As a simply implemented procedure, RECIST has both its advocates and critics. Publications both supportive and critical can be cited in the scientific literature (James et al. 1999; Mazumdar et al. 2004; Park et al. 2003). RECIST and WHO have their devotees as easily understood methods that allow simple ruler analysis of printed films as well as workstation use of electronic calipers to produce comprehensible results. Clinical imaging usually provides correlative or mostly secondary trial end points, so mRECIST and WHO criteria provide pragmatically adequate tools that satisfy a noncritical role relative to other data and clinical outcome that take primacy. They are accommodating of a variety of imaging acquisition circumstances and place minimal added demand on routine clinical practices. In sum, they have been perceived as simple tools adequate to imaging’s supportive role. To date, few widely available alternatives exist that are as easily executed or of provably greater benefit to justify further expense, time demands, or operational complexity (Tran et al. 2004).

As RECIST was framed in the context of individual slices, the research community is currently reexploring the obvious gaps in both RECIST and WHO criteria, which admittedly were constrained by the limits of earlier technology. To list the most obvious shortcomings, neither linear- nor bilinear-based methods address intratumor heterogeneity and its change over time, as it might occur after tumor-targeted therapy (e.g., Yttrium-90 microsphere embolization, radiofrequency ablation, etc.), nor do they reflect appropriate measures for tumor metabolism (Fig. 1). They do not incorporate multislice integrated understanding, register information about time-sequence change of shape or morphologic complexity, or address statistical uncertainties arising from low-intensity lesion edges. The techniques do not provide methodological distinctions between tumors of inherently high contrast compared with their surrounding tissue (e.g. lung), nor do they prescribe specific approaches to the use of contrast materials usually needed to enhance intra-abdominal soft tissue findings (Kamel and Bluemke 2002). Most importantly, little attention was paid to acknowledge in the guidelines the inconsistencies inherent in the expert observer. The reader makes his/her measurements unassisted by anything other than the most rudimentary form of image processing technology (often simply the use of electronic calipers on a workstation display). Neither RECIST nor WHO provide especially rigorous guidance on the subject of observer variability aside from recommending review panels and independent observers. Disagreement among observers has been noted to be as high as 15–40 % in these contexts and may not be ideally remedied by consensus (Belton et al. 2003). Besides providing only nominal guidance on slice thickness, RECIST does not address at any length image acquisition components that inevitably result in significant lesion contrast differences within and between studies, such as lack of uniformity of machine settings for kVp (peak kilovolts) and mAs (milliampere seconds) in CT, and pulse sequences in MRI. As a simply adoptable, widely applicable method, posing no impediment to accrual from a wide range of CT and MRI sites, RECIST has served a useful historic purpose in grouping image data into the rough four-group response classifications (CR, PR, SD, and PD). But since diameter measurements are best determined on smoothly shaped, distinct tumor boundaries, an ideal circumstance encountered infrequently, measurement variability inherent in such judgments is not adequately reflected in the recorded data. Tumors with irregular or diffuse boundaries pose the most significant challenge to data extraction and are highly observer dependent. Indeed, tumor boundary distinctiveness varies on a disease- or organ-specific basis. Observer recognition of boundaries may be further complicated by necrosis-caused internal heterogeneity that permeates the tumor or expresses itself asymmetrically on the lesion edge. Especially in the liver, tumor boundary sharpness in both CT and MRI may be enhanced by injection of contrast agents. But contrast agent pharmacokinetics are variable, and image acquisition routines are often compromised because they are usually prescribed by time from contrast administration, rather than the more definitive, but harder to obtain, contrast arrival time within specific organs.

Whole-body PET using FDG is an imaging modality enabling detection of cancerous disease by tracing increased accumulation of FDG in tumor tissue. The introduction of combined PET-CT scanners has made a new modality available for WB imaging, combining the functional data of PET with the detailed anatomical information of CT imaging in a single examination (Beyer et al. 2000).

[18F]-fluoro-2-deoxy-D-glucose-PET provides a functional metabolic map of glucose uptake in the WB. FDG is a glucose analogue that is labeled with the positron emitting radioisotope fluorine-18 that is produced by a cyclotron. The resulting radiopharmaceutical agent F-18 FDG is taken up by metabolically active tumor cells using facilitated transport similar to that used by glucose. The rate of uptake of FDG by tumor cells is proportional to their metabolic activity. Since FDG is a radiopharmaceutical analog of glucose, it also undergoes phosphorylation to form FDG-6-phosphate like glucose. However, unlike glucose, it does not undergo further metabolism, thereby becoming “trapped” in metabolically active cells (Kapoor et al. 2004). In general, PET is limited by poor anatomic detail, and therefore, anatomical correlation with some other form of imaging, such as CT, is desirable for differentiating normal from abnormal radiotracer uptake and accurate lesion localization.

First study results indicate, that a fusion of both modalities (PET and CT) improves diagnostic accuracy as well as lesion localization and report promising results for the staging of different oncological diseases compared to PET and CT alone (Lardinois et al. 2003; Pelosi et al. 2004). The total standard uptake value (SUV) of the entire axial slices of the liver as well as of the individual lesion correlated well with the laboratory and tomographic imaging results (Wong et al. 2004). However, PET-CT in some cases holds a risk of diagnostic misinterpretation, e.g., due to increased FDG-uptake in muscle or fat tissue, reduced spatial resolution or incorrect lesion localization caused by an inadequate fusion of the PET and CT data due to breathing artifacts. Furthermore, some tumor entities show no or only infrequent FDG-uptake (e.g., HCC), which suggests that FDG-PET-CT does neither represent the imaging modality of choice for detection nor response assessment in these tumor entities.

Imaging plays a major role in detecting and follow-up of metastatic disease in the liver, which strongly influences the treatment strategy. Contrast-enhanced CT has been reported as the most sensitive test for the detection of hepatic metastases. However, it has a considerable rate of false-positive findings, lowering the positive predictive value (Nelson et al. 1989; Soyer et al. 1992). False-positive results from FDG-PET in the liver are rare and occur primarily in hepatic abscesses. Delbeke et al. reported a lower sensitivity (91 vs 97 %) but higher specificity (95 vs 50 %) for FDG-PET resulting in a superior overall diagnostic accuracy compared to contrast-enhanced CT (Delbeke et al. 1997). In the study of Topal et al., PET was shown to be capable of detecting liver metastases with 99 % sensitivity (Topal et al. 2001). Several studies have compared the accuracy of FDG-PET and CT in the detection of hepatic metastases (Arulampalam et al. 2004; Bohm et al. 2004; Ogunbiyi et al. 1997). Overall, FDG-PET was more accurate than CT. Ogunbiyi et al. reported high sensitivity (95 %) and specificity (100 %) of FDG-PET for detecting liver metastases. In their study, the sensitivity and specificity of CT were 74 and 85 %, respectively (Ogunbiyi et al. 1997). In a meta-analysis, Kinkel et al. compared ultrasonography, CT, MRI, and FDG-PET in the detection of hepatic metastases from colorectal, gastric, and esophageal cancer (Kinkel et al. 2002). In this study, the sensitivity of the modalities with specificity higher than 85 % was 55 % for ultrasonography, 72 % for CT, 76 % for MRI, and 90 % for FDG-PET.