The detection of non-iodine-avid disease is an important issue in the management of differentiated thyroid cancer. This is commonly in the context of a raised thyroglobulin with negative radioiodine whole-body imaging. In some patients, the presence of thyroglobulin antibodies invalidates the serum thyroglobulin result [5
]; additional methods are required to detect non-iodine-avid recurrence. In the context of known non-iodine-avid oligometastases, the detection of additional sites of disease is a pivotal issue in the decision-making process regarding resection.
F-FDG PET has been shown to be an effective tool for detecting thyroid cancer recurrence or metastases [8
]. A recent meta-analysis of 17 studies between 1990 and 2008 including a total of 571 patients found an overall sensitivity for PET of 0.835 (95% CI 0.791–0.873) and specificity of 0.843 (95% CI 0.791–0.886) [22
]. Within the pooled subgroup of patients with a raised thyroglobulin levels and negative imaging, 131
I imaging sensitivity was 0.885 (95% CI 0.828–0.929) and specificity 0.847 (95% CI 0.715–0.934).
F-FDG PET-CT has been reported to have higher accuracy than 18
F-FDG PET or CT alone in head and neck cancer [23
]. Similarly, studies of the use of integrated PET-CT in thyroid cancer have suggested an improved diagnostic ability [22
]. In view of the potential for TSH stimulation to improve detection rates of sites of thyroid cancer [12
], a TSH-stimulated integrated PET-CT scan may offer the optimum diagnostic accuracy.
A key difficulty in interpreting this series, and others [8
], is in interpreting the disease status of patients who have a raised thyroglobulin level with negative imaging, including radioiodine whole-body scintigraphy and PET-CT. In order to make an estimate of the accuracy of PET-CT we adopted a cut-off for defining a true-negative PET-CT scan as a negative scan and no rise in thyroglobulin over at least a 12-month period following the scan with no clinical or radiological evidence of disease. Although similar to the methodology of other series [8
], this definition is limited by the often indolent nature of thyroid cancer, along with the ability of exogenous thyroid hormone-mediated TSH suppression to limit disease progression. It is a common clinical experience that a low level of thyroglobulin may persist for many years without any evidence of progression and negative imaging; in this situation it is not possible to be dogmatic regarding the source of the thyroglobulin. The presence of detectable thyroglobulin does not necessarily mean disease is presents, thyroglobulin levels may also be raised owing to persistent residual thyroid tissue, resulting in a false-positive raised thyroglobulin level [26
]. While it is important to recognise this limitation when interpreting the results, it is necessary to define a clinically relevant definition of a true-negative population in order to assess the usefulness of PET-CT.
Figure 4 Example of a true-positive recombinant thyroid-stimulating hormone positron emission tomography-CT (PET-CT) leading to radiotherapy. An 83-year-old female had a papillary thyroid cancer treated with a thyroidectomy and two radioiodine ablation treatments (more ...)
The patients who underwent rTSH PET-CT in this series had all received at least one 131I ablation dose. A high proportion (42 out of 47 patients; 89%) had clear documentation of being refractory to radioiodine with negative scintigraphy following 131I ablation or on a follow-up whole-body scan in the presence of a raised thyroglobulin level. The PET-CT was performed in four of the remaining five for reasons other than an isolated raised thyroglobulin level. The majority of the scans performed (46/58) were to investigate a raised thyroglobulin level. In the remainder, PET-CT was used to clarify the nature of equivocal anatomical abnormalities on other imaging, and in four cases to stage disease in the presence of known metastases. All patients within this latter group had a raised unstimulated thyroglobulin level. There was no significant difference between unstimulated thyroglobulin levels for PET-CT scans performed for investigation of a raised unstimulated thyroglobulin level and PET-CT scans performed for the other indications (p=0.35). Therefore, the overall cohort of patients has been included in the analysis to determine the accuracy of PET-CT.
Out of the 58 rTSH PET-CT scans performed, 25 (43%) were regarded as positive. This is broadly in line with other studies, although there has been some variation of reported positivity rates: Leboulleux et al [12
] reported positive PET scans in 30 out of 63 (48%) scans, Razfar et al [11
] in 75 out of 121 scans (62%), Saab et al [21
] in 9/15 (60%) scans, and Zoller et al [14
] in 35/47 (74%) scans. These variations are likely to be accounted for by the heterogeneous nature of the patient population in all of these studies.
The overall sensitivity of 69% and specificity of 76% of rTSH PET-CT in our series is lower than that reported in the meta-analysis [22
]. Reasons for this are likely to relate to methodological variations between studies, including rates of histological confirmation, definition of a true-negative group, as well as differences in the thyroglobulin levels of patients included. One previous large retrospective study demonstrated a fall in sensitivity and specificity of PET-CT at lower levels of unstimulated thyroglobulin of 93.8% and 93.9%, respectively, above 10 µg l−1
, and 74.6% and 0%, respectively, below 10 µg l−1
]. Unstimulated thyroglobulin levels were below 10 µg−1
in 35 scans in our series, which may explain the relatively low sensitivity and specificity compared with the meta-analysis results [22
]. In a similar way to the series reported by Razfar et al [11
], the sensitivity and specificity dropped in our series below the cut-off of 10 µg−1
: 87% and 100%, respectively, above 10 µg l−1
, compared with 44% and 73% below 10 µg l−1
. These results suggest that rTSH PET-CT has a greater accuracy at higher levels of unstimulated thyroglobulin. However, it is difficult to define a value of unstimulated thyroglobulin at which the investigation is no longer useful. There was no significant difference between unstimulated thyroglobulin levels when comparing positive and negative PET-CT scans (p
=0.29), nor between true-positive PET-CT scans and all other scans (p
=0.12). In addition, in 6 out of 35 scans performed with an unstimulated thyroglobulin of <10 µg l−1
, a beneficial change in management was judged to have occurred. Unstimulated thyroglobulin levels of <10 µg l−1
were present in four cases of true-positive PET-CT. This is similar to other studies that have found true-positive results at low thyroglobulin levels [13
]. These findings, despite a fall in sensitivity and specificity, are important when considering that a localised recurrence/metastasis remains potentially curable.
Figure 5 Example of a false-positive recombinant thyroid-stimulating hormone positron emission tomography (PET)-CT. A 33-year-old female had a papillary thyroid carcinoma treated with thyroidectomy followed by radioiodine ablation 10 years previously. An unstimulated (more ...)
The thyroglobulin level in patients with differentiated thyroid cancer is dependent upon the tumour being capable of producing and releasing thyroglobulin, tumour size, any remnant thyroid tissue and on response to TSH stimulation [26
]. If a tumour is highly sensitive to TSH suppression, thyroglobulin levels may remain low or undetectable when adequate doses of thyroid hormones are administered. Therefore, an unstimulated thyroglobulin level is a very indirect indicator of tumour volume. The measurement of thyroglobulin under conditions of TSH stimulation may provide a more accurate reflection of tumour status. Levels of stimulated thyroglobulin were only obtainable for 27 PET-CT scans. However, there was a statistically significant difference between stimulated thyroglobulin levels from true-positive scans and the remainder (p
=0.046). There were no true-positive PET-CT scans with a stimulated thyroglobulin level below 8 µg l−1
. This suggests that the stimulated thyroglobulin level may provide a more accurate method of deciding which patients are likely to benefit from a PET-CT scan.
In this series PET-CT directly influenced patient management in 19 (33%) of 58 scans. In 17 (29%) cases this was judged to be beneficial. This is lower than the impact on management reported by others (40–48%) [11
]. This difference may again relate to a higher proportion of scans in our series being performed with a low unstimulated thyroglobulin level. Two patients underwent a histologically negative neck dissection following the PET-CT result. In view of this, histological confirmation is now obtained prior to proceeding to definitive intervention whenever possible.
The findings in our study are limited by the retrospective nature of the study and the consequent potential for selection bias, the limited availability of histological correlation histological correlation, and the absence of a stimulated thyroglobulin level for 31 PET-CT scans. Further larger prospective studies are required to validate our findings. The cost-effectiveness of PET-CT in the management of differentiated thyroid cancer is uncertain. A cost–benefit analysis could be usefully incorporated into future studies.