To perform any therapeutic transarterial procedure in the liver in a safe and efficient manner, it is essential to be acquainted with the hepatic arterial anatomy (Covey et al. 2002). This is particularly important when ⁹⁰Y-microspheres could be inadvertently deposited in excessive amounts in organs other than the liver, such as the stomach, duodenum, gall bladder, pancreas, mesentery and, to a lesser degree, vascular structures. Serious complications that include GI ulceration, bleeding, gastritis, duodenitis, cholecystitis, pancreatitis, radiation dermatitis and pneumonitis, could occur (Leong et al. 2009; Riaz et al. 2009).

Evaluation of the extrahepatic uptake of ^99mTc-MAA using planar scintigraphy analysis can be challenging at times, leading to the misinterpretation of possible extrahepatic side effects. The location of several organs within a relatively small region in the upper abdomen demands the analysis of tomographic images. SPECT and CT or MRI fused images can be employed for the detection of the extrahepatic location of ^99mTc-MAA and, thus, detection of the potentially dangerous misplacement of the radiospheres.

As was discussed above, SPECT with integrated low-dose CT increases the sensitivity and specificity of ^99mTc-MAA SPECT when detecting extrahepatic arterial shunting. Planar ^99mTc-MAA scintigraphy, non-attenuation-corrected SPECT, and SPECT/CT have been reviewed retrospectively for extrahepatic ^99mTc-MAA deposition in ninety diagnostic hepatic angiograms obtained from 76 patients with different types of cancer (Ahmadzadehfar et al. 2010). The sensitivity for detecting extrahepatic shunting with planar imaging, SPECT, and SPECT/CT was 32, 41 and 100 %, respectively, while specificity was 98, 98 and 93, respectively. Moreover, therapy planning was changed according to the results of planar imaging, SPECT, and SPECT/CT in 7.8, 8.9 and 29 % of patients, respectively (Ahmadzadehfar et al. 2010). These results have been confirmed recently, suggesting the superiority of SPECT/CT over planar imaging for the identification of digestive, extrahepatic ^99mTc-MAA uptake sites, including the gall bladder, which were found in more than one-third of cases (Lenoir et al. 2012).

Accordingly, ^99mTc-MAA SPECT/CT should now form part of the clinical assessment prior to RE in order to better identify the risk of digestive shunts. In cases of GI uptake on ^99mTc-MAA SPECT/CT, a repeated angiographic assessment is mandatory as the source of gastroduodenal flow can frequently be identified and corrected. However, the described series (Lenoir et al. 2012) revealed a high frequency of vascular uptake at the level of portal vein thrombosis, hepatic artery, falciform artery or coil embolisation sites, without any consequences for the therapeutic management of patients. Experts must recognise these uptake sites as possible sources of error (Fig. 2).

RE with ⁹⁰Y-microspheres is based on delivering a high dose of internal radiation to the tumour while maintaining a safe dose of radiation to sensitive tissues such as the lung and non-tumoural liver. The radiation dose is therefore non-homogeneous, and the precise dose received will vary within the tumour and normal liver parenchyma. For the calculation of the activity that allows the highest dose in the tumour while sparing normal liver tissue, some models have been proposed. One of these is the partition model. This model, which is based on medical internal radiation dosimetry (MIRD), partitions the lungs, tumour and non-tumoural liver into separate compartments for radiation dose modelling.

Assuming that LS and the relative distribution of ^99mTc-MAA in the tumoural and non-tumoural liver compartments (tumour targeting or tumour-to-non-tumour ratio, T/N) are similar to that of ⁹⁰Y-microspheres during subsequent treatment, the activity of ⁹⁰Y-microspheres to be administered can be estimated using the percentage of LS and T/N determined from the ^99mTc-MAA images (Ho et al. 1996).

T/N can be determined from the following equation:
where:

Although no standardised technique has been defined, several methods have been described for the calculation of SPECT/CT-based T/N ratios (Campbell et al. 2009; Flamen et al. 2008; Gulec et al. 2007; Kao et al. 2012). However, European Association Nuclear Medicine procedure guidelines for the treatment of liver cancer with intra-arterial radioactive compounds specify that the T/N ratio cannot be accurately determined unless an attenuation-corrected SPECT scan is used (Giammarile et al. 2011). According to our experience, we calculate the T/N ratio by obtaining the activities of ten identical ROIs drawn on the attenuation-corrected SPECT images obtained by means of SPECT/CT. Five ROIs are of relevant tumour lesions and another five involve relevant non-tumoural areas of the liver. As the number and size are the same for tumoural and non-tumoural tissue, the ratio between the sum of image counts represents an estimation of the T/N ratio. If the tumour is poorly delineated with the CT images obtained from the SPECT/CT acquisition, corregistered SPECT and diagnostic contrast-enhanced CT or MRI fused images should be obtained (Fig. 3).

According to Lau et al. (1994), the tumour targeting (T/N ratio) predicted by the ^99mTc-MAA images showed good correlation with the T/N of radiation doses measured with a calibrated beta-probe during open surgery and also with the T/N determined from the liquid scintillation counting of biopsies taken after infusion of the microspheres (Lau et al. 1994). Consequently, we could hypothesise that tumours with high ^99mTc-MAA uptake receive a high dose of ⁹⁰Y-microspheres and, therefore, may respond correspondingly better than tumours demonstrating a low uptake of ^99mTc-MAA. However, this hypothesis has not been entirely corroborated. Some authors have not found a significant correlation between tumour ^99mTc-MAA uptake and tumour response (Hamami et al. 2009). In this regard, we studied 138 target lesions from a group consisting of 17 hepatocellular carcinoma, 14 colorectal cancer and seven neuroendocrine metastasis patients treated with ⁹⁰Y RE (Rodríguez-Fraile et al. 2008). The response to RE was based on CT or MRI changes in the maximal diameter observed in each target lesion at 3 and 6 months after RE. In agreement with the previous studies, we did not find any significant correlation between the tumour size changes during the follow-up and the ^99mTc-MAA SPECT T/N ratio for each target lesion. In contrast, other authors have shown that the T/N ratio predicts the tumoural response and even the survival (Dancey et al. 2000; Flamen et al. 2008). Using a T/N ratio threshold of unity, Flamen et al. showed that positive and negative predictive values for tumour response prediction were 71 % (17/24) and 87 % (13/15), respectively (Flamen et al. 2008) (Fig. 4).

The liver volume that shows an increased uptake after selective injection of ^99mTc-MAA into the hepatic artery reflects the vascularised volume to be treated with ⁹⁰Y-microspheres. For this reason, an important goal of the ^99mTc-MAA hepatic perfusion study is to identify the liver volume fed by a selective arterial branch (liver target volume). The calculation of this volume is critical because it directly affects the radiation-absorbed dose estimation. This is especially relevant when planning for selective (lobar) or superselective (segmental or subsegmental) ⁹⁰Y RE, and in patients with complex anatomical liver vascularisation. ^99mTc-MAA SPECT/CT vascularised volume measurement is a more functional and reliable method than current volume calculations made using anatomical images based on CT or MRI data. Moreover, an unexpectedly large volume of target liver, slightly or non-vascularised, was revealed for some patients in the ^99mTc-MAA SPECT/CT images (Garin et al. 2011). This is probably due to the existence of microvascular communications between different anatomic segments, most likely via intratumoural arterioportal shunts with low arterial blood flow that went unnoticed on angiography but were identified with ^99mTc-MAA SPECT/CT. The existence of an unexpected non-vascularised liver target volume requires the site for delivering ⁹⁰Y-microspheres to be redefined.

Some factors can affect the vascular flow of liver tumours and may need to be taken into account when performing the pre-treatment assessment of candidate patients for ⁹⁰Y RE. Consequently, these factors can also affect the ^99mTc-MAA scintigraphy distribution, and should be considered during the evaluation of nuclear medicine procedures.

⁹⁰Y RE implies the manipulation of radioactive material to ensure the accurate administration of the previously calculated activity of ⁹⁰Y for each individual. Additionally, the facility in which treatment is administered must be appropriately staffed, be fitted with radiation safety equipment as well as procedures available for waste handling and disposal, and for handling contamination. Personnel must also be able to monitor for accidental contamination and to control its spread. Trained medical staff with supporting physicist and nursing staff should undertake the administration of ⁹⁰Y-microspheres. Planar scintigraphy, SPECT or PET acquisition, performed a few hours after the administration, isuseful to locate the activity (Giammarile et al. 2011).