Abstract
119
Objectives: Cancer of unknown primary (CUP), which accounts for 3% to 5% of all malignancies, represents a heterogeneous group of tumors first presenting with metastases in which the site of origin cannot be identified at the time of diagnosis. Since the primary lesion is not specified in CUP, it is difficult to provide meaningful survival statistics. A powerful tool for diagnosis and staging in oncology is 18F-FDG PET/CT. Features extracted from PET images, maximum standardized uptake value (SUVmax) and total lesion glycolysis (TLG), have provided potentially more reliable prognostic value for various cancer types. Recently, the intratumoral heterogeneity of 18F-FDG uptake has been associated with treatment failure and survival time in various malignant tumors. Hence, we investigated the correlation between SUVmax, TLG, and intratumoral heterogeneity in 18F-FDG PET/CT and the prognosis for CUP patients.
Methods: A cohort of 23 CUP [M : F=12 : 11; age=64.5 ± 14.2] patients had been enrolled for whole-body 18F-FDG PET/CT scan at the time of diagnosis between September 2004 and November 2015. The locations of the largest lymph nodes were as follows: 19 in the neck, one mediastinal node, two inguinal nodes, and one para-aortal node. Metastasis locations included five bone and one bone plus pulmonary, and metastasis occurred in one case due to peritoneal dissemination. The median follow-up period was 22.5 months for those who had received appropriate treatments currently recommended. PET/CT images were retrospectively evaluated visually and semi-quantitatively using 3D volume analysis software (SYNAPSE VINCENT, FUJIFILM Medical, JAPAN). A semi-quantitative analysis was performed using SUVmax, TLG, and heterogeneity factor (HF). For the analysis of focal lesions, volume of interest (VOI) was determined for the largest lymph node among all metastatic lymph nodes using the fusion of images from PET and CT. The metabolic tumor volume (MTV) was determined as the total number of voxels with a threshold SUV greater than or equal to 40% of the SUVmax in the VOI. TLG was calculated as MTV multiplied by SUVmean. HF was measured by first determining the tumor volume using a series of SUV thresholds (40%, 50%, 60%, 70%, and 80% of SUVmax). After plotting the volume-threshold function of the tumor and performing linear regression analysis, the HF was calculated and modified into absolute values. Receiver operating characteristic (ROC) curves and the area under the curve—used to determine overall survival—were calculated to determine cut-off values of several parameters for use of 18F-FDG PET/CT in prognosis. Overall survivals among different prognostic categories were calculated by Kaplan-Meier method using the log-rank test.
Results: HF (0.72±1.06, range from 0.04 to 4.86) and TLG (136.6±170.3, range from 7.04 to 762.8) were significantly correlated with overall survival. On the other hand, SUVmax (8.35±3.17, range from 3.24 to 16.47) was not correlated with overall survival. ROC curves and the area under the curve revealed cut-off values for HF, TLG, and SUVmax of 0.33, 55.6, and 7.09, respectively. Patients were further stratified into two prognostic groups based on these cut-off values. Based on the results of Kaplan-Meier analysis with the log-rank test, we found that HF, TLG, and metastasis (M) status significantly affected overall survival. Patients with CUP that had HF<0.33 or TLG<55.6 or M0 (no distant metastasis) status had better disease-free survival rates.
Conclusion: The intratumoral heterogeneity and TLG of FDG uptake in the largest metastatic lymph node of CUP patients significantly correlated with tumor aggressiveness; these may be predictive of patient prognosis. Conversely, SUVmax was not correlated with prognosis. These novel findings may be useful for prognosis development and improving patient care. Research Support: I have no financial relationships to disclose.