Early prediction of treatment outcome for lenvatinib using ¹⁸F-FDG PET/CT in patients with unresectable or advanced thyroid carcinoma refractory to radioiodine treatment: a prospective, multicentre, non-randomised study

The key eligibility criteria were as follows: age > 18 years, Eastern Cooperative Oncology Group performance status (ECOG PS) 0–2, measurable pathologically confirmed differentiated thyroid carcinoma, lesions refractory to radioiodine treatment, and no prior therapy with tyrosine kinase inhibitors, including sorafenib and lenvatinib. The definition of a disease refractory to radioiodine treatment is described later. The exclusion criteria were as follows: medullary or anaplastic carcinoma, prior treatment with chemotherapy, synchronous malignancies except for early-stage lesions curable by endoscopy, fasting blood glucose > 150 mg/dL, ECOG PS 3–4, thromboembolism requiring treatment, pregnant or nursing women, and other reasons as judged by the investigator.

This study was conducted prospectively as a non-randomised multicentre study. All relevant institutional review boards approved the study protocol. All patients provided written informed consent before participation. This trial was registered in the University Hospital Medical Information Network (UMIN) Clinical Trials Registry (number 000022592). This study was conducted in accordance with the principles of the Declaration of Helsinki. We recruited patients from 1 June 2016 to 31 March 2020. Data were collected and analysed by the investigators.

Eligible patients underwent FDG PET/CT, contrast-enhanced CT, and blood tests at baseline within 28 days of treatment initiation. Eligible patients were orally administered lenvatinib 24 mg once daily. One week after treatment initiation, the patients underwent FDG PET/CT and blood tests. Contrast-enhanced CT was performed at least 4 weeks after treatment initiation as the gold standard for evaluation. The treatment response was further confirmed by additional CT performed at least 8 weeks later. The patients were observed until disease progression or death occurred. The procedure is summarised in Fig. 1.

In this study, the patient was considered refractory to radioiodine treatment if they presented any of the following:

The highest metabolic activity within the tumour, that is, the maximum standardised uptake value (SUVmax) obtained by PET/CT, is considered the most common marker of metabolic changes in the tumour. As the main purpose of this study was to determine whether the change in SUVmax at an early time point after initiation of treatment can predict treatment outcome, the primary endpoint was to evaluate the discrimination power of SUVmax obtained by PET/CT performed 1 week after initiation of lenvatinib treatment compared with the response evaluation obtained by contrast-enhanced CT performed at least 4 weeks after treatment initiation. Evaluation was performed using the area under the receiver operating characteristic (ROC-AUC) curve.

The secondary endpoint was to determine the optimal SUVmax cut-off value which determined the closest distance from the upper left corner to the ROC curve obtained from the baseline PET/CT. The other endpoint was to perform the analysis using total lesion glycolysis (TLG), the product of mean SUV and metabolic tumour volume (MTV). MTV is the sum of the tumour volumes that show a higher SUV than a certain threshold. In this study, the previously proposed semi-automated method was used to measure the SUV of the liver to determine the threshold value [12]. The threshold was determined using the SUVmean and its standard deviation (SD) in the liver’s VOI as follows: threshold = SUVmean + 3 × SD [13].

Although this was a multicentre study, all PET/CT scans were performed at the Hokkaido University Hospital using the same PET/CT device. FDG was produced using an in-house cyclotron or was purchased from Nihon Medi-Physics Co. Ltd. (Tokyo, Japan). Patients were instructed to fast for at least 6 h before FDG administration to achieve a blood glucose level < 150 mg/dL. Blood glucose levels were measured prior to FDG administration. ¹⁸F-FDG (3.7 MBq/kg) was administered intravenously, and patients were allowed to rest in a quiet and dimly lit room. Approximately 50 min after ¹⁸F-FDG injection, an integrated PET/CT system was used to acquire the imaging data (Gemini TF PET/CT scanner; Philips Healthcare, OH, USA). PET/CT was performed in accordance with guidelines published by the National Cancer Institute [14]. PET/CT scanning is typically performed from the level of the vertex of the skull or orbits through the upper thighs, unless the patient has known metastatic lesions (e.g. tibial lesions). Whole-body unenhanced CT was performed before emission acquisition for anatomical localisation and attenuation correction. The CT parameters included an axial slice thickness of 2 mm, tube voltage of 120 kV, fixed (120 mA) or variable tube current, and table speed of 13.5 mm/s. Emission data were acquired in three-dimensional mode using a state-of-the-art time-of-flight technique and stored in list mode. Emission scanning was initiated 60 min (allowance: 55–65 min) after FDG administration for 2 min per bed position. The trans axial and axial fields of view were 576 and 180 mm, respectively. PET images were reconstructed for attenuation-corrected and non-attenuation-corrected data using a blob-based, ordered-subset, expectation–maximisation algorithm. The PET, CT, and fused images were displayed in 4-mm slices. Contrast-enhanced CT examinations were performed at each institution. The reconstructed image data were stored electrically in the hospital information system at the Hokkaido University Hospital. The list-mode data were electronically stored in the Department of Radiology at Hokkaido University Hospital.

FDG PET/CT and contrast-enhanced CT data were interpreted clinically by a board-certified nuclear medicine physician and a diagnostic radiologist, respectively, at our institution. For FDG PET/CT analysis, a non-physiological increase in FDG uptake compared with the surrounding background was classified as positive for the disease. A nuclear medicine physician measured the SUVmax within the tumour using a spherical volume of interest. The SUV was defined as the measured activity concentration (kBq/mL) multiplied by body weight (g) divided by the injected activity (kBq). Contrast-enhanced CT images were analysed using the Response Evaluation Criteria for Solid Tumours (RECIST) criteria.

As the response rate to lenvatinib was reported to be 64.8% [10], we assumed the expected response rate to lenvatinib in this study to be 60%. The significance level was set at 5%. Using Obchowski’s method [15] with a diagnostic ability of SUVmax area under the curve (AUC) = 0.84, 20 patients were required: 12 responders and 8 non-responders, to achieve a power of 80%. As a 10% loss to follow-up or discontinuation of treatment within 1 week from the initiation of lenvatinib treatment was expected, we included 21 patients.

Twenty-two patients were recruited from 1 June 2016 to 31 March 2020. The survival data cut-off date was 30 April 2022. One patient was excluded after registration because tumour invasion into the internal carotid artery was detected during the pre-treatment evaluation. Although the patient was in good general condition, lenvatinib treatment was discontinued because tumour invasion could be a serious risk factor for fistula or bleeding [16]. Therefore, twenty-one patients were included for analysis. Patient characteristics are presented in Table 1. The median length of follow-up time was 32.5 months (range 4.7–70.4 months). The median patient age was 69 years (range 55–82). The patients were in good general condition, with an ECOG PS 0 or 1. All the patients received TSH-suppressive therapy for cancer management. Median TSH at treatment was 0.45 μIU/mL (range < 0.01–15.61, normal range 0.35–4.94). The thyroglobulin test is used as a tumour marker to assess the aggressiveness of thyroid cancer. The median thyroglobulin level at treatment initiation was 378.2 ng/mL (range 1.45–70,909, normal range < 46 ng/mL). All the patients had distant metastases. The most frequent metastases were in the lungs and lymph nodes (15 patients in each group). All patients had papillary carcinoma. The median SUVmax and TLG before treatment were 27.7 and 376.5, respectively.

As mentioned above, treatment was discontinued in one patient. Lenvatinib treatment (24 mg) was initiated in twenty-one patients. During the first week, the lenvatinib dose was reduced or stopped in three patients because of common terminology criteria for adverse events (CTCAE) grade 3 hypertension. The relative dose intensity during the first week of treatment was 0.95 mg/body/week. After 1 week of treatment, the patients underwent FDG PET/CT. The typical images are shown in Fig. 2. In this patient, FDG PET/CT showed a rapid treatment response after 1 week of treatment, whereas CT performed after 4 weeks of lenvatinib treatment did not show tumour shrinkage. Contrast-enhanced CT was performed at least 4 weeks after treatment initiation, and additional CT for confirmation was performed at least 8 weeks later. Of the 19 patients with target lesions, the response rate was 63% (including one complete response (CR) and 11 partial responses (PR)). Stable disease and progressive disease (PD) were seen in six patients and one patient, respectively. The PFS of all the patients is shown in Fig. 3. The median PFS in all the patients was 23.8 months (95% CI 19.9–27.6 months). The median PFS durations in CR + PR and stable disease + PD were 26.1 months and 16.6 months, respectively (P = 0.166).

ROC curve analysis of SUVmax after 1 week of lenvatinib treatment was performed as the primary endpoint. The ROC curve for SUVmax is shown in Fig. 4a. The AUC was 0.714 (95% CI 0.452–0.917) based on 2000 bootstrap replicates, and the best cut-off value for treatment response (CR + PR) for SUVmax was 15.211. The sensitivity and specificity of this cut-off value were 0.583 and 0.857, respectively.

ROC curve analysis of the TLG was performed as the secondary endpoint (Fig. 4b); AUC was 0.643 (95% CI 0.357–0.893). The best cut-off value for the treatment response for TLG was 772.72. The sensitivity and specificity of this cut-off value were 1.000 and 0.286, respectively.

PFS was evaluated based on the results of ROC curve analysis. Patients with an SUVmax cut-off value (15.211) tended to have longer PFS. The median PFS was 26.3 months in patients with an under-cut-off value and 19.7 months in patients with an over-cut-off value (P = 0.078, Fig. 5a). Regarding the TLG, only three patients had a TLG cut-off value (772.72). The median PFS was 26.1 months in patients with an under-cut-off value and 16.5 months in those with an over-cut-off value (P = 0.052, Fig. 5b).