PET Changes Management and Improves Prognostic Stratification in Patients with Head and Neck Cancer: Results of a Multicenter Prospective Study

Patients

This prospective study was conducted at 3 Australian PET centers. Eligible patients had previously untreated carcinoma of the nasal cavity, nasopharynx, oral cavity, oropharynx, hypopharynx, or larynx, or had metastatic disease involving cervical lymph nodes from an unknown primary. Patients had to have an Eastern Cooperative Oncology Group performance status of less than or equal to 2 and had to be at least 18 y of age and available for follow-up for at least 12 mo after treatment. Patients were excluded if they had previously undergone surgical resection or radiation therapy for head and neck cancer, had concurrent active cancer, had symptomatic or radiologic evidence of distant metastatic disease, were receiving concurrent treatment with any other anticancer therapy, had uncontrolled diabetes mellitus, or were pregnant. The institutional ethics review boards at the participating hospitals approved the study, and informed consent was obtained from all patients. The study was conducted in accordance with the ethical standards of the Helsinki Declaration of 1975.

Conventional Staging Procedures

All patients were to have undergone examination under anesthesia and biopsy to confirm the diagnosis. Contrast-enhanced CT of the neck was required within 6 wk of the PET scan. Depending on clinical suspicion, the referring clinicians performed MRI of the neck, CT of the chest and other regions, and chest radiography before PET. Pre-PET investigations were performed according to institutional protocols at the referring centers and were not standardized. Investigators were required to record if lesions seen on PET were adequately assessed by prior anatomic imaging methods.

PET Scan Procedure and Image Interpretation

Patients fasted for a minimum of 6 h before the PET study. ¹⁸F-FDG at a dose of 120–440 MBq was administered intravenously. After a minimum uptake period of 45 min, PET emission data were acquired from the skull vertex to at least the lower abdomen. At sites with PET/CT scanners, CT scans were performed for the purposes of anatomic localization and attenuation correction of image data. PET transmission data were used for attenuation correction if the CT data were not available. Experienced, credentialed PET specialists with full knowledge of conventional imaging results interpreted the PET scans. Lesion interpretation was based on the final PET reports. SUV_max was measured for at least 1 lesion in a large subgroup of patients. SUV_max was calculated by measuring the maximum concentration of ¹⁸F-FDG in a lesion and correcting for the body weight and injected dose of ¹⁸F-FDG.

Documentation of Management Plans

Before receiving the results of the PET scans, the referring clinicians were required to document a management plan for the patient, as if PET findings were not available, but with access to all other clinical and conventional imaging results. This management plan outlined the modalities planned such as surgery, radiation therapy, chemotherapy, or a combination. Information was also collected on the planned extent of these therapies. After the release of the PET results, a second management plan was recorded, including any changes in intended management as a result of the PET scan findings. The actual implemented management plan, based on follow-up data, was also recorded.

Impact of PET on Patient Management

The impact of PET on patient management plans was assessed by comparing pre-PET management plans with post-PET management plans for individual patients and by asking the referring clinician if the management had been changed on the basis of PET results.

Patient Follow-up and Prediction of Disease-Free Survival on Basis of PET

Patients underwent follow-up for 12 mo after the completion of therapy. The date of tumor progression was recorded. Where possible, the results of posttreatment imaging were obtained. Clinicians were asked to assess the clinical response at 12 wk after the completion of therapy. The impact of detection of additional lesions by PET (when compared with conventional imaging) was analyzed for the effect on disease-free survival. The utility of SUV_max for predicting disease-free survival was also analyzed.

Statistics

A sample size of at least 60 patients was initially planned. This assumed that if 12 patients (20%) had data insufficient for analysis, and if 20% of the remaining patients were to have a change in management plan, the 95% confidence intervals would be 9%–31%. Bivariate tables were analyzed using the χ² test or Fisher exact probability test. Survival analyses were performed using the Kaplan–Meier method, and the groups were compared using the log-rank test.

Patient Demographics

Between December 17, 2003, and June 3, 2005, 72 patients were enrolled in the study. One patient was lost to follow-up before it could be confirmed that his treatment plan had been initiated. Thus, 71 patients were available for data analysis. The median patient age was 56 y (range, 35–86 y), 49 of whom were men (69%) and 22 of whom were women (31%). Tumor sites classified by ICD-10-AM code are shown in Table 1.

Lesions Detected on PET, Compared with Conventional Imaging

A total of 156 lesions were detected in the pre-PET evaluation. Not all of the pre-PET investigations were documented, as some did not show additional lesions beyond that seen on CT. Based on the documented investigations, all patients underwent a CT scan of the neck, although in 4 cases, the MRI results, rather than the CT results, were used for lesion documentation. Six patients underwent CT brain studies, 28 underwent a CT scan of other regions, 16 underwent an MRI, 8 underwent ultrasound, 6 underwent plain radiograph, and 16 had lesions documented on the basis of clinical examination. Lesion interpretation for the pre-PET investigations was based on radiology reports and information provided by referring physicians. In instances where multiple lymph nodes could not be easily separated from each other, that group was treated as 1 lesion. Primary lesions were detected in 56 patients, with 2 primaries identified in one case. The primary lesion was not detected in 15 patients.

Forty patients (56.3%) were evaluated with stand-alone PET. Thirty-one patients (43.7%) underwent PET/CT scans.

PET detected 171 lesions. In 28 patients (39.4%), PET detected 43 additional lesions, 9 of which were primary lesions; 27 were regional lymph nodes, including instances of multiple lymph nodes in 1 location; and 7 were distant metastases. Of the 171 lesions detected on PET, 160 were interpreted as consistent with malignancy, and 11 lesions were equivocal.

Pre-PET evaluation did not detect a primary lesion in 15 patients. Seven of the primary lesions detected only by PET relate to 7 of these 15 patients (46.7%). The other 2 additional primaries detected on PET were second primaries detected in 2 patients in whom the first primary lesion had been detected before PET. Figure 1 shows an example of a primary lesion detected on PET.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1. 

CT (A), PET (B), and combined PET/CT (C) images from a patient with metastatic cervical nodal disease of unknown primary. PET detected lesion in right lobe of thyroid gland (arrowhead). The patient was subsequently found to have primary thyroid cancer and went on to have thyroidectomy.

In 15 patients (21.1%), 28 lesions were detected on the pre-PET evaluation but not detected on PET. Of these 28 lesions, 5 were primary lesions, 22 were regional lymph nodes, and 1 was a possible distant metastatic lesion. Most of these lesions (18) were detected on the CT scan. The remaining 10 were detected on 1 or more modalities other than CT (MRI, biopsy, clinical examination). All 15 patients had CT scans before PET. In 2 cases, surgical biopsies performed between the pre-PET evaluation and the PET scan resulted in lesions not being detected on PET. In one case, a localized cervical lymph node was excised, and in the other, only a primary lesion (left tongue) was detected before the PET scan and this was excised before the PET scan.

The American Joint Committee on Cancer TNM stage was documented on the basis of the pre-PET evaluation and again after the PET scan. Overall, TNM stage was altered in 22 patients (31%). Six patients (8.5%) had a change in T stage based on PET (1 upstaged, 1 changed to stage TX [primary stage not assessable], and 4 changed from TX). Fourteen patients (19.7%) had a change in N stage (9 upstaged, 3 downstaged, and 2 changed from NX [nodal stage not assessable]). Nine patients (12.7%) had a change in M stage (1 upstaged, 2 downstaged, and 6 changed from MX [distant metastases not assessable]).

Additional lesions were detected in 18 of 40 patients scanned using stand-alone PET and in 10 of 31 patients scanned with PET/CT. No significant difference between these groups (χ² test) was shown. Stand-alone PET detected 95 lesions in 40 patients (mean, 2.38); PET/CT detected 76 lesions in 31 patients (mean, 2.45). Management was changed in 12 patients (30%) in the stand-alone PET group and in 12 patients (39%) in the PET/CT group. This difference in management was not significant by χ² analysis (χ² = 0.27, P = 0.60).

Post-PET Change in Management Plan

Management plans were altered on the basis of the PET result in 24 patients (33.8%) (95% confidence interval, 22.8%−44.8%). PET detected additional lesions in 19 of these 24 patients. In the other 5 patients, all of whom were treated with radiation therapy, the radiation volume was changed in 3, and the radiation dose was changed and chemotherapy abandoned in 2. An example of a patient whose radiation volume was changed is shown in Figure 2. The pre-PET and post-PET management plans and changes are shown in Table 2. Detailed information on the management plans for all patients before PET and after PET are outlined in Table 3.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2. 

CT (A), PET (B), and combined PET/CT (C) images from patient in whom PET detected additional left supraclavicular nodal disease (arrowhead) that was not reported as pathologic on CT scan. This finding resulted in change in radiation target volume.

Referring clinicians rated the impact of PET on patient management as high in 13 (18.3%), medium in 11 (15.5%), and low in 47 (66.2%). These results were significant by χ² analysis (χ² = 69.8, P < 0.001).

Referring clinicians were also asked to record if the management plan intent was curative or palliative before PET and if a change on the basis of the PET results was warranted. The number of curative-intent patients remained similar, with 70 before PET and 65 after PET. Seven patients (9.9%) had treatment intent altered by PET (6 patients were changed from curative to palliative, 1 from palliative to curative).

Actual Treatment

At the 12-wk posttreatment follow-up, actual treatment was compared with the treatment planned after PET. This information was available for all 71 patients, including 53 whose actual treatment was as planned after PET (74.6%) and 18 whose actual treatment differed from that planned after PET (25.4%). Review of these 18 patients showed that in 17 cases, the actual treatment implemented was consistent with the PET results. In the majority, planned treatment was varied because of treatment-related toxicity, disease progression, or refinement of the management plan in light of the patient's response to initial treatment.

In only 1 case, the treatment was not consistent with the PET findings: a second primary lesion was identified at the time of surgery and an (unplanned) excisional biopsy of this left tongue lesion was performed. Neither PET nor CT had identified this lesion. Thus, actual treatment was consistent with PET findings in almost all patients.

Posttreatment Clinical Response

Patients were also grouped according to clinical response at 12-wk after treatment and to whether the pretreatment PET scan did or did not detect additional lesions.

Patients who had additional lesions seen on PET were less likely to achieve a complete response as judged by the referring clinician (59%) than those with no additional lesions (76%). To generate adequate numbers for statistical analysis, the noncomplete responders were combined and compared with the complete responders. The result was not significant (P = 0.18, Fisher exact probability test).

Disease-Free Survival

At 12-mo follow-up, disease-free survival data were available for 64 patients (with 7 having been lost to follow-up or having died without progression being identified before 12 mo). These data demonstrated progression in 4 of the 4 available patients classified as palliative-intent after PET and in 20 of the 60 available patients classified as curative-intent after PET. This difference was statistically significant (P = 0.017, Fisher exact probability test).

The effect of additional lesions detected on PET was evaluated using a Kaplan–Meier survival analysis (Fig. 3). A trend toward poorer disease-free survival in those patients who had additional lesions detected on PET was demonstrated, although this did not quite achieve significance (P = 0.06, log-rank test). This analysis was performed on the 69 patients for whom disease-free survival data were available.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3. 

Kaplan–Meier analysis of disease-free survival in patients with additional lesions detected on PET vs. those with no additional lesions (P = 0.06, log-rank test). One patient who had follow-up data for more than 700 d was censored at 500 d for purposes of charting.

An evaluation of the prognostic value of SUV_max (a measure of the intensity of tumor uptake on PET) was also performed. In this analysis, the single-highest SUV_max measured from a patient's various lesions was used. Fifty-three patients had sufficient data to perform this analysis. In 42 cases, the SUV_max related to a primary lesion, and in 11, a lymph node region. Patients were divided into high- and low-SUV_max groups based on the median SUV_max of 6.5. The disease-free survival of the high- and low-SUV_max groups was compared with Kaplan–Meier survival analysis (Fig. 4). The analysis showed a clear trend toward poorer disease-free survival in the high-SUV_max group, although this did not achieve significance (P = 0.13, log-rank test).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4. 

Kaplan–Meier analysis of disease-free survival in patients divided into high and low SUV groups (P = 0.13, log-rank test). One patient who had follow-up data for more than 700 d was censored at 500 d for purposes of charting.