Patterns of failure in anaplastic and differentiated thyroid carcinoma treated with intensity-modulated radiotherapy

Background The radiotherapy (rt) volumes in anaplastic (atc) and differentiated thyroid carcinoma (dtc) are controversial. Methods We retrospectively examined the patterns of failure after postoperative intensity-modulated rt for atc and dtc. Computed tomography images were rigidly registered with the original rt plans. Recurrences were considered in-field if more than 95% of the recurrence volume received 95% of the prescribed dose, out-of-field if less than 20% received 95% of the dose, and marginal otherwise. Results Of 30 dtc patients, 4 developed regional recurrence: 1 being in-field (level iii), and 3 being out-of-field (all level ii). Of 5 atc patients, all 5 recurred at 7 sites: 2 recurrences being local, and 5 being regional [2 marginal (intramuscular to the digastric and sternocleidomastoid), 3 out-of-field (retropharyngeal, soft tissues near the manubrium, and lateral to the sternocleidomastoid)]. Conclusions In dtc, locoregional recurrence is unusual after rt. Out-of-field dtc recurrences infrequently occurred in level ii. Enlarged treatment volumes to level ii must be balanced against a potentially greater risk of toxicity.


INTRODUCTION
External-beam radiotherapy (rt) is an established treatment modality in the management of anaplastic thyroid cancer (atc).These undifferentiated tumours are extremely aggressive, recurring both locally and distantly.Radiotherapy decreases local recurrence, reduces the morbidity associated with atc progression, and might increase cancer-specific survival [1][2][3][4][5] .
Uncontrolled locoregional disease is a major cause of morbidity and mortality in dtc 13 .
Despite several studies investigating the role of rt in dtc and atc, no clear consensus has been reached concerning the optimal rt volume, and institutional practice varies.At the Princess Margaret Cancer Centre, the standard is to include levels iii-vi, with levels v and ii partially included, in the clinical target volume for both atc and dtc.Larger volumes are described in the literature, especially for atc (including in the currently open Radiation Therapy Oncology Group 0912 trial), often extending from the mastoid to the carina 6,7,14,15 .In the case of atc, the use of larger fields must be balanced against the potential for increased acute and late toxicities in this population with a very poor prognosis.
Surgery to achieve complete resection of disease is occasionally possible for atc, particularly when it is a smaller component within a tumour mass consisting of dtc.For dtc, the philosophy at Princess Margaret has been to control disease in the postoperative thyroid bed, especially in the region of the larynx and tracheoesophageal groove, wherein salvage surgery would involve extensive procedures such as laryngectomy.We consider that nodal recurrences of dtc can usually be surgically salvaged.However, in the intensity-modulated rt (imrt) era, the locoregional pattern of failure after rt has not been well established, and it is possible that treating a larger volume could prevent additional recurrences in both atc and high-risk dtc.
The purpose of the present study was to determine the patterns of failure after postoperative imrt in atc and dtc patients and to determine the location of those failures in relation to the rt volume.

Study Design and Patient Population
A research ethics board-approved retrospective chart review was performed for all patients diagnosed with atc or dtc from 2006 to 2012.That period was chosen to limit the review to imrt-treated patients.Patients who were treated with postoperative rt for atc and dtc were included.Patients who received 40 Gy in 16 fractions-and higher cumulative doses-were included.The adjuvant dose for dtc patients was 60-66 Gy in 30-33 fractions to the highrisk volume and 54-56 Gy in 30-33 fractions to lower-risk volumes.Patients with atc were treated with a range of fractionations, including 60 Gy in 40 fractions delivered twice daily in 1.5 Gy fractions 16 .We excluded patients who received primary rt, patients treated for recurrent disease, and patients who received split-course 40 Gy in 10 fractions or other palliative regimens (n = 245).All pathology was centrally reviewed at our institution.
All patients underwent computed tomography (ct) simulation with a thermoplastic mask for immobilization.Radiation treatment plans were created using the Pinnacle treatment planning system (Philips Medical Systems, Andover, MA, U.S.A.).All patients were treated using an imrt technique.The rt volumes routinely included the thyroidectomy bed (levels iii-vi, with levels ii and v partially included) for both atc and dtc 17 .
Follow-up was at the discretion of the treating physician.Patients with dtc were seen as necessary until acute toxicity resolved, and then every 6 months for 5 years, with care usually being shared between the surgeon and other members of the team.Patients with dtc routinely received additional therapy with radioactive iodine.Thyroglobulin levels were routinely assessed, and in dtc patients, we aimed for suppression of thyroid-stimulating hormone.Toxicity was reported by retrospective review and was graded according to the Common Terminology Criteria for Adverse Events, version 4.0.

Recurrence Analysis
Local recurrences were defined as occurring within the thyroid bed.Regional recurrence was defined as occurring in the lymph node regions or soft tissues of the neck or the upper mediastinum.
Computed tomography imaging demonstrating local or regional recurrence was rigidly registered with the original planning CT imaging.Priority was given to accuracy of registration in the area of the recurrence.We performed separate registrations in cases in which 2 recurrence volumes were identified at a significant distance from one another.The recurrence volumes were then contoured, with the original dose distribution overlaid.The dose delivered to the area of the recurrence volume was calculated.In instances in which the recurrence volume extended outside the limit of the original patient volume, the recurrence volume was limited to the boundaries of the original ct images.The dose to that truncated volume was calculated.
Recurrences were defined as in-field if more than 95% of the recurrence received 95% of the dose; marginal, if 95%-20% received 95% of the dose; and out-of-field, if less than 20% received 95% of the dose 18 .When a clinical target volume was created, with a lower dose prescribed to elective nodal volumes (typically 54-56 Gy in 30 or 33 fractions), the 95% isodose of that dose was used in the analysis for nodal recurrence.

Statistical Analysis
Pat ient demog raphics a nd disease a nd t reat ment characteristics are summarized using descriptive statistics.Time-to-event statistics for overall survival (os), locoregional recurrence (lrr), and distant recurrence (dr) were calculated from the date of diagnosis to the event date (death for os, local or regional recurrence for lrr, and distant recurrence for dr) or to the last date of follow-up.The Kaplan-Meier method was used to calculate os.The cumulative incidence for lrr and dr was calculated using the competing-risks approach.

RESULTS
We identified 30 patients with dtc and 5 with atc who completed postoperative rt.The results are presented here by histologic type.

Patient, Tumour, and Treatment Characteristics
All 5 of the identified patients had atc within a larger non-anaplastic tumour.The anaplastic component was detected either preoperatively by fine-needle aspiration or within the final thyroidectomy specimen.Table i presents the patient, tumour, and treatment characteristics.

Patterns of LRR
All 5 patients with atc experienced recurrence after rt.Table ii sets out the surgical and rt treatment details.There were 7 recurrence volumes: 2 local and 5 regional.Both local recurrences were in-field.Of the regional recurrences, 2 were marginal (1 intramuscular to the anterior belly of the digastric, 1 in the lateral aspect of the sternocleidomastoid at the level of the cricoid), and 3 were out-of-field (1 retropharyngeal, 1 lateral to the sternocleidomastoid at the level of the hyoid, and 1 in the soft tissues anterior to the upper manubrium).Median time from completion of rt to lrr was 8.  disease, and clinical deterioration, no patient underwent salvage surgery for recurrence.

Toxicity
Of the 5 patients, 3 experienced acute grade 3 esophagitis.For nutritional support, 1 patient required a gastrostomy tube which was inserted before rt and remained in situ.In 1 patient who required it, a tracheostomy tube was inserted after surgery and removed before the start of rt.No grade 3 or greater dermatitis was reported.

Patient, Tumour, and Treatment Characteristics
All 30 dtc patients had tumours with papillary histology.The rt dose in 4 patients (13%) was 60 Gy in 30 fractions; in the remaining 26 patients (87%), it was 66 Gy in 33 fractions.The radioactive iodine dose was available in all but 1 of the 25 patients who received it.Most patients had positive margins, extrathyroid extension, and involved lymph nodes.

Patterns of LRR
Of the dtc patients, 4 experienced regional recurrence; there were no local recurrences.Table ii sets out the surgical and rt details for those patients.Of the regional recurrences, 1 was in-field in level iii, and 3 were out-of-field, all in level ii.Of the level ii recurrences, 2 were found at the level of the hyoid, and 1, mid-way between the hyoid and the C1 transverse process.There were no marginal recurrences.
Median time from completion of rt to recurrence was 29 months (range: 2-52 months).Of the 4 patients with lrr, 2 underwent salvage neck dissection.The others had progressive metastatic disease and were unsuitable for salvage surgery.

Toxicity
Acute grade 3 radiation dermatitis was reported in 1 of the 30 patients.Acute grade 3 esophagitis was reported in 4 patients, and grade 4 esophagitis, in 1 patient.Late grade 3

DISCUSSION
The optimal volume for external-beam radiation in the imrt era for atc and dtc is a significant clinical challenge.Little published evidence is available to justify what might be considered optimal rt volumes in the treatment of atc and dtc.Several studies have investigated the role of rt in both atc and dtc, but the rt volume is not always described, and no randomized comparisons have been published.In addition, because of the rarity of disease presentations requiring consideration of rt, many of the studies span decades and do not address modern imrt techniques.

ATC
Radiation volumes for atc frequently include the entire neck and the superior mediastinum 19,20 .In a series from Germany, the field borders most commonly extended from the mastoid to the tracheal bifurcation 21 .The Radiation Therapy Oncology Group 0912 trial (https://www.rtog.org/ClinicalTrials/ProtocolTable/StudyDetails.aspx?study=0912), a randomized study currently accruing patients with atc, mandates generous coverage of the tracheoesophageal groove, levels ii-vi, the upper mediastinum to the level of the carina, plus level i and the retropharyngeal nodes at the discretion of the treating physician.
When treating patients with atc, the primary goal of rt is local control to prevent the morbidity and mortality associated with disease recurrence and progression.Patients with atc are at risk of dying of uncontrolled local disease even in the presence of distant metastasis 22 .At Princess Margaret, policy has been to limit the volume to the thyroid bed and adjacent nodal regions, including levels iii, iv, and via-b, with parts of level v (the nodal levels considered at our institution to be most at risk for spread in both the adjuvant and primary rt setting), but omitting level ii and the superior mediastinum 23 .Extending the treated volume superiorly might increase the radiation dose to normal structures such as the parotids, mandible, and submandibular glands, increasing the risk of toxicity.An attempt has been made to balance the need for local control against the potential for severe toxicity in this group of patients with generally very poor outcomes.
In a study of 39 patients treated to a median total dose of 50 Gy, with a larger volume (as in Radiation Therapy Oncology Group 0912), 28% of patients developed grade 3 or greater dysphagia requiring a feeding tube or parenteral nutrition, 18% needed emergency tracheostomy, and almost half experienced recurrent laryngeal nerve palsy 21 .Using imrt and a smaller volume, we found that, of 5 patients, 1 required a gastrostomy, and 1, a tracheostomy.Acute grade 3 esophagitis developed in 3 patients, including in 1 patient who received hyperfractionated twice-daily rt.In the case of rapidly dividing tumours such as atc, hyperfractionated rt might have improved activity without increased toxicity.In a previous report from Princess Margaret, 9 patients with unresectable atc received twice-daily rt to 60 Gy.Toxicity was found to be acceptable, although the treatment was reserved for patients with a good performance status 23 .In a larger study of hyperfractionated rt from the Royal Marsden Hospital, acute toxicity was found to be unacceptably high, with grade 3 or greater esophagitis occurring in 79% of patients, leading to discontinuation of the treatment regimen 19 .However, only a small proportion of their patients received postoperative rt.Whether hyperfractionated rt in the adjuvant setting improves outcomes for atc patients remains unknown.
The marginal and out-of-field atc recurrences identified in the present study appeared in the tissues lateral to the sternocleidomastoid, intramuscularly in the digastric, in the retropharyngeal nodes, and in soft tissue near the manubrium.To routinely cover all those structures in the clinical target volume would result in a significantly larger volume and would probably result in unacceptable toxicity.The location of the observed recurrences suggests that the pattern of spread is not an orderly progression through nodal echelons, but an aggressive pattern that is not entirely predictable.

DTC
Patients with dtc are treated using a volume similar to that described for atc.In general, recurrences in nodal regions are deemed to be amenable to salvage surgery.The use of rt to the whole cervical lymph node region is reserved for patients at high risk of relapse who have extranodal extension or multiple nodal recurrences despite surgical neck dissections and radioactive iodine treatment.
Published reports about the pattern of nodal involvement at surger y, or the pattern of failure after surgery alone, often combine nodal levels ii-iv [24][25][26][27][28] .It is therefore not possible to tailor rt volumes on the basis of those studies.More recently, Kruijiff et al. 29 examined the location of 94 postoperative recurrences in 1183 patients with papillary thyroid carcinoma by individual nodal levels in the neck.They found that 12% recurred in level ii; 18%, in level iii; 18%, in level iv; 17%. in level v; 32%, in level vi; and 2%, in the superior mediastinum.Such detailed pattern-of-failure analyses after rt are sparse.Azrif et al. 15 reported the patterns of failure after rt in 49 patients treated with either non-coplanar lateral fields (with the superior border above the hyoid bone) or with anterior-posterior parallel opposed portal fields (with a superior border at the tip of the mastoid process).Superior mediastinal nodes were not included.Patients experienced 4 upper mediastinal, 8 nodal, and 6 local recurrences.Because the target volumes were defined using orthogonal simulator fields, no analysis of the location of the nodal recurrences with respect to the dose distribution was possible.However, the 4 superior mediastinal recurrences were likely out-of-field, given that treatment to that area was purposefully omitted.Kim et al. 14 reported on 11 dtc patients treated with rt limited to the operative bed or gross relapse, and 12 patients who received additional elective rt to the cervical and mediastinal lymph nodes.Out-of-field recurrences were defined broadly as "outside the rt field".There were 7 lrrs in the limited-rt group (6 out-of-field, 1 in-field) and only 1 lrr (in-field) in the elective-rt group.In another report from Korea, 3 of 6 recurrences were situated in level ii, and all were reported as in-field, although no dosimetric analysis was performed 30 .In our pattern-of-failure analysis under imrt, all 3 out-of-field dtc recurrences occurred in level ii.
The toxicity from a large rt volume for dtc was recorded in a multicentric study of three-dimensional conformal rt that prescribed 50.4-54Gy to lymph nodes from the mastoid to the tracheal bifurcation.Acute grade 3 toxicity was reported in only 14% of patients (3 of 22), and all incidences had resolved at first follow-up 31 .Kim et al. recorded grade 3 or greater toxicity in only 8% (1 of 12) patients treated with large elective rt volumes 14 .The present study of a larger group of dtc patients supports the observation that current treatments produce grade 3 or greater toxicities in a small number of patients.Using imrt, we documented acute grade 3 or greater esophagitis in 17% of patients (5 of 30).Two patients required tracheostomy and gastrostomy tube insertion during follow-up.Further technical improvements in imrt or optimized volumetric modulated arc therapy with an aim to reduce the dose to organs at risk could further reduce treatmentrelated toxicity in this patient population 6,31,32 .
In the era of imrt, expanding the rt volume for patients with high-risk dtc to include level ii might improve the recurrence rate, with lessened toxicity.On the other hand, the overall locoregional failure rate remains relatively low even when accounting for competing risks (17% at 5 years), and the benefit of a larger rt field for the population as a whole would likely be small.

Limitations
Our study has several limitations.First, the sample size was relatively small, and the prescribed doses and fractionations in the atc group were heterogeneous.
Further, the categorization of failures using the "20%-95%" analysis has its own limitations.For example, it is not possible to know whether a marginal recurrence started within the 95% isodose volume, growing outside of it, or outside the 95% isodose volume, growing within.We occasionally had to truncate the part of the recurrence volume situated outside the original planning ct imaging, partly because of the limitations of rigid registration between the planning and recurrence ct imaging when the patient contours differed significantly.The reduced volume could lead to an overestimation of the proportion of the recurrence that was covered by the 95% isodose line.However, in reviewing the 2 patients for whom modified volumes were used, deformable registration would not have changed the categorization of the recurrences.
Another limitation is that toxicities might have been underreported because they were not prospectively scored during treatment.With respect to atc, we recognize that a more common clinical scenario is that of unresectable disease treated with upfront rt.However, such patients have a poor prognosis and might not receive routine serial follow-ups and ct imaging, such that we chose to omit them from the analysis.Our results should not be extrapolated beyond the adjuvant setting.
Finally, 2 patients in the present analysis had lowvolume lung metastases at presentation.Nevertheless, follow-up duration was 2-6 years in those patients, showing that local control is essential even in the presence of distant metastasis.

CONCLUSIONS
Out-of-field regional recurrences in patients with highrisk dtc treated with a limited rt volume were located in nodal regions adjacent to the treated volumes.However, the overall regional failure rate remains modest, and expanding the rt fields to cover additional nodal echelons would benefit only a small number of patients and could increase toxicity.Out-of-field atc recurrences were scattered in a non-contiguous fashion, raising the question of whether an extended rt volume would improve locoregional control.
in 1 patient, and 1 patient with lung metastases developed late grade 4 tracheal obstruction and hemorrhage requiring emergency tracheostomy and bronchial artery embolization.In total, 2 patients experienced permanent tracheostomy tube insertion, one before rt and the other, 6 years later.Gastrostomy tubes were required in the same 2 patients, permanently in the former and for 3 months in the latter.

FIGURE 1
FIGURE 1 Overall survival for differentiated (DTC) and anaplastic thyroid cancer (ATC).Patients with ATC experienced significantly worse survival.

TABLE I
Patient, tumour, and treatment characteristics a Available for 28 of the 30 patients.Dx = diagnosis; NA = not applicable; RT = radiotherapy; RAI = radioactive iodine.
Table i presents complete patient, tumour, and treatment characteristics.

TABLE II
Treatment and recurrence details for patients with locoregional recurrence after radiotherapy Histology