salvage re-irradiation for recurrent head and neck cancer

C

DcplrpatusCro

OA(

Int. J. Radiation Oncology Biol. Phys., Vol. 68, No. 3, pp. 731–740, 2007Copyright © 2007 Elsevier Inc.

Printed in the USA. All rights reserved0360-3016/07/$–see front matter

doi:10.1016/j.ijrobp.2006.12.055

LINICAL INVESTIGATION Head and Neck

SALVAGE RE-IRRADIATION FOR RECURRENT HEAD AND NECK CANCER

NANCY LEE, M.D.,* KELVIN CHAN, B.A.,* JUSTIN E. BEKELMAN, M.D.,* JOANNE ZHUNG, B.A.,*JAMES MECHALAKOS, PH.D.,† ASHWATHA NARAYANA, M.D.,* SUZANNE WOLDEN, M.D.,*ENNAPADAM S. VENKATRAMAN, PH.D.,‡ DAVID PFISTER, M.D.,§ DENNIS KRAUS, M.D.,¶

JATIN SHAH, M.D.,¶ AND MICHAEL J. ZELEFSKY, M.D.*

Departments of *Radiation Oncology, †Medical Physics, and ‡Biostatistics, and Divisions of §Head and Neck Oncology, Departmentof Medicine, and ¶Head and Neck Surgery, Department of Surgery, Memorial Sloan-Kettering Cancer Center, New York, NY

Purpose: To present a retrospective review of treatment outcomes for recurrent head and neck (HN) cancerpatients treated with re-irradiation (re-RT) at a single medical center.Methods and Materials: From July 1996–September 2005, 105 patients with recurrent HN cancer underwent re-RTat our institution. Sites included were: the neck (n � 21), nasopharynx (n � 21), paranasal sinus (n � 18), oropharynx(n � 16), oral cavity (n � 9), larynx (n � 10), parotid (n � 6), and hypopharynx (n � 4). The median prior RT dosewas 62 Gy. Seventy-five patients received chemotherapy with their re-RT (platinum-based in the majority of cases).The median re-RT dose was 59.4 Gy. In 74 (70%), re-RT utilized intensity-modulated radiation therapy (IMRT).Results: With a median follow-up of 35 months, 18 patients were alive with no evidence of disease. The 2-yearloco-regional progression-free survival (LRPFS) and overall survival rates were 42% and 37%, respectively.Patients who underwent IMRT, compared to those who did not, had a better 2-year LRPF (52% vs. 20%, p <0.001). On multivariate analysis, non-nasopharynx and non-IMRT were associated with an increased risk ofloco-regional (LR) failure. Patients with LR progression-free disease had better 2-year overall survival vs. thosewith LR failure (56% vs. 21%, p < 0.001). Acute and late Grade 3–4 toxicities were reported in 23% and 15%of patients. Severe Grade 3–4 late complications were observed in 12 patients, with a median time to developmentof 6 months after re-RT.Conclusions: Based on our data, achieving LR control is crucial for improved overall survival in this patientpopulation. The use of IMRT predicted better LR tumor control. Future aggressive efforts in maximizing tumorcontrol in the recurrent setting, including dose escalation with IMRT and improved chemotherapy, arewarranted. © 2007 Elsevier Inc.

Head and neck, Recurrent, Re-irradiation, IMRT.

tu4muwaamctmra

o5

INTRODUCTION

espite aggressive efforts to cure head and neck (HN)ancer, locoregional (LR) recurrence remains a seriousroblem (1, 2). Recurrent HN cancer poses a great chal-enge, particularly when patients have received prior HNadiotherapy. In general, when recurrence is detected inatients who were previously irradiated, surgical resectionnd/or postoperative radiation therapy (RT) and/or chemo-herapy (CT) are offered (3–7). When patients present withnresectable recurrent disease or are unable to undergourgery, alternative nonsurgical treatments, ranging fromT alone to re-irradiation (re-RT) with or without CT are

ecommended (8–10). Best supportive care has also been anption for patients who are medically unfit for treatment.

Reprint requests to: Nancy Lee, M.D., Department of Radiationncology, Memorial Sloan-Kettering Cancer Center, 1275 Yorkve., Box 22, New York, NY 10021. Tel: (212) 639-3341; Fax:

212) 717-3104; E-mail: [email protected]
Presented at the 48th Annual Meeting of the American Society A
731

Chemotherapy alone has traditionally been consideredhe standard of care for recurrent and previously irradiatednresectable HN cancer. Response rates between 10% and0% were observed, with a poor median survival of 5–9onths (2, 11, 12). Combination chemotherapy has been

sed, and a slight increase in 2–3-year overall survival (OS)as reported, at around 5% to 10% (2, 13, 14). The results

re suboptimal, and the majority of these patients die withctive LR disease. Even among those who die with distantetastases, a large portion of the patients still harbor un-

ontrolled disease at the primary site and suffer from in-ractable pain, bleeding, and infection. Therefore, single andulti-institutional clinical trials have focused on the use of

e-RT in combination with CT to improve disease controlnd OS for these patients (8, 10, 15, 16).

f Therapeutic Radiology and Oncology (ASTRO), November–9, 2006, Philadelphia, PA.Conflict of interest: none.Received Nov 17, 2006, and in revised form Dec 16, 2006.

ccepted for publication Dec 22, 2006.

cwBpetalatunt

wstnctuin

rtvtiicepc

P

pcbdspioh

hcanscb

R

wosatwttastctneastyco

Tisr(

R

TG

R

R

R

732 I. J. Radiation Oncology ● Biology ● Physics Volume 68, Number 3, 2007

Achieving a tumoricidal re-RT dose with or withouthemotherapy, while focusing on controlling LR disease,as used to treat inoperable recurrent HN cancer (17, 18).ecause of a fear of unacceptable late normal-tissue com-lications, different centers have limited their re-RT toither interstitial and intracavitary brachytherapy or stereo-actic radiosurgery (16, 19–22). Although brachytherapynd radiosurgery are good options, most recurrences arearge, irregularly shaped, and situated in a region that is notlways amenable to either of these procedures. Furthermore,he risks of anesthesia may preclude many patients fromndergoing brachytherapy. Due to these limitations, alter-ative re-RT delivery methods using external beam radio-herapy have been utilized.

Escalated tumoricidal doses are often difficult to achieveith conventional radiotherapy techniques for these rea-

ons: (1) Without radiation-beam modulation, conventionalechniques deliver unacceptable doses to the surroundingormal tissues, potentially resulting in severe late compli-ations; and (2) Physicians may choose to protect normalissues such as the spinal cord or brainstem at the expense ofnderdosing portions of a tumor. Therefore, strategies tomprove the delivery of external-beam radiotherapy areeeded (23, 24).With the advent of highly conformal intensity-modulated

adiation therapy (IMRT), a desired dose can be delivered tohe tumor target with minimal radiation delivery to unin-olved normal tissues (25, 26). Intensity-modulated radia-ion therapy offers highly conformal tumor coverage, allow-ng for a tumor-dose escalation which may lead tomprovement in LR control, and perhaps a reduction in lateomplications. Therefore, we undertook the current study toxplore our experience with conventional RT and IMRT inatients who underwent salvage re-RT for recurrent HNancer from 1996 to 2005.

METHODS AND MATERIALS

atient and tumor characteristicsFrom July 1996 to September 2005, 155 recurrent HN cancer

atients who received prior HN radiation presented at our clinic foronsideration for re-RT. Exclusion criteria included: use of salvagerachytherapy, performance status of �60, melanoma, RT-in-uced sarcoma, and the presence of distant metastases at time ofalvage. The present analysis consists of 105 remaining HN canceratients who underwent external-beam radiotherapy with curativentent, including both conventional RT and IMRT. Of note, onlyne patient out of 105 had new primary disease. All patients hadistologic confirmation before re-RT (Table 1).Patients underwent pretreatment evaluations consisting of a

istory and physical, complete blood count (CBC), chemistry,hest x-ray (CXR), and dental evaluation, and the following im-ging studies: computed tomography (CT) and/or magnetic reso-ance imaging (MRI) of the head and neck, and/or positron emis-ion tomography (PET) before re-irradiation. A urinalysis,reatinine clearance, ECG, and audiogram were also performed
efore treatment with CT. fi
adiation and chemotherapyPatients were immobilized with a thermoplastic HN mask and

hen necessary including the shoulders, to ensure reproducibilityf re-irradiation. Computed-tomography simulation with a 3-mmlice thickness was performed in all patients treated with IMRTnd those who received three-dimensional (3D) conformal radio-herapy. Intravenous contrast was used when indicated. Patientsere planned using the Memorial Sloan-Kettering Cancer Center

reatment-planning system with six megavoltage photons (27). Athe discretion of the treating physician, 14 patients underwentltered fractionation (12 with hyperfractionation, and two withplit-course hypofractionation), while the rest of the cohort wasreated with once-a-day radiotherapy. In general, beams werehosen to ensure that at least 95% of the dose encompassed thearget volume. Every effort was made to avoid re-RT of criticalormal tissues such as the spinal cord and brainstem. No majorfforts were made to spare the parotid glands, because patientslready complained of xerostomia before re-RT. In 74 patients,alvage radiation utilized IMRT. Of note, there was an evolution inhe use of IMRT in the treatment of recurrent HN cancer. Since theear 2000, we have routinely implemented IMRT, except in a fewases where the treating physician had not felt that IMRT would bef further benefit when compared to conventional radiotherapy.The median prior HN re-RT dose was 62 Gy (range, 28–78 Gy).

he median time from the first course of radiotherapy to re-rradiation was 38 months (range, 5–380 months). The medianalvage re-irradiation dose was 59.4 Gy (range, 30–70 Gy). Thee-RT was delivered with concurrent chemotherapy in 45 patients43%).

adiation target volumesFor unresectable tumors, gross target volume (GTV) was de-

Table 1. Patient and tumor characteristics at timeof re-irradiation

Characteristics n Percent or range

otal patients 105 100%enderMen 69 66%Women 36 34%Median age 58 31–84

ecurrent disease siteNasopharynx 21 20%Neck 21 20%Paranasal sinus 18 17%Oropharynx 16 15%Larynx 10 10%Oral cavity 9 8%Parotid 6 6%Hypopharynx 4 4%

ecurrent tumor in RT fieldPrimary site only 63 60%Neck only 20 19%Both primary and neck 22 21%

ecurrent histologySquamous-cell carcinoma 91 86%Adenoid cystic carcinoma 5 5%Mucoepidermoid

carcinoma5 5%

Adenocarcinoma 4 4%

ned as any visible evidence of disease on physical examination or

oMCcGaasopws1

S

soshngmtga

T

ttrm

F

semaa

S

cargTsOaAuumaprb(

I((tplT(S

P

mo

R

bppahG(eb

C

mcbtmctrm

R

drwdrmrArwpr

733Salvage re-irradiation for head and neck cancer ● N. LEE et al.

n any imaging modality, i.e., CAT, MRI, and/or PET scans.agnetic-resonance-imaging fusion with a treatment-planningAT scan was done when possible. In postoperative patients, thelinical tumor volume (CTV) included at least the preoperativeTV and the postoperative bed. For unresectable patients, as well

s those who underwent resection, there was no intention to treatny other subclinical sites of disease. All target volumes and theirurrounding critical normal structures were outlined slice by slicen axial CAT scans. To account for setup error and organ motion,lanning target volumes (PTVs) were added to the GTV and CTV,ith a margin ranging from 1–2 cm. In regions near critical

tructures such as the spinal cord, the PTV margin was as small asmm.

urgeryAll patients were discussed by our multidisciplinary team, con-

isting of a HN surgeon, medical oncologist, and radiation oncol-gist. When the recurrent tumor was deemed resectable, the HNurgeon performed gross total resection. Thirty-six patients (34%)ad surgery immediately before re-RT. Eleven patients had upfronteck dissections before re-RT. Among the 25 patients who hadross total resection of their recurrent disease, 7 had negativeargins, 3 had close margins, and 15 had positive margins. Al-

hough there was no gross residual disease, postoperative RT wasiven because of a high suspicion of microscopic residual diseaset time of surgery.

oxicity assessmentAcute toxicities and late complications were assessed retrospec-

ively by reviewing all charts. All symptoms and complicationshat were documented in patient records during and followinge-RT were recorded. Toxicity grading was according to the Com-on Toxicity Criteria, version 3.0 (28).

ollow-upAll patients were evaluated weekly during treatment and sub-

equently followed every 1–2 months for the first 2 years, andvery 4–6 months in subsequent years, by the radiation oncologist,edical oncologist, and/or HN surgeon. Baseline CAT and MRI

nd/or PET scans were performed typically between 2–4 monthsfter treatment.

tatistical methodsDescriptive statistics were calculated for patient and tumor

haracteristics, treatment features, and toxicities. Response ratespproximately 2 months after re-RT were determined by clinical oradiographic examination. Patients who had clinical or radio-raphic evidence of progression of disease were scored as failures.he primary endpoints were 2-year actuarial LR progression-freeurvival (LRPFS), distant metastases-free survival (DMFS), andS. Freedom from LR or distant progression was defined as an

bsence of disease on physical and radiographic examination.ctuarial estimates for LRPFS, DMFS, and OS were calculatedsing Kaplan-Meier methods. Log-rank tests were used to comparenadjusted survival estimates. We used a Cox proportional hazardsodel to examine potential prognostic factors for LRPFS, DMFS,

nd OS, including (1) age (continuous variable); (2) re-RT to therimary only vs. other; (3) aggressiveness of the disease (recur-ence followed by immediate re-RT, or recurrence more than onceefore re-RT); (4) surgery (any vs. none); (5) CT (any vs. none);
6) re-RT technique (conventional including 3D conformal vs. r
MRT); (7) re-RT dose (�50 Gy vs. �50 Gy); (8) re-RT sitenasopharynx, pharynx, or other); and (9) histology at re-RTsquamous-cell carcinoma [SCC] vs. other). Time-to-event dura-ions were measured from the first day of re-RT. Among survivingatients, all were closely followed except for six patients who wereost to follow-up because they reside outside of the United States.he median follow-up for all surviving patients was 35 months

range, 2.4–80 months). Statistical analyses were performed usingtata, version 9 (Stata Corp., College Station, TX).

RESULTS

atient characteristicsThe median patient age was 58 years. Most patients wereen, and had SCC with disease limited to the primary site

nly (Table 1).

adiationFive patients did not complete their prescribed re-RT

ecause of progression of disease, of whom four patientsrogressed locally, and one progressed distantly. All fouratients who had locally progressive disease died shortlyfter discontinuation of re-RT. One patient was found toave lung metastases, and re-RT was discontinued at 39.6y. The median cumulative radiation dose was 121.4 Gy

range, 88–137 Gy). The cumulative radiation dose deliv-red to the spinal cord was limited to 50 Gy, and to therainstem, 60 Gy (Table 2).

hemotherapyChemotherapy was given at the discretion of the treatingedical oncologist. Seventy-five patients received various

hemotherapeutic agents with their re-RT, in various com-inations (Table 2). Carboplatin-and-taxol chemotherapy ishe most frequent combination used during induction che-otherapy before re-RT. Single-agent cisplatin followed by

arboplatin and 5-fluorouracil (5FU) chemotherapy werehe most commonly used agents if given concurrently withe-RT. Finally, carboplatin and 5FU were the most com-only used agents given adjuvantly after re-RT.

e-irradiation responseBoth clinical and radiographic examinations were used to

etermine response rates, 2 months after completion ofe-RT. At the time of this analysis, 18 patients were aliveith no evidence of disease. Forty patients had no responseata available (36 patients had their measurable diseaseesected before re-RT, 2 patients died shortly after treat-ent, and 2 patients had no follow-up data). Therefore,

esponse data were only available for 65 of 105 patients.mong these, 10 achieved complete response, 7 had partial

esponse, and 17 had stable disease. The total response rateas 52% (34 of 65 patients). Thirty-one patients had LRrogressive disease that developed during or shortly aftere-RT. The total radiation dose was not related to the
esponse rates.

L

m4mD

odr

ret(ff

wrc(

A

t0pcGrv

P

R

S

C

C

C

C

Fl

F


RPFS, DMFS, and OSThe median LRPFS was 10 months (range, 0.1–77onths), with 1-year and 2-year LRPFS rates of 48% and

2%, respectively (Fig. 1). The median DMFS was 49onths (range, 0.1–77 months) with 1-year and 2-yearMFS rates of 80% and 72%, respectively.For the entire cohort, patterns of failure included: LR

nly (n � 43), distant failure only (n � 9), and both LR andistant failures (n � 14). Among those who had surgical

Table 2. Treatment details of prior radiation and re-irradiation

Treatment for 105 patients n (range)

rior radiation doseMedian and range of total dose 62 Gy (28–78)Interval between prior RT and

re-irradiation38 months (5–380)

e-irradiationMedian and range of total dose 59.4 Gy (30–70)Median and range of dose per fraction 1.8–2.0 GyConventional technique 15Simple 3D conformal 10Complex 3D conformal 6IMRT 74Median and range of cumulative total

dose121.4 Gy

Incomplete re-irradiation course 5urgerySurgery immediately prior to re-RT 36

hemotherapyPrior chemotherapy 4Concurrent chemotherapy 45Prior and concurrent chemotherapy 14Concurrent and adjuvant chemotherapy 10Prior, concurrent, and adjuvant

chemotherapy2

hemotherapeutic agents used as inductionchemotherapy prior to re-RT

Carboplatin and taxol 7Cisplatin and 5FU 3Taxol 3Carboplatin and 5FU 2Carboplatin and docetaxel 2Cisplatin and capecitabine 1Carboplatin alone 1Gemcitabine 1

hemotherapy agents used during re-RTCisplatin 24Carboplatin and 5FU 20Carboplatin and taxol 13Carboplatin 6Taxol 5Cisplatin and taxol 1Mitomycin C 15FU 1

hemotherapeutic agents used duringadjuvant phase of re-RT

Carboplatin and 5FU 6Cisplatin and 5FU 3Carboplatin and taxol 1Cisplatin 1Taxol 1

Abbreviations: 3D � three-dimensional; re-RT � re-irradiation.

esection before re-RT (n � 36), 11 patients experienced LR o

ecurrence only, 4 failed distantly only, and 6 patientsxperienced both LR and distant failures. Lastly, amonghose who did not undergo surgical resection before re-RTn � 69), 32 patients experienced LR failure only, 5 patientsailed distantly only, and 8 patients had both LR and distantailures.

The median OS was 15 months (range, 0.1–80 months),ith 1-year and 2-year survival rates of 56% and 37%,

espectively (Fig. 2). Patients with LR control had signifi-antly better 2-year OS compared to those with LR failure56% vs. 21%, p � 0.001).

nalysis of prognostic factors for LR progressionPatients who underwent IMRT, compared to non-IMRT

echniques, had better 2-year LRPFS (52% vs. 20%, p �.001; Fig. 3). As shown in Table 3, univariate analysis ofotential prognostic factors revealed that the use of IMRT,ompared to non-IMRT techniques and radiation dose �50y vs. �50 Gy, were predictive of better LRPFS (hazard

atio [HR] for LR progression, 0.36; 95% confidence inter-al [CI], 0.21–0.61; p � 0.001; and HR, 0.36; 95% CI,

ig. 1. Kaplan-Meier estimate of 1-year (56%) and 2-year (42%)ocoregional progression-free probabilities.

ig. 2. Kaplan-Meier estimate of 1-year (56%) and 2-year (37%)
verall survival probabilities.

0ybD0ibcff(

A

tIhp00tnwphm(oss

S

�2iOrb

Ff(

AR

A

S

C

R

D

S

H


.20–0.65; p � 0.001, respectively). On multivariate anal-sis, the use of IMRT techniques was still associated withetter LRPFS (HR, 0.37; 95% CI, 0.19–0.76; p � 0.006).ose was no longer significantly predictive of LRPFS (p �.13; Table 3). There was also a trend toward better LRPFSn patients who underwent surgical resection immediatelyefore re-RT. The use of chemotherapy was not signifi-antly associated with longer LRPFS. Because there were soew distant-metastasis events, no prognostic factors wereound to be significantly associated with longer DMFSanalysis not shown).

ig. 3. Kaplan-Meier estimate of 2-year locoregional progression-ree probabilities for intensity-modulated radiation therapyIMRT) vs. nonintensity-modulated radiation therapy patients.

Table 3. Univariate and multivariate analysis of prognos

Univa

n Unadjusted

ge 105 1.01 (0e-RT locationPrimary only 63 RefOther 42 1.08 (0

ggressiveness of diseaseFirst recurrence followed by re-RT 60 RefMultiple recurrences prior to re-RT 45 1.51 (0

urgery(compared to no surgery) 36 0.61 (0

hemotherapy(compared to no chemotherapy) 75 2.03 (1

T techniqueNon-IMRT 31 RefIMRT 74 0.36 (0

ose�50 Gy 21 Ref�50 Gy 84 0.36 (0

iteNasopharynx 21 RefPharynx 30 2.20 (0Other 54 1.44 (0

istologySCC 91 RefOther 14 0.52 (0

Note: Hazard ratio (HR) �1 is indicative of better LRPFS; HR �1 is

nalysis of prognostic factors for overall survivalAs shown in Table 4, univariate analysis demonstrated

hat patients who underwent IMRT, compared to non-MRT techniques, and those who had nonsquamous-cellistology, compared to squamous-cell histology, wereredictive of better OS (HR for death, 0.57; 95% CI,.35– 0.93; p � 0.03; and HR for death, 0.27; 95% CI,.11– 0.69; p � 0.01, respectively). Advancing age, pa-ients who recurred more than once before re-RT, andon-nasopharyngeal pharynx sites were predictive oforse OS. There was a trend toward better OS amongatients who received �50 Gy (p � 0.07) and those whoad surgery immediately before re-RT (p � 0.06). Onultivariate analysis, only nonsquamous-cell histology

p � 0.03) and radiation dose (p � 0.04) were predictivef better OS. Lastly, non-nasopharyngeal sites were as-ociated with a worse OS compared to nasopharyngealites on multivariate analysis (MVA) (p � 0.05).

ubset analysis of non-nasopharyngeal SCC patientsSubset analysis of non-nasopharyngeal SCC patients (n71) showed that there was significant improvement in

-year LRPFS (45% vs. 19%; p � 0.04) in favor of surgerymmediately before re-RT (Fig. 4). In addition, the 2-yearS for patients who underwent surgical resection before

e-RT, compared to those who did not, was significantlyetter (36% vs. 12%, p � 0.008).

ors for locoregional, progression-free survival (LRPFS)

alysis Multivariate analysis

5% CI) p value Adjusted HR (95% CI) p value

03) 0.632 1.01 (0.98–1.03) 0.624

Reference83) 0.786 1.29 (0.64–2.57) 0.476

Reference55) 0.123 1.06 (0.58–1.95) 0.847

08) 0.088 0.74 (0.30–1.78) 0.497

86) 0.031 2.21 (0.92–5.36) 0.078

Reference61) �0.001 0.37 (0.19–0.76) 0.006

Reference65) 0.001 0.52 (0.23–1.20) 0.127

Reference1) 0.055 2.92 (1.17–7.27) 0.02194) 0.316 1.68 (0.70–3.99) 0.243

Reference15) 0.106 0.58 (0.23–1.46) 0.246

tic fact

riate an

HR (9

.98–1.

erence.63–1.

erence.89–2.

.34–1.

.07–3.

erence.21–0.

erence.20–0.

erence.9–4.9.71–2.

erence.23–1.

indicative of worse LRPFS.

A

tpTm

L

G

cqpNin

pri1dhtaeOwp

hiwG

AR

A

S

C

R

D

S

H

is ind

Ffb


cute toxicityAcute toxicities that were more severe than expected with

he first course of radiation were not seen. Twenty-fouratients (23%) complained of Grade 3 acute toxicities.hese toxicities included tongue swelling (1), skin (5), anducositis (18).

ate toxicityLate complications are summarized in Table 5. Severe

rade 3 and 4 complications were seen in 12 patients. These

Table 4. Univariate and multivariate analys

Univa

n Unadjusted

ge 105 1.03 (1e-RT locationPrimary 63 RefOther 42 1.39 (0

ggressiveness of diseaseSingle recurrence prior to re-RT 60 RefMultiple recurrences prior to re-RT 45 1.83 (1

urgery(compared to no surgery) 36 0.61 (0

hemotherapy(compared to no chemotherapy) 75 1.77 (1

T TechniqueNon-IMRT 31 RefIMRT 74 0.57 (0

ose�50 Gy 21 Ref�50 Gy 84 0.60 (0

iteNasopharynx 21 RefPharynx 30 4.35 (2Other 54 1.97 (0

istologySCC 91 RefOther 14 0.27 (0

Note: Hazard ratio (HR) �1 is indicative of better OS; HR �1

ig. 4. Kaplan-Meier estimate of 2-year locoregional progression-ree probabilities for patients with and without surgical resection

cefore re-irradiation.

omplications included trismus, stricture and dysphagia re-uiring feeding-tube dependence, cranial neuropathy, tem-oral-lobe necrosis, hearing loss, and unilateral blindness.eck fibrosis was commonly reported; however, the major-

ty of these patients already experienced moderate subcuta-eous fibrosis due to their first course of radiation.

DISCUSSION

Re-irradiation using external-beam methods has been em-loyed by multiple institutions. Reported LR control ratesange from 25% to 60%, while OS ranges from 15% to 93%n series where re-RT was used as a treatment modality (2,0, 17, 18, 29, 30). Many factors could contribute to theifferences in reported results, such as patient selection,istology at recurrence, and whether surgery and/or chemo-herapy were given with re-RT. In addition, different radi-tion techniques can also contribute to the observed differ-nces. We observed a 2-year LRPFS of 42%, and a 2-yearS of 37%, in our series. We further observed that patientsith LR control had significantly better 2-year OS com-ared to those with LR failure (56% vs. 21%, p � 0.001).

Given the promising re-RT results when compared toistorical control results with chemotherapy, several multi-nstitutional studies also focused on re-RT in combinationith chemotherapy. The Radiation Therapy Oncologyroup (RTOG) conducted a Phase II trial where a split-

rognostic factors for overall survival (OS)

alysis Multivariate analysis

5% CI) p value Adjusted HR (95% CI) p value

05) 0.003 1.02 (0.997–1.04) 0.094

Reference25) 0.177 1.69 (0.92–3.08) 0.088

Reference99) 0.016 1.11 (0.63–1.97) 0.72

03) 0.064 0.73 (0.34–1.54) 0.404

11) 0.046 1.83 (0.86–3.89) 0.114

Reference93) 0.026 0.68 (0.35–1.32) 0.252

Reference05) 0.073 0.44 (0.20–0.98) 0.043

Reference47) �0.001 4.44 (1.84–10.73) 0.00110) 0.071 2.42 (1.04–5.63) 0.04

Reference69) 0.006 0.32 (0.11–0.88) 0.027

icative of worse OS.

is of p

riate an

HR (9

.01–1.

erence.86–2.

erence.12–2.

.36–1.

.01–3.

erence.35–0.

erence.34–1.

erence.00–9..94–4.

erence.11–0.

ourse re-RT (60 Gy) and concurrent chemotherapy (5FU

aSwPmms5smtsnTctaarpH

t

Crpsoatt3bs(

cqcfrtpshscstria

ceraswbamtgsspdjitmsa

otrc

G

G

G

G


nd hydroxyurea) were given to inoperable, recurrent HNCC (31). The investigators reported a 2-year OS of 16.9%,ith a median follow-up of 22.7 months. A second RTOGhase II trial also employed a split-course re-RT and che-otherapy, but with taxol and cisplatin. The update with aedian follow-up of 23.6 months showed a 1-year progres-

ion-free survival of 35%, and 1-year and 2-year OS rates of0.2% and 25.9%, respectively (32). The results from thesetudies showed the importance of the addition of a localodality such as re-RT to improve survival. However, as

here are inherent limitations in both retrospective and pro-pective Phase I and II studies, a randomized study isecessary to definitively answer this important question.herefore, the RTOG has undertaken a Phase III trial,omparing the above split-course re-RT and cisplatin andaxol chemotherapy vs. cisplatin and taxol chemotherapylone for inoperable patients, where the chemotherapy-lone arm is the standard treatment. If the results of thisandomized trial are positive, then re-RT should be incor-orated into the standard treatment for recurrent inoperableN cancer.For recurrent, operable HN cancer, salvage surgery is the

Table 5. Late complications following re-irradiation

Grade and type ofcomplication Number of patients

rade 1 complicationsDysphagia 3Dry eye syndrome 1Hoarseness 4Neuropathy 1Tinnitus 2Temporal-lobe edema 1Trismus 4

rade 2 complicationsCranial neuropathy 4Dry eye 1Dysphagia 10Eye dysfunction 4Hearing 1Neck fibrosis Common (mostly due to

first course of RT)Ophthalmoplegia 1Radiation osteitis 1Stricture 3Temporal-lobe necrosis 1Trismus 7

rade 3 complicationsNeck fibrosis Common (mostly due to

first course of RT)Cranial neuropathy 1Hearing loss 3Dysphagia 3Stricture 1Temporal-lobe necrosis 1Trismus 3

rade 4 complicationsUnilateral blindness 1Temporal-lobe necrosis 3

reatment of choice (4, 7). Data from the University of b

hicago demonstrated that surgery before re-RT and CT forecurrent squamous-cell HN cancer was an independentrognostic factor predicting for LRPFS, progression-freeurvival (PFS), and OS compared to those who were notperated upon (17). Other series also validated these results,nd reported 3–4-year OS ranging from 40–60% for pa-ients who underwent surgical resection vs. unresected pa-ients who had a 4–5-year OS rate between 10–17% (31,3–35). Similarly, our study showed that the 2-year OS wasetter in non-nasopharyngeal SCC patients who underwenturgery before re-RT (36%) vs. the inoperable cohort (12%)p � 0.001).

Given the important role of surgery in recurrent HNancer reported by multiple institutions, the next logicaluestion is whether adjuvant therapy after surgical resectionan further improve treatment outcomes. A randomized trialrom France was designed to answer this question, and theesults were recently reported in abstract form (36). In thisrial, recurrent, resectable HN patients who had receivedrevious radiation were randomized to surgery alone vs.urgery followed by postoperative chemotherapy (5FU andydroxyurea) and radiation (60 Gy). Preliminary resultshow that PFS significantly improved in patients who re-eived postoperative chemoradiotherapy (p � 0.01). Notatistically significant difference in OS was observed be-ween the two groups. Acute mucositis (Grade �3) waseported in 29% of patients, and a moderate and acceptablencrease in Grade �3 late complications (trismus, mucosa,nd fibrosis) was also observed.

There has been growing interest in the addition of con-urrent chemotherapy to re-RT for both operable and inop-rable recurrent HN cancer patients who received prioradiation (2, 15, 17, 30, 37). Various chemotherapeuticgents have been used in different combinations. Our owneries used a mixture of different agents, and chemotherapyas typically offered to those patients who were shown toe at a very high risk of LR recurrence. Other factors, suchs a patient’s underlying medical condition to tolerate che-otherapy, were also used to determine the use of chemo-

herapy. Concurrent chemotherapy and re-RT were onlyiven to 43% of the patients in our series. Because of thetringent selection factors for the use of chemotherapy astated above, it is not surprising that chemotherapy did notredict for improved outcome in our series. In addition, theifferent combinations (prior and/or concurrent and/or ad-uvant) of different chemotherapy agents given with re-RTn our series limited the ability to detect any significance inhis retrospective review. Given the available data on che-otherapy’s role to enhance radiotherapy, a prospective

ingle-institution trial on the use of concurrent chemother-py and re-RT is underway.

The delivery of re-RT using conventional techniques isften limited by the tolerance of surrounding critical struc-ures. The current recommendation, based on the RTOGe-RT trials, limits the combined spinal-cord dose (first-ourse RT and re-RT) to �50 Gy (31, 32). The major
enefit of using an IMRT technique over conventional ra-

dwsacbAGFsntwpda

(tdbptfartsb

owwwOpwaw1Gr(c

clrsi4bwippiIntivlparLmwm

oa

Fnbvcl


iation portals in re-RT is to allow superior target coverage,hile preserving the ability to spare adjacent normal tissues

uch as the spinal cord. Few studies have focused on IMRTs a salvage radiotherapy technique to treat recurrent HNancer (23, 24). Our series showed that IMRT predicted foretter LRPFS when compared to conventional radiotherapy.recurrent oropharyngeal-cancer patient (prior RT dose, 70y) who was treated with IMRT for his re-RT is shown inig. 5. The patient was found to have biopsy-proven, unre-ectable, recurrent disease involving the retropharyngealodal region and base of the skull. Therefore, we elected toreat the patient with re-RT, using IMRT to 70 Gy. Thereas excellent coverage where the plan was able to meet therescription dose of 70 Gy (mean dose, 74 Gy; maximumose, 78 Gy). The maximum point doses to the brainstem

ig. 5. Example of a patient with recurrent oropharyngeal carci-oma who underwent intensity-modulated radiation therapy to thease of the skull to a total dose of 70 Gy. (A) Axial view; (B) doseolume histogram. Notice that the 70 Gy isodose line (orange)overs the tumor target while sparing the brain stem (blue isodoseine).

nd spinal cord were 9 Gy (prior dose, 50 Gy) and 8 Gy s

prior dose, 40 Gy), respectively. Treatment with conven-ional radiotherapy would have resulted in significant un-erdose to the gross tumor while attempting to protect therainstem and spinal cord. Of note, although IMRT didredict for better LRPFS on MVA, one must keep in mindhe intricate association of IMRT with radiation dose. Theact that IMRT resulted in better LRPFS could also bettributed in part to its ability to deliver a higher tumoricidaladiation dose. Dose-escalation studies using IMRT in pa-ients who received prior HN radiotherapy are being con-idered, because a dose �50 Gy was predictive foretter OS.Nasopharyngeal carcinoma was another prognostic factor

n MVA that was associated with improved LRPFS and OShen compared to tumors from the pharynx or other sitesithin the head and neck. Others reported similar findings,ith local control rates ranging from 15–60%, and a 5-yearS rate up to 60% (22, 24, 38, 39). One recent study on 31atients with locally recurrent nasopharyngeal carcinomaho underwent re-RT using IMRT reported 1-year LR PFS

nd OS rates of 56% and 63%, respectively (23). Patientsith rT1–3 disease had an excellent local control rate of00% vs. 35% in patients with recurrent T4 (rT4) tumors.rade 3 late toxicities, mostly ototoxicity and cranial neu-

opathy, occurred in six (19%) of the patients. Chua et al.23) reported a 25% Grade 3 1-year actuarial late-compli-ation rate.

Because of the high cumulative doses of radiation,hronic toxicities as a result of re-RT should not be over-ooked. Late complications due to re-RT up to 30% wereeported, which range from mild trismus or neck fibrosis toevere complications such as carotid hemorrhage resultingn death (8, 17, 18). In our series, we observed a 4% Grade

late-complication rate, where one patient had unilaterallindness, and three had temporal-lobe necrosis. The patientith unilateral blindness presented with recurrent cancer

nvolving the paranasal sinuses, and was treated with op-osed lateral fields at time of recurrence. The other threeatients with temporal-lobe necrosis had extensive diseasenvolving the base of skull. All three patients underwentMRT. Two patients had recurrent nasopharyngeal carci-oma, while one had recurrent adenoidcystic carcinoma ofhe hard palate. Due to the extensive recurrent diseasenvolving the base of the skull for these three patients, it wasery difficult to minimize the dose delivered to the temporalobes, even with the use of IMRT. Alternative strategies,erhaps with the selection of different radiation-beam anglesnd minimizing the PTV margin with the help of image-guidedadiotherapy, should be pursued to reduce these complications.astly, intensity-modulated proton therapy may further mini-ize late complications in patients with recurrent HN cancerho received prior HN radiation. Trials are being planned inultiple centers with this technology.Our study has several important limitations. (1) It is based

n a nonrandomized heterogeneous cohort of patients fromsingle institution; therefore, definitive causal inferences

hould not be drawn. (2) Because toxicities were retrospec-

tswmllst

aptt

1

1

1

1

1


ively determined based on chart reviews rather than pro-pectively collected, it is possible that not all complicationsere captured. (3) With our median follow-up of 35onths, further follow-up is needed to fully assess the

ong-term complications associated with re-RT. Nonethe-ess, the strength of our study is that it is one of the largesttudies to examine the role of external-beam radiotherapy in
he treatment of recurrent HN cancer. i
REFEREN

J Clin Oncol 2005;23:3562–3567.

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3

CONCLUSIONS

Based on this single institution’s retrospective review,chieving LR control is crucial for improved OS in this patientopulation. The use of IMRT predicted for better tumor con-rol. Future aggressive efforts in maximizing tumor control inhe recurrent setting, including dose escalation with IMRT and
mproved chemotherapy, are warranted.
CES

1. Brockstein B, Haraf DJ, Rademaker AW, et al. Patterns offailure, prognostic factors and survival in locoregionally ad-vanced head and neck cancer treated with concomitant che-moradiotherapy: A 9-year, 337-patient, multi-institutional ex-perience. Ann Oncol 2004;15:1179–1186.

2. Wong SJ, Machtay M, Li Y. Locally recurrent, previouslyirradiated head and neck cancer: Concurrent re-irradiation andchemotherapy, or chemotherapy alone? J Clin Oncol 2006;24:2653–2658.

3. Wong LY, Wei WI, Lam LK, et al. Salvage of recurrent headand neck squamous cell carcinoma after primary curativesurgery. Head Neck 2003;25:953–959.

4. McLaughlin MP, Parsons JT, Fein DA, et al. Salvage surgeryafter radiotherapy failure in T1–T2 squamous cell carcinomaof the glottic larynx. Head Neck 1996;18:229–235.

5. Agra IM, Carvalho AL, Ulbrich FS, et al. Prognostic factors insalvage surgery for recurrent oral and oropharyngeal cancer.Head Neck 2006;28:107–113.

6. Agra IM, Carvalho AL, Pontes E, et al. Postoperative com-plications after en bloc salvage surgery for head and neckcancer. Arch Otolaryngol Head Neck Surg 2003;129:1317–1321.

7. Parsons JT, Mendenhall WM, Stringer SP, et al. Salvagesurgery following radiation failure in squamous cell carci-noma of the supraglottic larynx. Int J Radiat Oncol Biol Phys1995;32:605–609.

8. Creak AL, Harrington K, Nutting C. Treatment of recurrenthead and neck cancer: Re-irradiation or chemotherapy? ClinOncol (R Coll Radiol) 2005;17:138–147.

9. Stewart FA. Re-treatment after full-course radiotherapy: Is it aviable option? Acta Oncol 1999;38:855–862.

0. Kasperts N, Slotman B, Leemans CR, et al. A review onre-irradiation for recurrent and second primary head and neckcancer. Oral Oncol 2005;41:225–243.

1. Forastiere AA, Metch B, Schuller DE, et al. Randomizedcomparison of cisplatin plus fluorouracil and carboplatin plusfluorouracil versus methotrexate in advanced squamous-cellcarcinoma of the head and neck: A Southwest OncologyGroup study. J Clin Oncol 1992;10:1245–1251.

2. Forastiere AA, Shank D, Neuberg D, et al. Final report of aphase II evaluation of paclitaxel in patients with advancedsquamous cell carcinoma of the head and neck: An EasternCooperative Oncology Group trial (PA390). Cancer 1998;82:2270–2274.

3. Forastiere AA, Leong T, Rowinsky E, et al. Phase III com-parison of high-dose paclitaxel � cisplatin � granulocytecolony-stimulating factor versus low-dose paclitaxel � cispla-tin in advanced head and neck cancer: Eastern CooperativeOncology Group Study E1393. J Clin Oncol 2001;19:1088–1095.

4. Gibson MK, Li Y, Murphy B, et al. Randomized phase IIIevaluation of cisplatin plus fluorouracil versus cisplatin pluspaclitaxel in advanced head and neck cancer (E1395): Anintergroup trial of the Eastern Cooperative Oncology Group.

5. Haraf DJ, Weichselbaum RR, Vokes EE. Re-irradiation withconcomitant chemotherapy of unresectable recurrent head andneck cancer: A potentially curable disease. Ann Oncol 1996;7:913–918.

6. De Crevoisier R, Bourhis J, Eschwege F. Modified fraction-ated radiotherapy in head and neck squamous cell carcinoma(HNSCC) and re-irradiation in recurrent head and neck carci-nomas. Cancer Treat Res 2003;114:199–212.

7. Salama JK, Vokes EE, Chmura SJ, et al. Long-term outcomeof concurrent chemotherapy and reirradiation for recurrent andsecond primary head-and-neck squamous cell carcinoma. Int JRadiat Oncol Biol Phys 2006;64:382–391.

8. Dawson LA, Myers LL, Bradford CR, et al. Conformal re-irradiation of recurrent and new primary head-and-neck can-cer. Int J Radiat Oncol Biol Phys 2001;50:377–385.

9. Puthawala AA, Syed AM. Interstitial re-irradiation for recur-rent and/or persistent head and neck cancers. Int J RadiatOncol Biol Phys 1987;13:1113–1114.

0. Syed AM, Puthawala AA, Damore SJ, et al. Brachytherapy forprimary and recurrent nasopharyngeal carcinoma: 20 years’experience at Long Beach Memorial. Int J Radiat Oncol BiolPhys 2000;47:1311–1321.

1. Tan EH, Adelstein DJ, Saxton JP, et al. Concurrent chemo-radiotherapy for salvage in relapsed squamous cell head andneck cancer. Cancer Invest 1997;15:422–428.

2. Chua DT, Sham JS, Hung KN, et al. Salvage treatment forpersistent and recurrent T1–2 nasopharyngeal carcinoma bystereotactic radiosurgery. Head Neck 2001;23:791–798.

3. Chua DT, Sham JS, Leung LH, et al. Re-irradiation of naso-pharyngeal carcinoma with intensity-modulated radiotherapy.Radiother Oncol 2005;77:290–294.

4. Chen YJ, Kuo JV, Ramsinghani NS, et al. Intensity-modulatedradiotherapy for previously irradiated, recurrent head-and-neck cancer. Med Dosim 2002;27:171–176.

5. Xia P, Verhey LJ. Delivery systems of intensity-modulatedradiotherapy using conventional multileaf collimators. MedDosim 2001;26:169–177.

6. Lee N, Xia P, Quivey JM, et al. Intensity-modulated radio-therapy in the treatment of nasopharyngeal carcinoma: Anupdate of the UCSF experience. Int J Radiat Oncol Biol Phys2002;53:12–22.

7. Chui CS, Spirou SV. Inverse planning algorithms for externalbeam radiation therapy. Med Dosim 2001;26:189–197.

8. Services U.S. Department of Health. Common terminologycriteria for adverse events (CTCAE): version 3.0. In: NationalInstitutes of Health National Cancer Institute, editor; 2003.

9. Wang CC, McIntyre J. Re-irradiation of laryngeal carci-noma—Techniques and results. Int J Radiat Oncol Biol Phys1993;26:783–785.

0. Kramer NM, Horwitz EM, Cheng J, et al. Toxicity and out-come analysis of patients with recurrent head and neck cancertreated with hyperfractionated split-course reirradiation andconcurrent cisplatin and paclitaxel chemotherapy from twoprospective phase I and II studies. Head Neck 2005;27:406–
414.

3

3

3

3

3

3

3

3

3


1. Spencer SA, Harris J, Wheeler RH, et al. RTOG 96–10:reirradiation with concurrent hydroxyurea and 5-fluorouracilin patients with squamous cell cancer of the head and neck. IntJ Radiat Oncol Biol Phys 2001;51:1299–1304.

2. Horowitz EM, Harris J, Langer CJ, et al. Concurrent splitcourse hyperfractionated radiotherapy, cisplatin, and pacli-taxel in patients with recurrent, previously irradiated squa-mous cell carcinoma of the head and neck: Update of RTOG9911. J Clin Oncol 2005;23:519s.

3. De Crevoisier R, Domenge C, Wibault P, et al. Full dosereirradiation combined with chemotherapy after salvage sur-gery in head and neck carcinoma. Cancer 2001;91:2071–2076.

4. De Crevoisier R, Bourhis J, Domenge C, et al. Full-dosereirradiation for unresectable head and neck carcinoma: Ex-perience at the Gustave-Roussy Institute in a series of 169
patients. J Clin Oncol 1998;16:3556–3562.
5. Machtay M, Rosenthal DI, Chalian AA, et al. Pilot study ofpostoperative reirradiation, chemotherapy, and amifostine af-ter surgical salvage for recurrent head-and-neck cancer. Int JRadiat Oncol Biol Phys 2004;59:72–77.

6. Janot F, De Raucourt D, Castaing M, et al. Re-irradiationcombined with chemotherapy after salvage surgery in headand neck carcinoma: A randomized trial from the GETTECand GORTEC groups. J Clin Oncol 2006;24:18s.

7. Spencer SA, Wheeler RH, Peters GE, et al. Concomitantchemotherapy and reirradiation as management for recurrentcancer of the head and neck. Am J Clin Oncol 1999;22:1–5.

8. Wang CC. Re-irradiation of recurrent nasopharyngeal carci-noma—Treatment techniques and results. Int J Radiat OncolBiol Phys 1987;13:953–956.

9. Hwang JM, Fu KK, Phillips TL. Results and prognostic fac-tors in the retreatment of locally recurrent nasopharyngeal
carcinoma. Int J Radiat Oncol Biol Phys 1998;41:1099–1111.

salvage re-irradiation for recurrent head and neck cancer

Documents