Clinical Utility of ¹⁸F-FDG PET/CT in Assessing the Neck After Concurrent Chemoradiotherapy for Locoregional Advanced Head and Neck Cancer

Patient Eligibility and Characteristics

This retrospective Health Insurance Portability and Accountability Act (HIPPA)–compliant study was approved by the Institutional Review Board. Patient consent was not required. From our database, we identified patients with HNSCC treated with definitive chemoradiotherapy between March 2002 and December 2004 who met these inclusion criteria: histologically confirmed HNSCC with neck lymph node metastases, no evidence of distant metastases (M0) (26), definitive treatment with concurrent chemoradiotherapy, and posttherapy ¹⁸F-FDG PET/CT no later than 6 mo after the end of therapy. Patients with cancer of the nasopharynx, paranasal sinuses, and salivary glands were excluded. Patients who were treated with irradiation only and those treated with palliative intent were also excluded.

Sixty-five patients with 84 heminecks met these criteria (Table 1), including 58 men and 7 women (mean age, 57 ± 9 y; range, 28–75 y). The primary disease sites were the oropharynx (n = 49: 24 tonsillar, 24 base of tongue, 1 soft palate), larynx (n = 9), or hypopharynx (n = 7).

Chemoradiotherapy and Neck Dissections

Treatment consisted of external-beam radiotherapy with concurrent chemotherapy. Twenty-eight patients underwent conventional radiotherapy, and 37 underwent intensity-modulated radiotherapy (IMRT). Radiotherapy was generated by a 6-MV linear accelerator. Opposed lateral fields were used in patients who underwent conventional radiotherapy. Electron beams were used to boost selected nodal regions. Patients were treated with a total cumulative dose of 66–72 Gy in once-daily fractions of 180–200 cGy or to a total dose of 70–72 Gy using a concomitant boost radiotherapy technique. During the first and fourth weeks of radiotherapy, patients received chemotherapy: high-dose cis-platinum (100 mg/m² over 2 d, intravenously) or carboplatin (60–70 mg/m² daily, intravenously) and 5-fluorouracil (600 mg/m² daily, continuous infusion) for 4 d when there was concern for ototoxicity or renal toxicity or because of patient preference (9,27). Patients who underwent IMRT were treated with a total dose of 70 Gy over 33 d, given daily. After chemoradiotherapy and ¹⁸F-FDG PET/CT, 17 patients underwent neck dissection (16 unilateral, 1 bilateral). The decision to perform neck dissection was based on high-risk features (e.g., ≥N2 disease), posttreatment clinical assessment and imaging findings, and patient preference.

¹⁸F-FDG PET/CT and Structural Imaging

All patients were imaged using a standard clinical PET protocol: 555 MBq (15 mCi) of ¹⁸F-FDG injected intravenously. Images were acquired from the skull base to the upper thighs at a minimum of 45 min after ¹⁸F-FDG injection. Studies were acquired on combined PET/CT tomographs, either Biograph (Siemens Medical Solutions, Inc.) (28) or Discovery LS (GE Healthcare) (29). Emission images were acquired for 4 min per bed position. The CT data were used for attenuation correction and anatomic localization. The median time interval between the end of therapy and ¹⁸F-FDG PET/CT was 12 wk (range, 8–27 wk). In 35 patients, the ¹⁸F-FDG PET/CT scans were obtained between 8 and 12 wk, in 28 patients between 13 and 20 wk, and in 2 patients >20 wk after the end of therapy. The median time interval from the ¹⁸F-FDG PET/CT scan to neck dissection was 29 d (range, 5–61 d). Structural imaging was performed with standard CT or MRI protocols using intravenous contrast material.

Image Interpretation

All ¹⁸F-FDG PET/CT studies were evaluated retrospectively by one of the investigators who was unaware of other imaging findings, clinical findings, or patient outcome. Whenever available, baseline PET/CT was used for comparison. The PET/CT findings after treatment were catalogued as normal, abnormal, or equivocal and were then cross-referenced with the original clinical PET/CT report. In case of disagreement between the independent investigator and the original report, images were reviewed by a second investigator, and a consensus was reached. In all cases, attenuation-corrected images were reviewed on a picture archiving and communication system (PACS) workstation (AW suite; GE Healthcare), displaying a maximum-intensity-projection image and at least transaxial PET, CT, and PET/CT fusion images. ¹⁸F-FDG uptake was considered abnormal when it was focal (rather than diffuse), outside normal anatomic structures seen on companion CT, and of intensity greater than background blood-pool activity or uptake in adjacent normal tissue. Maximum standardized uptake values (SUVs) were measured for these lesions. Background SUV was also obtained for the contralateral normal neck side and at the treated disease site.

For all other imaging studies, the official clinical reports from CT and MRI studies, generated by staff radiologists at this institution, were used. Only imaging studies obtained within 1 mo before or after PET were considered.

Histopathologic Examination of Neck Dissection Specimens

Archived histopathologic material was reviewed by a single pathologist. Lymph nodes were assessed for the presence or absence of tumor cells in addition to the presence and extent of necrosis, nodal fibrosis, radiation atypia, and several other parameters. Viable tumor cells were defined as those epithelial cells present within a lymph node, adjacent fibroadipose tissue, skeletal muscle, or other structures, which were morphologically identifiable and recognizable as squamous. These cells show no evidence of irreversible injury—that is, they have intact nuclei, as opposed to the presence of karyolysis or karyorrhexis.

Data Analysis

An effort was made to interpret the images in a binary fashion as either normal/probably normal or abnormal/probably abnormal. Imaging findings were considered true-positive (TP) for residual disease or metastases if confirmed by any one of the following: a positive histopathology from biopsies or neck resections, the presence of a detectable lesion in the corresponding site on conventional imaging studies, and an increase in lesion size on follow-up imaging studies. ¹⁸F-FDG PET/CT findings were classified as FP if apparent abnormalities did not meet any of the above criteria, as true-negative (TN) if the scan was negative and no disease was detected, and as FN if the scan was negative but metastatic disease was found in the neck dissection specimen or became clinically apparent within 24 mo after the date of the PET/CT or at the time of death in patients who died of progressive disease before that time point.

Data are presented on the basis of heminecks and classified according to the presence or absence of residual lymphadenopathy (>1 cm in short-axis diameter) as our experience and prior studies (21,30) suggested that residual disease is unlikely (prevalence, <5%) in heminecks without residual enlarged nodes.

Statistical Analysis

All data are presented as mean ± 1 SD. The follow-up time was calculated from the date of treatment completion to the date of the last contact or death. Time to failure (local, regional, or distant) was calculated from the date of treatment completion to the date of a relevant event. The positive predictive value (PPV), negative predictive value (NPV), sensitivity, and specificity were calculated. Unpaired t tests were performed to compare data between 2 groups of heminecks—for instance, SUV in TP versus FP heminecks. The Fisher exact test was used to compare the diagnostic performance of ¹⁸F-FDG PET/CT between groups. A P value < 0.05 was considered significant.

Patient Outcome

For all 65 patients, the median follow-up time was 37 mo (range, 8–60 mo). When excluding individuals who died of progressive disease, the median follow-up time was 43 ± 10 mo (range, 27–62 mo). Residual cancer in neck lymph nodes was ultimately proven in 7 individuals. Nineteen patients died during follow-up (median time to death, 19 mo; range, 8–41 mo); 15 of these 19 individuals died of their disease, 3 of unknown cause, and 1 of a ruptured carotid artery secondary to radionecrosis. Of the 15 patients who died of disease, 13 died from distant metastases, 1 from local disease, and 1 from local and regional disease. With regard to the 46 surviving patients, 1 had local recurrence of tonsil cancer at 27 mo, 1 had local recurrence in the skin of the neck at 13 mo, 1 had tonsil cancer at 5 y after successful treatment of initial piriform sinus cancer, and 1 had sigmoid adenocarcinoma at 28 mo of follow-up.

Heminecks with Residual Lymphadenopathy

A total of 29 heminecks, in 26 patients, showed residual lymphadenopathy on contrast-enhanced CT/MRI; all of these abnormalities were also clearly noted on the CT component of the ¹⁸F-FDG PET/CT (Fig. 1 and Tables 2, 3, and 4). The initial nodal stage in these patients was N1 in 2 patients, N2A in 2 patients, N2B in 14 patients, N2C in 7 patients, and N3 in 1 patient. Their primary disease sites were the base of tongue (n = 12), tonsil (n = 7), larynx (n = 2), soft palate (n = 1), or hypopharynx/piriform sinus (n = 4). Five of these 29 heminecks were found to harbor metastatic disease, whereas the remaining 24 heminecks were free of disease. In 4 of the 5 heminecks with metastatic disease in the specimen, ¹⁸F-FDG PET/CT was abnormal (TP), and in 1 case ¹⁸F-FDG PET/CT was negative (FN). With regard to the 24 heminecks without evidence of residual disease, PET/CT findings were FP in 6, TN in 17, and equivocal in 1 hemineck (however, the latter patient showed distant metastasis). The details on these 29 heminecks with residual lymphadenopathy are presented in Figure 1 and Tables 2–4⇓⇓.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1. 

Flow chart shows distribution of findings in 84 heminecks studied. ND = neck dissection; f/u = follow-up.

Ten of the 29 heminecks with residual lymphadenopathy showed abnormal ¹⁸F-FDG uptake. The original nodal stage in these cases was N1 in 1, N2A in 1, N2B in 6, and N2C in 2 cases. Their site of primary disease was the base of tongue (n = 5), larynx (n = 2), tonsil (n = 2), or hypopharynx (n = 1). A neck dissection was performed on 7 of these 10 PET-positive heminecks. Histopathology showed residual squamous cell carcinoma in 3 PET TP specimens (Fig. 2), whereas necrotic cells or nonviable tumor cells but abundant inflammation (histiocytosis, foreign body giant cell reaction, increased number of plasma cells) was found in the 4 PET FP specimens (Fig. 3). The median time between the end of therapy and PET/CT was 9 wk in the FP and 14 wk in the TP cases.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2. 

Transaxial PET (A) and CT (B) images of 69-y-old man 14 wk after definitive therapy. (A) Focus of increased ¹⁸F-FDG uptake (arrow) in left hemineck and no focal abnormal uptake in right hemineck. (B) Enlarged residual nodes in left (arrow) and right hemineck. Pathology revealed squamous cell tumor in left hemineck but no residual tumor in right hemineck. Thus, PET scan was TP in left hemineck and TN in right hemineck.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3. 

Transaxial PET (A) and CT (B) images of 40-y-old man at 13 wk after definitive therapy. (A) Focus of increased ¹⁸F-FDG uptake (arrow) in left hemineck. (B) Corresponding enlarged residual node (arrow). Pathology revealed giant cell reaction but no viable tumor. Thus, PET was FP.

Nineteen heminecks with residual lymphadenopathy, in 18 patients, were either negative on ¹⁸F-FDG PET/CT (n = 18: 17 TN, 1 FN) or considered equivocal (n = 1). Their original nodal stage was N1 in 1, N2A in 1, N2B in 8, N2C in 7, and N3 in 1 case. The primary disease site was the base of tongue (n = 7), tonsil (n = 5), soft palate (n = 1), larynx (n = 2), or hypopharynx (n = 3). The time from the end of therapy to PET/CT was 10 ± 3 wk. Six of these heminecks underwent neck dissections, and 13 were observed. One dissected hemineck showed metastatic squamous cell carcinoma in 3 of the 20 nodes harvested. The SUV and size of the largest node were 2.4 and 2.0 cm, respectively, with the ¹⁸F-FDG PET/CT scan performed at 9 wk after therapy (Fig. 4). The remaining 5 neck dissection specimens showed benign or necrotic nodes with no viable tumor.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4. 

Transaxial PET (A) and CT (B) images of 46-y-old man at 9 wk after definitive therapy. (A) No definite focus of increased ¹⁸F-FDG uptake in neck. (B) Enlarged residual node is seen in right hemineck on CT (arrow). Pathology revealed metastatic squamous cell carcinoma. Thus, PET was FN.

One hemineck with a residual 3.3-cm lymph node was deemed equivocal on ¹⁸F-FDG PET/CT performed at 8 wk after treatment because of mild asymmetry and slight, yet focal, elevated tracer uptake in the residual node (SUV = 2.2). The patient died of distant metastases 13 mo after treatment, without evidence of recurrent neck disease.

Heminecks Without Residual Lymphadenopathy

There were 55 heminecks (in 42 patients) without residual enlarged lymph nodes (Fig. 1). Their original nodal stage was N1 in 12 cases, N2A in 1 case, N2B in 11 cases, N2C in 17 cases, and N3 in 1 case. The site of primary disease was the tonsil (n = 19), base of tongue (n = 12), larynx (n = 8), soft palate (n = 1), or hypopharynx (n = 2).

Five patients underwent neck dissections because of extensive lymphadenopathy on initial staging (N2B or N3 disease). The specimens of the 2 heminecks with FP ¹⁸F-FDG PET/CT showed extensive necrosis and inflammation with giant cell reaction. The specimen of the TP PET/CT scan showed 2 metastatic nodes with a maximum diameter of 4 mm. The remaining 50 heminecks were followed for a median of 37 mo (range, 8–60 mo). Excluding 2 patients who died of rapid development of distant metastasis or early treatment complication (both at 8 mo), the median follow-up was 38 mo (range, 15–60 mo). The single FN ¹⁸F-FDG PET/CT occurred in a patient who presented with isolated treatment failure in a neck node at approximately 20 mo.

Timing of PET/CT and Role of SUV

Diagnostic accuracy data are given in Tables 2 and 3. SUV measurements, per se, did not distinguish well between TP and FP cases (Table 4). Prior studies suggested an SUV of 3.0 as a suitable cut-off to exclude residual cancer in neck nodes. When applying this SUV to our data, the sensitivity and specificity were 57% and 84%, and the PPV and NPV were 25% and 96%, with an accuracy of 82%. Applying this SUV only to enlarged residual nodes, the sensitivity and specificity would have been 66% and 81%, with a PPV and NPV of 44% and 91% and an accuracy of 78%. Thus, SUV-based analysis did not improve the diagnostic accuracy of scan interpretation.

Sensitivity and specificity tended to be higher for PET/CT studies obtained at >12 wk after the end of therapy, but there was considerable overlap (Table 3). Although the fraction of FP findings was somewhat higher for PET/CT scans acquired at ≤12 wk versus at >12 wk, the NPV was similarly high in both groups.

Evaluation for Residual Local Disease

The mean posttreatment SUV at the primary site was 3.3 ± 0.9 (range, 1.4–5.5). In 3 patients, residual ¹⁸F-FDG uptake was considered clearly abnormal, including the left base of the tongue (SUV, 5.3), the left supraglottic larynx (SUV, 3.7), and the right tonsil (SUV, 4.3) on ¹⁸F-FDG PET/CT scans performed at 9, 9, and 18 wk after completion of treatment. These findings were FP and the patients remained free of local disease (median follow-up, 34 mo; range, 21–50 mo). The FP ¹⁸F-FDG uptake resolved on subsequent studies at 14, 17, and 27 wk after therapy.

In an additional 3 patients, the ¹⁸F-FDG PET/CT findings on scans performed at 8, 14, and 27 wk after completion of therapy were equivocal for residual primary tumor, including right tonsil (SUV, 5.9), posterior larynx (SUV, 3.6), and left supraglottic larynx (SUV, 6.2). None of these patients underwent biopsy or salvage therapy, as all 3 patients did not show evidence of local disease on clinical examination performed within 1 mo of the PET/CT scan. Two patients (with SUVs of 5.9 and 6.2) are alive without evidence of recurrent head and neck cancer at 35 and 60 mo of follow-up. The third patient (with SUV of 3.6) had neck nodal metastases detected on the same posttreatment ¹⁸F-FDG PET/CT and died of disease at 23 mo.

Fifty-nine patients showed no abnormal ¹⁸F-FDG uptake at the primary site. In 2 of these 59 cases, ¹⁸F-FDG PET/CT was FN (SUVs of 4.7 and 3.0 on PET/CT at 10 and 15 wk), because local failures at the primary site in the base of tongue were diagnosed 3 mo after the posttreatment scan. The remaining 57 scans, obtained at 8–27 wk (mean, 12 wk) after therapy, were TN (median follow-up, 37 mo; range, 15–60 mo). Excluding patients who died of a treatment complication, rapid distant metastasis, or second primary, the minimum follow-up was 23 mo.

When excluding the 3 patients with equivocal scans, negative ¹⁸F-FDG PET/CT had 95% specificity and 97% NPV for residual disease at the primary site. The SUV tended to be lower in TN (3.3 ± 0.9) and FN cases (3.8 ± 1.2) than that in equivocal (5.2 ± 1.4) and FP (4.4 ± 0.8) cases, but there was considerable overlap.

Comparison with Earlier Studies

Our results are in general concordance with data from some prior studies (21,22,32) but are at variance with data of other investigators (17,20,31). Some of these discrepancies might be explained by technical factors, the time interval between ¹⁸F-FDG PET and the end of treatment, our consistent use of combined PET/CT rather than PET alone, and the approach to image interpretation. It should also be noted that all PET/CT scans in our study were reviewed by 2 investigators. This may differ from the situation in daily clinical practice in which studies are read by several physicians, perhaps with varying experience and approach to the interpretation of ¹⁸F-FDG PET findings. Indeed, we encountered discrepancies between our interpretation and the original clinical report in 6 cases in which the original report was FP. However, although this is a retrospective study, the reviewing physician was only aware of the fact that all patients had completed definitive chemoradiotherapy for HNSCC but was unaware of the clinical findings, surgical management, histopathology, and ultimate outcome. Few prior studies have consistently used combined PET/CT in the posttreatment evaluation of patients with HNSCC in organ preservation protocols, and the results have been quite variable. For instance, Gourin et al. (31) studied 17 patients at 8–10 wk after the end of treatment and reported a sensitivity of 40%, a specificity of 25%, and a NPV of 50%. In contrast, Andrade et al. (32) reported a sensitivity and specificity of 100% for ¹⁸F-FDG PET/CT scans obtained at >8 wk after the end of chemoradiotherapy. Yao et al. (21) used PET/CT in a subset of their patients and reported a 100% NPV of this combined test. The reasons for these discrepant results remain unclear.

There is ongoing discussion with regard to the optimal time point for ¹⁸F-FDG PET after the end of chemoradiotherapy (16–18,21,22,32). Scanning too early after the end of definitive chemoradiotherapy increases the rate of FP, and potentially also FN, findings. For instance, Andrade et al. (32) noted an improvement in both sensitivity and specificity of ¹⁸F-FDG PET when the scan was obtained at >8 wk, as compared with 4–8 wk, after the end of treatment. It is the consensus opinion of our multidisciplinary team that PET/CT should be performed about 10–12 wk after the end of therapy unless clinical management requires it at an earlier time. This time point strikes a balance between the clinical desire for early, yet accurate, response assessment and the surgeon's desire not to operate on a neck that has developed extensive fibrosis and scar tissue as the result of chemoradiotherapy.

Our 3% prevalence for residual neck disease among patients with normal-sized nodes at the end of treatment is in the 0%–5% range previously reported by Peters et al. and Yao et al. (21,30). On the basis of their data and assuming that the high NPV of ¹⁸F-FDG PET or PET/CT can be reproduced by other institutions, Porceddu et al. (22) and Yao et al. (21) have outlined potential algorithms for patient management. The major difference in these algorithms is the more measured application of PET based on CT findings (22) versus routine use of PET on all patients except those with clear clinical progression (21). However, even in Yao's algorithm, nodal size is a decision criterion but is used at a later time point. Porceddu's approach is based on their earlier work showing a very high NPV for the lack of residual lymphadenopathy on follow-up CT (30). On the basis of our own data, we suggest a management strategy as outlined in Figure 5.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5. 

Our suggested algorithm for use of combined PET/CT in evaluation of patients with HNSCC after definitive chemoradiotherapy. *Either as single PET/CT with full diagnostic CT with intravenous contrast or PET/CT with low-dose CT supplemented by a separate diagnostic CT or MRI with intravenous contrast. **Observation should be considered only after informed discussion with the patient and when close monitoring is guaranteed either as part of institutional practice or in a clinical study. LN = lymph node; ND = neck dissection; ECE = extracapsular extension; CRT = chemoradiotherapy.

Limitations

We encountered FP findings in 6 of the 28 heminecks with residual enlarged lymph nodes, and the PPV in this group was only 40%. This is in the range of data reported by Yao et al. (21) with 43% and Brkovich et al. (33) with 33% but is lower than the PPVs of 71% and 90% reported by Porceddu et al. (22) and Andrade et al. (32), respectively. In each FP case, histopathologic analysis showed inflammation or granulomatous disease; both are known causes for increased ¹⁸F-FDG uptake in lymph nodes. Therefore, FP findings are unavoidable, and SUV numbers are not helpful in this regard. However, the number of FP findings can be minimized when CT and PET features (absence of residual lymphadenopathy, diffuse rather than focal uptake of ¹⁸F-FDG) are part of the study interpretation. PET/CT findings at 2 neck sites and 3 primary sites had to be classified as equivocal. This highlights the occasional difficulties in classifying imaging abnormalities in the treated neck as clearly positive for disease versus a posttreatment effect only, but we believe that with conscious effort the number of such equivocal readings can be minimized.

Some may consider the lack of histopathology for every treated hemineck a limitation of our study, leaving some uncertainty with regard to the possible presence of (microscopic) clusters of residual viable tumor cells. However, because neck dissection is no longer performed routinely after chemoradiotherapy, such results are simply not obtainable. We believe that long-term clinical follow-up provides a reasonable alternative for demonstrating the absence of residual disease.