucosal extension was possible, an extra 1 cm margin was added (dotted black circle). (B) The vGTV+margin and the entire anatomical compartment containing was defined as the high-risk Clinical Target Volume (CTV) (internal area bounded by the solid black line). (C) The intermediate-risk CTV (solid gray area) was defined as the anatomical compartments adjacent to the high-risk CTV. The low-risk CTV (dotted black area) was defined as the surgical bed and elective lymph node area.

‘Intermediate-risk CTV’: the anatomical compartments adjacent to the ‘high-risk CTV’, as follows (i) Nodal levels adjacent to the high-risk CTV; (ii) when the ‘high-risk CTV’ is attached to the parapharyngeal space, the ipsilateral part of the parapharyngeal space; (iii) parts of the pterygoid muscles and base of the tongue when the ‘high-risk CTV’ is attached to these compartments; (iv) scar tissues around the tracheal stoma, especially in patients with laryngeal or hypopharyngeal cancer; and (v) the entire free jejunal flap and its anastomoses in cases undergoing flap reconstruction has been performed after laryngectomy or pharyngolaryngectomy (Fig. 1C).

‘Low-risk CTV’: the entire operative bed and elective lymph node levels, including the level II–V nodes, supraclavicular nodes and retropharyngeal nodes. In patients with the primary tumor in the oral cavity, level Ib is also included electively (Fig. 1C).

The planning target volumes (PTVs) for high-, intermediate- and low-risk CTVs were created by a circumferential 5-mm expansion of the corresponding CTVs. Trimming of the PTVs within 3 mm beneath the skin surface was performed to enable proper dose evaluations during the IMRT planning.

Radiotherapy techniques and dose prescription

IMRT plans were created using Xio version 4.5.0 (Elekta, Stockholm, Sweden) for the first six patients, and with Pinnacle 3 (Philips, Amsterdam, the Netherlands) for the subsequent 116 patients.

A simultaneous integrated boost technique was adopted during the study period. Basically, the dose prescriptions to the high-, intermediate- and low-risk PTVs were 66 Gy, 60 Gy and 54 Gy, respectively, administered in 33 fractions. In principle, the plan was approved when 95% or more of the PTV received 95% of the prescribed dose (D95 = 95% dose prescription), after eliminating hot spots receiving ≥107% of the prescribed dose.

We used 6 MV X-rays, the fixed 9-beam arrangement and the step-and-shoot technique with a 5–10-mm width multi-leaf collimator. The entire PTV was treated by the extended-field technique at the same time, encompassing the whole neck.

Outcome measures and statistical considerations

Follow-up visits were scheduled monthly for the first 2 years after treatment, at least once every 3 months during the third year, and once every 6 months thereafter. Radiological examinations, including CT and/or MRI of the head and neck were performed at least twice within the first 6 months after treatment, and once every 6 months thereafter.

We performed time-to-event analyses using the Kaplan–Meier method for several outcome measures. The overall survival (OS) was calculated from the day of start of the adjuvant therapy until death from any cause. Progression-free survival (PFS) was calculated from the day of start of the adjuvant therapy until death from any cause or any relevant events including local, regional and distant recurrence. Distant metastasis free survival (DMFS) rate was calculated from the day of the start of the adjuvant therapy until death from any cause or distant metastasis. Loco-regional control (LRC) was calculated from the day of start of the adjuvant therapy until local and/or regional recurrence; patients who died without experiencing any of these events were censored at the time of the last follow-up. The statistical significance of differences between the groups was evaluated using the log-rank test. Univariate and multivariate analyses to identify the clinicopathological factors influencing the treatment outcomes were performed using Cox regression analysis. Results with two-sided P-values of ≤0.05 were regarded as significant. Statistical calculations were performed using the JMP software, ver. 10 (SAS, Chicago, USA).

Evaluation of the patterns of failure

The recurrent tumor volume (Vrec) was defined based on the CT/MRI findings demonstrating recurrence. The Vrec was coregistered with the images of the IMRT-planning CT using the planning supporting system, MIM maestro ver.2.0 (MIM, Ohio, USA).

We classified the loco-regional recurrences according to the definitions of Chao et al. (19,20), as follows; (i) ‘in-field’: if >95% of the Vrec was within the high-risk or intermediate-risk CTV; (ii) ‘marginal’: if 20–95% of the Vrec was within the high-risk or intermediate-risk CTV; and (iii) ‘out-field’: if <20% of the Vrec was within either the high-risk or intermediate-risk CTV.

Evaluation of adverse events

Acute and late adverse events were graded according to the National Cancer Institute Common Terminology Criteria of Adverse Events, version 4.0. Acute adverse events were defined as toxicities observed within 90 days of completion of the adjuvant treatment, including IMRT, and late adverse events were defined as toxicities observed thereafter.

Results

Patient characteristics

From March 2006 to December 2013, 122 patients with HNSCC underwent postoperative adjuvant IMRT. Table 1 outlines the patients’ characteristics. There were 87 (71%) males and 35 (29%) females, with a median age of 64 years (range, 23–79 years). The primary disease sites were distributed as follows; oral cavity in 59 (48%) patients, hypopharynx in 18 (26%), oropharynx in 14 (11%), larynx in 10 (8%) and unknown primary in 6 (7%) patients. Approximately a half of the patients had had oral cavity cancer. Sixty-seven patients (55%) received adjuvant postoperative IMRT after the initial surgery, whereas 55 patients (45%) had undergone two or more additional salvage surgeries before receiving adjuvant postoperative IMRT. Some patients underwent flap reconstruction during surgery to compensate for the major tissue loss. Free jejunal, free transverse rectus abdominis (TRAM), free anterolateral thigh (ALT) and other kind of flaps were used for 27, 17, 4 and 14 patients, respectively. The pathological stage of the patients excluding those who had underwent reoperation was distributed as follows: Stage I disease, 2 (2%) patients; Stage II disease, 3 (2%) patients; Stage III disease, 17 (14%) patients; Stage IV disease, 82 (67%) patients. In regard to the histopathological findings, 37 patients (30%) had a PSM at the site of resection of the primary disease, 94 (77%) patients had ECE in the resected lymph nodes, and 109 patients (89%) had either PSM or ECE.

Table 1.

Patient characteristics

Characteristics	No. of patients	%
Sex
Male	87	71
Female	35	29
Primary site
Oral cavity	59	48
Hypopharynx	31	26
Oropharynx	14	11
Larynx	10	8
Unknown primary	8	7
No. of surgical procedure
1	67	55
≤2	55	45
Type of flaps for reconstruction
None	60	49
Jejunum	27	22
TRAM	17	14
ALT	4	3
Others	14	11
Pathological T-classification
pT0	3	2
pT1	5	4
pT2	16	13
pT3	12	10
pT4	7	6
NA	45	37
Pathological N-classification
pN0	10	8
pN1	20	16
pN2	79	65
pN3	4	3
NA	8	7
Pathological staging (AJCC)
I	2	2
II	3	2
III	17	14
IV	82	67
N.A.	18	15
Positive surgical margin (PSM)
Yes	37	30
No or N.A.	85	70
Extracapslar extention (ECE)
Yes	94	77
No or N.A.	28	23
Lymphovascular invasion
Yes	45	37
No or N.A.	77	63

N.A., not available; TRAM, transverse rectus abdominis; ALT, anterolateral thigh; AJCC, American Joint Committee on Cancer.

Procedure of adjuvant therapy

The median interval from the latest surgery to the start of postoperative IMRT was 48 days (range, 27–123 days). Of the patients, 76 (62%) received concurrent chemotherapy, including CDDP in 68 patients, CBDCA in seven patients and CDDP/5-FU in one patient. The overall treatment time for IMRT ranged from 21 to 58 days (median, 50 days). The IMRT could not be completed in three patients, in whom distant metastases were detected during the treatment course.

Treatment outcomes

Of all the patients enrolled in this study, 64 (52%) were alive until the last follow-up visit. The median follow-up period for the surviving patients was 54 months (range, 25–115 months). The 3-year actuarial rates of OS, PFS, DMFS and LRC rates for the entire patient population were 58.9% (95% confidence interval [CI] 50.0–67.3), 48.1% (95% CI 39.4–57.0), 52.4%(95% CI 43.4–61.2%) and 71% (95% CI 62.2–78.6), respectively (Fig. 2). The clinicopathological factors influencing the treatment outcomes are listed in Table 2. Univariate analysis identified pathological stage (hazard ratio [HR] 0.53, 95% CI 0.27–0.96, P = 0.03), number of metastatic lymph nodes (0–3 vs. 4 or more) (HR 0.42, 95% CI 0.24–0.72, P = 0.002) and primary site in the oropharynx (HR 0.22, 95% CI 0.04–0.71, P = 0.007) as influencing the OS rate, and number of metastatic lymph nodes (HR 0.53, 95% CI 0.31–0.88, P = 0.01), primary site in the oropharynx (HR 0.19, 95% CI 0.03–0.61, P = 0.0023) as influencing the PFS rate, and number of metastatic lymph nodes (HR 0.56, 95% CI 0.33–0.95, P = 0.03), primary site in the oropharynx (HR 0.21, 95% CI 0.03–0.67, P = 0.005) as influencing the DMFS rate, and primary site in the oral cavity (HR 2.04, 95% CI 1.02–4.27, P = 0.04) or oropharynx (HR 0.21, 95% CI 0.01–0.98, P = 0.045) as influencing the LRC rate. Multivariate analysis identified number of metastatic lymph nodes and primary site in the oropharynx as significantly influencing the OS, PFS and DMFS rate.

Figure 2.

Treatment outcomes estimated using the Kaplan–Meier method. (A) Overall survival (OS) rate; (B) Progression-free survival (PFS) rate; (C) Distant metastasis free survival (DMFS) rate; (D) Loco-regional control (LRC) rate.

Table 2.

Univariate and multivariate analyses of treatment outcomes. The categories with underlined P value indicate that there was a statistical significance

Variables	OS			PFS			DMFS			LRC
Monovariate analysis	HR	95% CI	P value	HR	95% CI	P value	HR	95% CI	P value	HR	95% CI	P value
Age	1.01	0.003–1.05	0.16	1.01	0.99–1.03	0.4	1.01	0.98–1.04	0.51	1.01	0.98–1.04	0.71
Sex (male as referent)	1.36	0.76–2.64	0.31	1.53	0.87–2.88	0.14	1.44	0.80–2.79	0.23	0.81	0.40–1.73	0.57
No. of surgery		0.78–1.38	0.6	1.07	0.78–1.37	0.64	1.05	0.76–1.34	0.74	1.28	0.94–1.59	0.11
Clinical stage (1–3 as referent)	0.69	0.36–1.22	0.21	0.82	0.47–1.39	0.48	1.14	0.66–2.07	0.63	0.82	0.36–1.70	0.6
Pathological stage (1–3 as referent)	0.53	0.27–0.96	0.03	0.61	0.34–1.03	0.07	1.68	0.95–3.12	0.07	0.57	0.24–1.21	0.15
Pathological differenciation (non-p/d as referent)	0.69	0.21–1.69	0.46	0.74	0.26–1.68	0.51	0.83	0.29–1.89	0.69	0.61	0.10–2.00	0.46
Concurrent chemotherapy	0.69	0.41–1.17	0.17	0.85	0.52–1.41	0.52	0.88	0.53–1.52	0.65	0.59	0.29–1.71	0.13
PSM and/or ECE	1.27	0.56–3.65	0.6	1.24	0.58–3.20	0.61	1.26	0.56–3.62	0.61	1.95	0.59–12.0	0.31
No. of metastatic LN (0–3 as referent)	0.42	0.24–0.72	0.002	0.53	0.31–0.88	0.01	0.56	0.33–0.95	0.03	0.57	0.27–1.17	0.3
Primary site
Oral cavity as referent	1.32	0.79–2.24	0.29	1.23	0.76–2.01	0.4	1.23	0.73–2.06	0.43	2.04	1.02–4.27	0.04
Hypopharyngeal as  referent	1.18	0.65–2.04	0.57	1.2	0.68–2.00	0.53	1.32	0.74–2.26	0.34	0.62	0.23–1.40	0.26
Oropharyngeal as  referent	0.22	0.04–0.71	<0.01	0.19	0.03–0.61	<0.01	0.21	0.03–0.67	0.005	0.21	0.01–0.98	0.045
Larynx as referent	1.4	0.53–3.01	0.46	1.23	0.47–2.62	0.64	1.44	0.55–3.10	0.42	1.57	0.47–3.99	0.43
Unknown primary as  referent	0.65	0.16–1.76	0.44	1.15	0.40–2.59	0.77	0.64	0.16–1.75	0.43	0.46	0.03–2.15	0.389
Multivariate analysis
Pathological stage  (1–3 as referent)	0.51	0.24–1.01	0.06	—	—	—	—	—	—	—	—	—
No. of metastatic LN  (0–3 as referent)	0.5	0.28–0.87	0.01	0.55	0.33–0.91	0.02	0.58	0.34–0.99	0.047	—	—	—
Primary site
Oral cavity as referent	—	—	—	—	—	—	—	—	—	1.67	0.83–3.63	0.1512
Oropharyngeal as  referent	0.3	0.05–0.99	0.048	0.25	0.04–0.79	0.01	0.26	0.04–0.84	0.02	0.29	0.02–1.47	0.1544

OS, overall survival; PFS, progression-free survival; DMFS, distant metastasis free survival; LRC, loco-regional control; HR, hazard ratio; CI, confidence interval.

Patterns of failure

A total of 63 (52%) patients developed recurrence. Loco-regional recurrence occurred as the first recurrence in 38 (31%) patients; 8 patients (7%) had synchronous loco-regional recurrence and distant metastasis. The median time to recurrence from the start of adjuvant IMRT was 8.3 months (range, 2.6–38.5 months). Figure 3 outlines the patterns of loco-regional recurrence; 26 patients developed in-field recurrence, five patients developed marginal recurrence, and seven patients developed out-field recurrence. Among the patients with in-field recurrence, three patients had synchronous marginal recurrence, and another three patients had synchronous out-field recurrence.

Figure 3.

Venn's circles representing the pattern of loco-regional failure.

Among the 26 patients with in-field recurrence, 18 patients had recurrence in the high-risk CTV, six patients had recurrence in the intermediate-risk CTV, and two patients had synchronous recurrences in the high-risk and intermediate-risk CTVs. Eight out of the 26 patients with in-field recurrence had synchronous distant metastasis.

The details of the five patients who developed marginal recurrence are shown in Table 3. Of the five with marginal recurrence, three had the primary tumor in the oral cavity one had the primary tumor in the hypopharynx, and one had the primary tumor in the larynx. All but one of the patients with marginal recurrence had undergone two or more surgeries before the adjuvant IMRT. In regard to the site of the marginal recurrence, three of the patients showed subdermal recurrence, one showed local recurrence (tongue), and three showed recurrence in the nodal area. Three out of the five patients had synchronous distant metastasis. Only one of the five patients had isolated marginal recurrence (i.e. without synchronous loco-regional recurrence or distant metastasis). Figures 4 and 5 show the comparative images of IMRT dosimetry and sites of recurrence in the patient who developed marginal recurrence (designated as Patients #1 and #3, respectively, in the Table 3).

Figure 4.

Dosimetry of intensity-modulated radiotherapy (IMRT) (left) and positron emission tomography (PET)/ computed tomography (CT) imaging showing a subdermal marginal recurrence (red arrow) of a patient with hypopharyngeal cancer (Patient #1).

Figure 5.

Dosimetry of IMRT (left) and magnetic resonance imaging (MRI) showing a marginal periparotid recurrence of a patient with oral cavity cancer.

Table 3.

The details of the patients with marginal recurrence

Patient no.	Primary site	No. of surgery	Risk factors	Site of marginal recurrence	Synchronous recurrence
#1	Hypopharynx	2	ECE	Subdermal	None
#2	Larynx	1	ECE	Level II–III nodal area	In-field and distant
#3	Oral cavity	2	ECE	Level II nodal area (periparotid) and subdermal	Distant
#4	Oral cavity	3	ECE	Local (tongue) and subdermal	In-field and distant
#5	Oral cavity	2	PSM	Level II nodal area	In-field

ECE, extracapsular extension; PSM, positive surgical margin.

A total of seven patients developed out-field recurrence, including four with recurrence in the low-risk CTV, two with dermal metastasis, and one with recurrence outside the CTV in the neck. Three out of these seven patients with out-field recurrence had synchronous in-field recurrence and/or distant metastasis.

Thirty-three patients (27%) developed distant metastasis as the first recurrence. The primary disease site for these patients were as follows; oral cavity in 15 patients, hypopharynx in 12 patients, unknown in three patients, larynx in two patients, and oropharynx in one patient.

Adverse events

The worst-grade acute or late toxicities observed throughout the treatment and follow-up periods are listed in Table 4. The percentages of patients who developed Grade 3 or worse dermatitis, mucositis, xerostomia, loss of taste, swallowing dysfunction, osteomyelitis and hearing loss were 2%, 30%, 0%, 0%, 4%, 0% and 4%, respectively. None of the patients developed any Grade 4 or worse toxicities and there were no cases of severe toxicities associated with the flap reconstructions.

Table 4.

Acute and late toxicities

Toxicity grading	Dermatitis	Mucositis	Xerostomia	Taste loss	Swallowing disfunction	Osteomyelitis	Hearing loss
Grade 0–1	29	12	41	69	51	102	85
Grade 2	80	64	69	38	7	4	20
Grade 3	2	36	0	0	5	0	5
Grade 4–5	0	0	0	0	0	0	0
N.A.	1	0	2	5	49	6	2

Discussion

In the current study, the 3-year OS, PFS, DMFS and LCR rates were 58.9%, 48.1%, 52.4% and 71%, respectively. Thus, the treatment outcomes seemed satisfactory and comparable to previous reports (10,11,21–23), even though nearly a half of the patients had primary cancer of the oral cavity, whose outcomes are typically inferior to those of other head and neck cancers (11,24).

Treatment compliance was excellent, as the planned radiotherapy could be completed in all patients but three, in whom distant metastasis was detected during the adjuvant treatment. The proportion of patients with severe (Grade 3 or worse) acute/late adverse events was not unacceptably high, being <5% for all the toxicities investigated, except acute mucositis. Furthermore, it is also noteworthy that no severe reconstruction flap-related toxicities were encountered. In particular, of the 62 patients who underwent flap reconstruction, a free jejunal flap was used in 27 (22%) patients; free jejunal flaps are regarded as being radiosensitive, since the tolerance dose is low and vulnerability to high-dose radiation is high of their native site (small intestine) (25–27). Abdominal irradiation up to 60 Gy can cause radiation injury to the intestines, including enteritis, intestinal stricture and perforation (28). Most of the patients in the current study with free jejunal flaps had received irradiation of the entire reconstruction site at the dose of at least 60 Gy, however, no significant radiation-related toxicities were encountered. This finding is consistent with that reported by Barrett et al. (26), who reported finding no case of secondary ischemia or necrosis of the flaps in any of their patients after high-dose postoperative radiotherapy. Further large-scale or prospective studies aimed at exact analysis of the dose-volume histogram are necessary to evaluate the feasibility of high-dose postoperative therapy for those with flap reconstruction.

According to the results of the univariate analysis, the primary tumor in the oropharynx favorably influenced all the treatment outcomes analyzed; the same factor was still identified as an independent factor exerting a favorable influence on the OS even by multivariate analysis. These results are consistent with the reports from Chao and Eisbruch (21,29). Even though 10 out of the 14 patients with oropharyngeal cancer had Stage IV disease, the 3-year OS for these patients in the current study was excellent (86%; data not shown). Surgery followed by radiotherapy with or without chemotherapy is effective and might be favored for those with resectable oropharyngeal cancer, considering the excellent outcomes, as long as the patients are willing to accept treatment-related functional impairments.

The most frequent pattern of loco-regional failure in this study was in-field recurrence; marginal and out-field recurrence were not frequent, accounting for only 4% and 6% of all the patients, respectively, which was consistent with the previous findings (10,11,21–23), listed in Table 5 (7,10,11,19,21,23). The in-field recurrence rate in this study was relatively high compared to those in the previous reports (7,10,11,19,21,23). In the current study, the postoperative risk assessment revealed that 89% of the patients had the high-risk features (i.e. PSM and/or ECE positive). However, the proportions of patients with risk factors in the previous reports that listed in Table 1 were lower than those of this study, ranging from 41% to 72%. Patients with several clinicopathological factors, such as multiple lymph nodes metastases, oral cavity primary, lymphovascular invasion have also been regarded being at the risk factors for recurrence (30,31), and those with these factors may have constituted a substantial portion in the previous reports. In this study, patients with above-mentioned risk features had not been routinely indicated for postoperative IMRT. The difference of patient selection might have affected the pattern of failure of recurrence in this study. On the basis of the above findings, we conclude that IMRT in the postoperative setting did not lead to any unexpected failure patterns.

Table 5.

Patterns of loco-regional failures by previous reports

Author	No. of patients	Site (s)	Pattern of loco-regional failure			2-yr OS(%)	2-yr LRC (%)
			In-field	Marginal	Out-field
Chao et al.	74	OC, OPX, HPX	3	2	3	N.A.	90
Eisbruch et al.	73	OC, OPX, LYX HPX, UNK	10	4	0	77 (3-yr)	82 (3-yr)
Daly et al.	30	OC	7	2	2	60 (3-yr)	53 (3-yr)
Chen et al.	90	OC, OPX, LYX  HPX	11	6	0	79	80
Chan et al.	180	OC	26	7	5	65	78
Turaka et al.	28	OC, NPX, OPX LYX, HPX, PNS, NC	5	3	N.A.	76	66
Studer et al.	71	OC, OPX, LYX HPX, UNK, PNS	5	1	2	83	95
Current study	122	OC, OPX, LYX HPX, UNK	26	5	7	59 (3-yr)	71 (3-yr)

Loco-regional recurrence remains a matter of concern in treatment of postoperative head and neck cancer. Enhancing treatment intensity including dose escalation may lead to better LRC. Preoperative determination of tumors or a part of tumors that would exhibit biologically aggressive phenotype or radioresistant profile by using novel diagnostic imaging technique or novel molecular imaging such as hypoxia detecting PET would provide effective solution for reduction of in-field recurrence (32). Dose escalation confined to these sites would worth investigating, because dose escalation for the whole area of high-risk CTV could contribute to increased toxicities, and impair treatment compliance. Selection of chemotherapeutic agents or regimens is also important to obtain better treatment outcomes. Efficacy of modulating chemotherapy administration (i.e. weekly CDDP administration instead of tri-weekly) and usage of novel chemotherapy agents (i.e. molecular targeted agent such as cetuximab), focusing on aiming at reducing toxicities associated with chemotherapy, has been currently under investigation in the ongoing clinical trials (e.g. JCOG 1008 and RTOG 0920) (30). Among the five patients who developed marginal recurrence, more than a half had two clinical features in common; three patients had the primary tumor in the oral cavity and four had undergone more than two surgeries before the IMRT. In addition to the inherent resistance to radiotherapy (11,24), altered lymphovascular drainage resulting from multiple surgeries (33,34), might have contributed to the development of the marginal recurrence in these cases. Caution must be taken when delineating the CTVs in patients with these clinical features so as to reduce the risk of marginal recurrence; addition of a generous margin to CTVs might be useful. The frequent occurrence of distant metastasis, in some cases synchronously with loco-regional recurrence, reflects the biologically aggressive nature of head and neck cancer (11,35). The univariate and multivariate analysis in this study revealed that the patients with four or more pathological lymph nodes metastases and primary sites of non-oropharyngeal cancer exhibited worse DMFS rates than those without these features. Therefore, close investigation for distant metastases with tumor marker measurements and/or PET/CT imaging may facilitate identifying insidious distant metastasis for patients with high-risk clinicopathological features.

There were several limitations of this study. Potential biases, which are inherent to any retrospective analysis, could have affected the results of this study as mentioned above. However, the heterogeneous patient backgrounds and treatment procedures represent actual clinical practice of treatment for head and neck cancer patients. Since prospective trials to evaluate the treatment outcomes and toxicities are typically difficult in a field where multimodality treatment is necessary, it is considered that a retrospective study, like this one, still provides valuable information for clinical practice.

In conclusion, postoperative IMRT appears to be feasible for head and neck cancer patients, without unexpected patterns of loco-regional failure.

Funding

This study was supported in part by Grant-in-Aid for Scientific Research by The Ministry of Education, Culture, Sports, Science and Technology (JSPS KAKENHI Grant Number 16K10411). In addition, this study was also supported in part by the National Cancer Center Research and Development Fund (25-A-10).

Conflict of interest statement

None declared.