Silvestri Erica, Scolozzi Valentina, Rizzo Gaia, Indovina Luca, Castellaro Marco, Mattoli Maria Vittoria, Graziano Paolo, Cardillo Giuseppe, Bertoldo Alessandra, Calcagni Maria Lucia
Department of Information Engineering, University of Padova, Via G. Gradenigo 6/B, 35131, Padova, Italy.
Department of Diagnostic Imaging, Radiation Oncology and Haematology, Institute of Nuclear Medicine, Fondazione Policlinico Universitario A. Gemelli IRCCS - Università Cattolica del Sacro Cuore, Roma, Italy.
EJNMMI Res. 2018 Aug 29;8(1):88. doi: 10.1186/s13550-018-0439-8.
The validation of the most appropriate compartmental model that describes the kinetics of a specific tracer within a specific tissue is mandatory before estimating quantitative parameters, since the behaviour of a tracer can be different among organs and diseases, as well as between primary tumours and metastases. The aims of our study were to assess which compartmental model better describes the kinetics of F-Fluorodeoxygluxose(F-FDG) in primary lung cancers and in metastatic lymph nodes; to evaluate whether quantitative parameters, estimated using different innovative technologies, are different between lung cancers and lymph nodes; and to evaluate the intra-tumour inhomogeneity.
Twenty-one patients (7 females; 71 ± 9.4 years) with histologically proved lung cancer, prospectively evaluated, underwent F-FDG PET-CT for staging. Spectral analysis iterative filter (SAIF) method was used to design the most appropriate compartmental model. Among the compartmental models arranged using the number of compartments suggested by SAIF results, the best one was selected according to Akaike information criterion (AIC). Quantitative analysis was performed at the voxel level. K, V and K were estimated with three advanced methods: SAIF approach, Patlak analysis and the selected compartmental model. Pearson's correlation and non-parametric tests were used for statistics. SAIF showed three possible irreversible compartmental models: Tr-1R, Tr-2R and Tr-3R. According to well-known F-FDG physiology, the structure of the compartmental models was supposed to be catenary. AIC indicated the Sokoloff's compartmental model (3K) as the best one. Excellent correlation was found between K estimated by Patlak and by SAIF (R = 0.97, R = 0.94, at the global and the voxel level respectively), and between K estimated by 3K and by SAIF (R = 0.98, R = 0.95, at the global and the voxel level respectively). Using the 3K model, the lymph nodes showed higher mean and standard deviation of V than lung cancers (p < 0.0014, p < 0.0001 respectively) and higher standard deviation of K (p < 0.005).
One-tissue reversible plus one-tissue irreversible compartmental model better describes the kinetics of F-FDG in lung cancers, metastatic lymph nodes and normal lung tissues. Quantitative parameters, estimated at the voxel level applying different advanced approaches, show the inhomogeneity of neoplastic tissues. Differences in metabolic activity and in vascularization, highlighted among all cancers and within each individual cancer, confirm the wide variability in lung cancers and metastatic lymph nodes. These findings support the need of a personalization of therapeutic approaches.
在估计定量参数之前，必须验证最适合描述特定组织内特定示踪剂动力学的房室模型，因为示踪剂在不同器官、疾病以及原发性肿瘤和转移瘤之间的行为可能不同。我们研究的目的是评估哪种房室模型能更好地描述原发性肺癌和转移性淋巴结中F - 氟脱氧葡萄糖（F - FDG）的动力学；评估使用不同创新技术估计的定量参数在肺癌和淋巴结之间是否存在差异；以及评估肿瘤内的不均匀性。
21例经组织学证实为肺癌的患者（7名女性；年龄71±9.4岁），经过前瞻性评估，接受F - FDG PET - CT检查以进行分期。采用频谱分析迭代滤波（SAIF）方法设计最合适的房室模型。在根据SAIF结果建议的房室数量排列的房室模型中，根据赤池信息准则（AIC）选择最佳模型。在体素水平进行定量分析。使用三种先进方法估计K、V和K：SAIF方法、Patlak分析和选定的房室模型。采用Pearson相关性检验和非参数检验进行统计分析。SAIF显示出三种可能的不可逆房室模型：Tr - 1R、Tr - 2R和Tr - 3R。根据众所周知的F - FDG生理学，房室模型的结构应为链状。AIC表明Sokoloff房室模型（3K）是最佳模型。在全局和体素水平上，Patlak法和SAIF法估计的K之间均发现了极好的相关性（分别为R = 0.97和R = 0.94），3K法和SAIF法估计的K之间也发现了极好的相关性（分别为R = 0.98和R = 0.95）。使用3K模型时，淋巴结的V均值和标准差高于肺癌（分别为p < 0.0014和p < 0.0001），K的标准差也更高（p < 0.005）。
单组织可逆加单组织不可逆房室模型能更好地描述肺癌、转移性淋巴结和正常肺组织中F - FDG的动力学。在体素水平应用不同先进方法估计的定量参数显示了肿瘤组织的不均匀性。所有癌症之间以及每个个体癌症内部代谢活性和血管生成的差异，证实了肺癌和转移性淋巴结存在广泛的变异性。这些发现支持了治疗方法个性化的必要性。
