The detection of subclinical head and neck cancer recurrence or a second primary tumor may improve survival. In the current study, the authors investigated the clinical value of a follow-up program incorporating serial 18F‒fluorodeoxyglucose‒positron emission tomography integrated with computed tomography (PET/CT) in the detection of recurrent disease in patients with head and neck cancer.
A total of 240 PET/CT scans were reviewed in 80 patients with head and neck cancer who were treated with radiotherapy (RT) from July, 2005 through August, 2007. All patients were followed with clinical examination, PET/CT, and correlative imaging for a minimum of 11 months (median follow‒up, 21 months).
The sensitivity, specificity, and positive and negative predictive values of PET/CT-based follow-up for detecting locoregional recurrence were 92%, 82%, 42%, and 98%, respectively. Corresponding values for distant metastases or second primary tumors were 93%, 96%, 81%, and 98%, respectively. Eight patients (10%) developed disease recurrences or second primary tumors that were amenable to salvage surgery with negative surgical margins. The 2-year progression-free survival and 2-year overall survival rates were significantly different between patients who had a negative and those with a positive PET/CT result within 6 months of the completion of RT (93% vs 30% [P<.001] and 100% vs 32% [P<.001], respectively).
The post-treatment follow-up of head and neck cancer poses a diagnostic challenge for the clinician. Both radiation and surgery can significantly distort normal head and neck anatomy, which may reduce the diagnostic accuracy of physical examination, computed tomography (CT), and magnetic resonance imaging.1 After radiotherapy (RT), patients remain at significant risk for local, regional, or distant failure and second primary aerodigestive tumors.2 Although the optimal approach to clinical follow-up remains controversial, there is significant interest in the application of positron emission tomography (PET)/CT to overcome the limitations of physical examination or conventional imaging alone.3-5 On the basis of previously published data suggesting that18F‒fluorodeoxyglucose (FDG)‒PET offers a more accurate assessment of therapy response and earlier detection of recurrent disease than conventional imaging, our group instituted a protocol incorporating PET/CT in the follow-up of patients with head and neck cancers in July, 2005.6-11 The goals of this study were 1) to establish the diagnostic accuracy of follow-up PET/CT in the detection of locoregional recurrence, distant metastases, and second primary tumors; and 2) to determine the impact of PET/CT on the prognosis and management of suspected disease recurrence.
MATERIALS AND METHODS
Among 89 consecutive patients with stage II through IVB12 head and neck cancer who were treated with RT at our institution between July, 2005 and August, 2007, serial PET/CT imaging was performed in 80 patients (grading determined according to the American Joint Committee on Cancer 6th edition staging system). Of the 9 patients not included in the PET/CT follow-up group, 6 had intercurrent death, 2 patients who were reirradiated developed early disease progression, and 1 patient was noncompliant. Table 1 displays the patient characteristics of the subjects included in this analysis. This retrospective study was approved by the Institutional Review Board.
Table 1. Patient Characteristics
Squamous cell carcinoma
Definitive RT was given to 41 patients and adjuvant RT was administered to 39 patients. A total of 62 patients received concurrent chemotherapy and/or cetuximab. In general, patients receiving definitive RT received 70 to 72 Gray (Gy) whereas patients treated adjuvantly received 60 to 66 Gy. RT was delivered via intensity-modulated RT in 96% of patients and 4% of patients had received prior head and neck RT. Baseline PET before therapy was available for review for 43 patients.
Standard head and neck examination was performed in all patients during follow-up appointments at 3-month intervals for the first 2 years and at 6-month intervals thereafter. This included palpation of the primary tumor site and cervical lymph nodes and, when indicated, flexible nasolaryngoscopy.
The initial PET/CT scan was recommended at 2 to 4 months after the completion of RT and at 4-month to 6-month intervals thereafter. A total of 240 PET/CT scans (range, 1‒7 scans per patient) were performed between October, 2005 and September, 2008.
All patients were imaged using a Discovery LS 16 PET/CT imaging system (GE Medical Systems, Waukesha, Wis). A high‒standard FDG‒PET imaging protocol was used for all patients.13 A helical low‒dose CT was obtained (80 mA) along with PET images covering the head and neck region with a 10-minute acquisition per bed. Patients fasted for a minimum of 6 hours before radiotracer injection and a glucose level of <200 mg/dL was confirmed. To reduce the risk of interobserver variability, PET/CT scans were interpreted by a single attending physician who was board certified in nuclear medicine (L.K.) with a subspecialty focus on PET/CT. All the studies were reread retrospectively by this reader, who was blinded to the clinical history and other imaging study results. A positive lesion was defined as the presence of FDG activity that exceeded the activity observed in the adjacent background in a location outside the borders of the pathologic lesion(s) or physiologic variant(s). A negative lesion was defined as a lesion with an FDG uptake that was equal to or less than the background uptake. The background was taken from a region that was 1 cm away from the evaluated lesion to avoid interference of the scatter from the lesion. Probability of malignancy for each lesion was assigned a score using a 5‒point scale in which 0 indicates normal; 1, probably normal; 2, equivocal; 3, probably abnormal; and 4 definitely abnormal for the presence or absence of residual or recurrent disease. For a dichotomous reading scale, scores of 0, 1, and 2 were considered negative, whereas scores of 3 and 4 were considered positive.
PET/CT findings were divided into primary mucosal sites, regional lymph nodes, and distant organs. Approximately 91% of recurrences were pathologically confirmed, whereas biopsy of suspicious findings and treatment were not pursued in 2 patients with a poor performance status. Freedom from progression of the index cancer was confirmed either pathologically or through clinical follow-up of ≥6 months after the most recent scan. Using these metrics, sensitivity, specificity, positive predictive value, and negative predictive value were calculated both for the entire follow-up program and for each individual study. Sensitivity is equal to true‒positive/(true‒positive + false-negative). Specificity is defined as true‒negative/(true‒negative + false-positive). Positive predictive value equals true‒positive/(true‒positive + false-positive). Negative predictive value is equal to true‒negative/(true‒negative + false-negative). For the purposes of composite scoring for follow-up with PET/CT, any outlying event (true‒positive, false-positive, or false-negative event) was counted regardless of the number of true‒negative findings recorded before or after the event. Time to disease progression was calculated using the Kaplan-Meier method.14 A log-rank test was used to compare differences between groups.
Clinical Follow-Up and Outcomes After RT for Head and Neck Cancer
The median follow-up was 20.5 months (range, 6 months‒38 months). All surviving patients had a minimum of 11 months of clinical follow-up (median, 22 months) after treatment and no patients were lost to follow-up. The 2-year locoregional control, distant control, progression-free survival, and overall survival rates were 86%, 83%, 77%, and 84%, respectively. Kaplan-Meier survival curves are shown in Figure 1. Table 2 describes the patterns of failure.
Table 2. Patterns of Failure
Alive and no evidence of disease
Local failure only
Regional failure only
Distant failure only
Locoregional and distant failure
Second primary failure only
Of 240 PET/CT scans, 53 (22%) were interpreted as positive. The positive predictive value for positive scans was 64% overall. The positive predictive values for local, regional, and distant sites were 56%, 65%, and 82%, respectively. In contrast, for the remaining 187 negative scans, 0.9% were false-negative scans. Although not considered to have false-negative results, 6 patients developed PET‒positive, biopsy‒proven recurrent disease within 6 months of a negative scan. Overall, the sensitivity, specificity, and negative predictive value for PET/CT after RT were 94%, 90%, and 99%, respectively. The sensitivity, specificity, positive predictive value, and negative predictive value of each PET/CT scan for the detection of local, regional, and distant disease are demonstrated in Table 3.
As a result of PET/CT imaging and physical examination, 34 patients underwent biopsy, including 19 patients with true‒positive results, 10 patients with false-positive results, and 5 patients with both false‒positive and true‒positive results. Examples of true‒positive and false-positive results in a single patient are shown in Figure 2. No patient sustained long-term morbidity as a result of biopsy. Six patients with false-positive PET/CT scans were observed based on a low index of suspicion on correlative imaging and physical examination with complete resolution of FDG activity on subsequent scans.
For the diagnosis of recurrent disease, the sensitivity, specificity, and positive and negative predictive values for follow‒up PET/CT were 92%, 78%, and 59% and 96%, respectively. Corresponding values for local, regional, and distant disease are shown in Table 4. The overall accuracy of follow-up PET/CT after RT was similar in both the definitive and adjuvant settings (Tables 5 and 6).
Treatment and Clinical Outcome of Patients With Recurrent Tumors
Of the 24 patients with recurrent or second primary tumors, only 8 were suspected to have disease recurrence based on history and physical examination. Thirteen patients had isolated failures or second primary tumors at distant sites, which would not have been detected by CT imaging of the neck alone. In addition, 2 of 9 patients with isolated locoregional disease recurrence had CT scans that were interpreted as indeterminate. Twelve patients derived a potential benefit from early diagnosis. Three patients were diagnosed with second primary stage IA (T1N0M0) nonsmall cell lung cancer at 14 months, 16 months, and 24 months, respectively, after RT. A total of 6 patients were able to undergo complete surgical extirpation of all visible sites of disease with negative surgical margins and 2 of these patients remained free of disease recurrence after a follow-up of 6 months and 11 months, respectively. One patient with stage IA lung cancer was treated with radiofrequency ablation. Two patients with distant metastases were treated with surgery and chemotherapy on the basis of limited systemic disease burden. Two additional patients who were treated with salvage chemotherapy with or without consolidative RT had complete radiographic responses, suggesting a possible benefit of early diagnosis. Two patients were eligible for a curative intent phase 1/2 oligometastasis protocol using stereotactic body RT and concurrent targeted therapy with sunitinib. A final patient was considered eligible for surgery and/or reirradiation but refused therapy.
Prognostic Significance of Early PET/CT Positivity
Of the PET/CT studies performed within 6 months of the completion of RT, 57 were negative and 22 were positive. Negative PET/CT studies were associated with significantly improved 2-year locoregional control (97% vs 49%; P < .001), distant control (95% vs 46%; P < .001), progression-free survival (93% vs 30%; P < .001), and overall survival (100% vs 32%; P < .001) compared with patients with positive PET/CT studies. The Kaplan‒Meier survival curves are shown in Figure 3.
Recent advances in the treatment of head and neck cancer have translated into improved survival.2-4 Because current treatment protocols focus primarily on improving locoregional control, further gains in survival may come from improved treatment or the prevention of distant recurrence, second primary cancer, and comorbid medical illnesses.15-18 In this article, we demonstrated that with highly effective locoregional therapy delivered by a dedicated multidisciplinary team, head and neck cancer patients now fail mostly at distant disease sites. These results are consistent with the experience of other recent reports.2, 19-21 It is clear that PET/CT imaging detects disease recurrence earlier than expectant management. Whether the early detection of locoregional disease recurrences or distant metastases increases the cure rates of salvage therapy remains unknown. As a result, routine follow-up imaging is not yet recommended by the National Comprehensive Cancer Network (NCCN) head and neck cancer guidelines.22 In the current study, the early detection of 3 second primary lung cancers may translate into improved survival with the prompt institution of therapy, although screening for lung cancer also remains an area of controversy.23
The results of the current study provide additional data regarding the diagnostic accuracy of a follow-up PET/CT scan in a relatively large series of contemporary patients who were closely followed by a multidisciplinary team. Because we obtained an average of 3 PET/CT scans per patient, we can report the performance of the follow-up protocol in addition to individual scans. As the number of scans increases, the possibility of a false-positive scan clearly increases given the relatively low prevalence of disease recurrence in such frequent periods of follow-up. Therefore, the specificity and false-positive rates should be considered critically in comparison with other studies that report only a single scan in a treated patient.10, 11, 24-33
In a recent meta-analysis of studies investigating the effectiveness of a single PET scan after RT that included patients with suspected disease recurrence, the sensitivity, specificity, positive predictive value, and negative predictive value were 94%, 82%, 75%, and 95%, respectively.25 In our experience, false-positive findings in the head and neck remain an issue for PET or PET/CT in the follow-up care of patients with head and neck cancer. Clinical correlation and pathologic confirmation is indicated before the initiation of salvage therapy. The effect of false-positive biopsies on quality of life and patient well-being warrants further investigation.
In the current study, there was an excellent correlation between PET positivity and positive predictive value. Only 1% of negative scans were found to be correlated with positive disease, whereas 64% of positive scans were true‒positive scans. For clinicians, the standardization of PET/CT reporting into positive and negative might assist with management decisions. The astute clinician would understand that any radiologic test is not necessarily diagnostic without clinical correlation and must be viewed in the context of pretest probability and likelihood score. For example, when the PET scan is interpreted as positive but the history, physical examination, and CT scan are all negative, we usually recommend close patient follow-up. Conversely, even with a negative PET scan, a worrisome physical examination or CT scan warrants biopsy.
The prognostic significance of negative PET/CT findings after RT for head and neck cancer is not itself surprising.9, 11 However, the magnitude of the difference between negative and positive scans, even with several false-positive results in the positive group, should be further investigated. PET/CT negativity within 6 months of RT may be used as an early surrogate endpoint for future clinical trials investigating new treatment combinations. To our knowledge, to date, only 1 patient with N3 neck disease, who was unable to tolerate concurrent chemotherapy, developed a local recurrence after a negative initial PET/CT scan.
We acknowledge several limitations of the current study. The retrospective nature of this study is associated with several inherent biases. To increase statistical power, we included patients treated with both definitive and adjuvant RT to a variety of primary tumor sites. This population reflects the variety of patients typically referred for consideration of RT at a major academic health center. Although the follow-up was intensive, its duration was somewhat limited by the irregularities involved in the frequency of follow-up imaging in some patients. Nonetheless, these reported results should be considered a preliminary experience of a larger PET/CT follow-up program. Because the overwhelming majority of locoregional disease recurrences occur within the first 2 years, it is unlikely that the conclusions of the current study will change significantly with further follow-up.34
In conclusion, the results of the current study confirm the contributions of PET/CT in the early diagnosis of disease recurrence and second primary tumors after RT for head and neck cancer. Whether this ultimately translates into improved survival requires further study.