Abstract
The functional imaging approach of nuclear medicine offers important information for the characterization of a tumor's pathobiology. In oncology, positron emission tomography (PET) especially has had great impact on the staging of tumor patients and the assessment of therapy. Both the development of new, tumor-specific, tracers and the introduction of by software- and hardware-driven image fusion emphasize the potential of this modality for an all-embracing diagnostic modality.
3.1 Introduction
More than 90% of malignant pancreatic tumors are constituted by ductal adenocarcinomas, which are characterized by their poor prognosis (mean survival 13-15 months in case of locally confined disease; 3-6 months in case of systemic spread) (Gold and Goldin 1998; Pakzad et al. 2006). Currently no screening test for the early detection of pancreatic carcinoma exists. Al-though some cancers are detected coincidentally, more are detected at an advanced stage due to the initial lack of clinical symptoms.
As cure by surgery is only possible at the early stages, there is a need for timely detection, pref-erably by noninvasive diagnostic modalities (Hi-
gashi et al. 2003).
Apart from accurate detection, imaging modalities must also meet the following de-mands: differentiation from inflammatory pan-creatic disease, assessment of metastatic spread, therapy control, and the detection of recurrence. Conventional imaging work-up usually consists of endoscopy with X-ray examination (endo-scopic retrograde cholangiopancreatography, ERCP), (endo-)sonography, computed tomog¬raphy (CT), magnetic resonance imaging (MRI), or a combination of them that offers anatomical information at the high spatial resolution neces-sary for the assessment of the primary structures and possible infiltration of neighboring struc¬tures.
These methods can be subdivided into in-vasive procedures such as endoscopy or endo-sonography that also allow interventions (e.g., Stent-placement) or biopsies (Ponchon and Pil-leul 2002; Harewood and Wiersema 2002) and noninvasive imaging modalities, for which spiral CT is considered the standard imaging modal¬ity with a sensitivity between 69% and 92% and a specificity of 74%-100% for the detection of pancreatic carcinoma (Freeny 2001). While the introduction of multislice CT (MSCT) greatly improved the accuracy of CT (Catalano et al. 2003), MRI has also experienced technical im-provements in recent years. This is especially true for the performance of additional MRCP and MR angiography in one imaging session (so-called "one-stop-shop"), which leads to sen¬sitivities of up to 95% and specificities up to 82% for carcinoma detection (Lopez Hanninen et al.
2002; Ishiguchi et al. 2001).
Apart from the initial diagnosis of pancreatic carcinoma, imaging modalities are faced with the problem of differentiating tumor from chronic pancreatitis. Although morphologic imaging al-lows a sufficiently reliable detection of pancreatic lesions, a specific characterization, crucial for the differentiation of benign from malignant disease is oten impossible. Moreover, both MRI and CT are limited in the accurate assessment of prog-nostic factors such as resectability or exclusion of distant metastases (Hanbidge 2002).
Both difficulties are the result of the fact that imaging modalities based on the visualization of anatomic details do not pay sufficient attention to an important aspect of tumor biology, the patho-biochemical changes associated with malignant transformation.
In contrast, the assessment of cellular charac-teristics such as metabolism and receptor expres-sion is the domain of nuclear medicine. Thus, even in structures with similar morphology, the functional imaging approach can differentiate viable tumor tissue from e.g., fibrosis due to a difference in tracer uptake. Another advantage of the usually systemically administered tracers is that whole-body examinations are routinely performed, while most conventional imaging modalities are often limited to a specific region. As the pathophysiologic changes on the cellular level occur ahead of the anatomical changes on the macroscopic level, functional imaging can also provide important information concerning the assessment of therapy and prognosis.
3.2 Introduction to Nuclear Medicine Imaging
Conventional scintigraphy is usually based on the acquisition of emitted gamma-quants summed in anterior and posterior projection (planar scin-tigraphy). In order to increase spatial resolution and to gain information on three-dimensional tracer distribution, single photon emission to¬mography (SPECT) of the region in question can be performed and visualized in axial, coronal, or sagittal slicing. Somatostatin receptor scintigra-phy in pancreatic tumors with neuroendocrine differentiation is a good example of a well-estab¬lished tumor scintigraphy that benefits greatly from SPECT and especially hybrid SPECT/CT imaging (de Herder et al. 2005; Amthauer et al. 2005). Moreover, a bone scan may be performed if osseous metastases are suspected. More experi-mental approaches, still limited to studies, aim at the direct visualization and therapy of pancreatic adenocarcinoma by labeling antibodies with di-agnostic and therapeutic nuclides respectively
(Cardillo et al. 2004).
Positron emission tomography (PET) offers an even higher spatial resolution than SPECT and, in contrast to SPECT, allows three-dimen¬sional visualization of the whole body. Moreover, the correction of photons for scatter and attenu¬ation allows a determination of the concentra¬tion of the tracer in the target region of interest (ROI) or target volume of interest (VOI). This semiquantification is usually expressed as the so-called "standardized uptake value" (SUV), which reflects tracer uptake in relation to the activity administered (corrected for decay) and the body weight or mass of the patient. The resulting SUVs can then be used for intra- and interindividual comparisons.
While the traditional synthesis of functional scintigraphy and morphologic imaging as a side-by-side analysis has already greatly improved the information output of either examination, the pinnacle of this combined imaging approach is seen in the visualization of the information from both examinations in one image. For a clinical routine, software-based retrospective image fu¬sion of separately acquired sets of data (CT, MRT, SPECT, PET) is feasible and has been validated in several studies (Amthauer et al. 2004, 2005;
Ruf et al. 2006; Lemke et al. 2004). Meanwhile,
both hybrid SPECT/CT and PET/CT systems are available that allow an almost simultaneous ac-quisition of imaging data, resulting in inherently fused images.
3.3 Positron Emission Tomography
F18-fluorodeoxyglucose positron emission to-mography (FDG-PET) is increasingly becoming an important diagnostic pillar in oncology. Con-cerning pancreatic carcinoma, it has shown its value in the initial detection, the differentiation from pancreatitis, and the preoperative exclusion of distant metastases. In contrast to conventional imaging modalities, FDG-PET also allows a reli-able detection of disease recurrence. Moreover, the assessment of glucose metabolism permits as-sumptions on response to therapy and prognosis. Finally, not only due to the use of hybrid PET/CT devices but also the development of more tumor-specific tracers, another breakthrough in the field of functional imaging is promising.
3.3.1 Methodic Fundamentals for the Use of FDG-PET in Pancreatic Carcinoma
3.3.1.1 Physical Aspects
In PET, both the physical prerequisites as well as the biochemical and radiochemical properties of the PET tracer have to be taken into account. The generation of a measurable signal depends on the one hand on the relation between systemic spa¬tial resolution and tumor size and on the other hand on the relation between tracer uptake and tumor metabolism.
In essence, spatial resolution is determined by the positron energy of the nuclide utilized (ef-fective positron range), the diameter of the PET scanner (non-colinearity), and the detector itself (material and size of a single detector-element). In the case of the most commonly used PET nu-clide, fluor-18, current clinical scanners achieve a spatial resolution of approximately 4 mm. In principle, smaller structures can be detected pro-vided that the tracer accumulation is sufficient to overcome the resulting partial volume effect (Cherry et al. 2003).
3.3.1.2 Mechanisms of Cellular Glucose Uptake
The use of radioactive glucose for tumor imag¬ing is based on the observations of Otto Warburg et al. (1924) who noted an increased glycolysis in malignant cells. In the case of FDG, labeling is performed by replacing the hydroxyl group at the second carbon atom with the positron emit¬ter F18. Although FDG also experiences cellular uptake via glucose transporters and consecutive phosphorylation, in contrast to regular glucose it is not subject to the further path of glycoly-sis. The result is an intracellular accumulation of the tracer, the so-called "metabolic trapping." As this accumulation is proportional to the glu¬cose intake of the target tissue, metabolically ac¬tive tumor tissue can be visualized. However, it has to be noted that, depending on the tumor entity, varying degrees of enzymatic activity and transport molecules exist, ultimately influencing cellular glucose accumulation (Arora et al. 1992; Smith 1999; Smith 2000). Concerning pancreatic cancer cells, an increased expression of trans-membranous transport proteins such as GLUT-1 or the increased enzymatic activity of hexokinase have been identified as influential factors (Reske et al. 1997; Higashi et al. 1998, 2002; Pessin and Bell 1992).
As the PET signal is generated by the activity accumulated within a voxel of 4 mm edge length, it must be noted that the above-mentioned par¬tial volume effects not only affect structures smaller than 4 mm, but also that the visualiza¬tion of larger structures is basically the result of the respective partial volume effects of neighbor¬ing voxels. This fact emphasizes that apart from the glucose avidity of a single cell, cellularity, i.e., the cellular content per volume unit, is of great importance for tumor imaging. As pancreatic carcinoma is often accompanied by desmoplas-tic reactions, the bad ratio of cellular content and extracellular matrix may cause a limited sensitiv¬ity of PET (Higashi et al. 1998). One study re¬ported several cases of tumors as large as 6 cm that did not show an elevated glucose metabo¬lism (Higashi et al. 2003). The low cellularity as a cause of nondetection is especially plausible for the scirrhous type of adenocarcinoma, but conflicting reports exist with regards to cystic tumors (Kasperk et al. 2001; Berger et al. 2004; Sperti et al. 2005)
3.3.1.3 Clinical Factors
Hyperglycemia prior to the FDG examination has been reported to negatively affect tracer up¬take (Diederichs et al. 1998). Although the ac¬tual impact of hyperglycemia is still the subject of debate, a diabetic state has to be expected in a large number of pancreatic cancer patients, either caused by the tumor itself or due to preexisting chronic pancreatitis. The results of our own analy¬sis of 174 patients with pancreatic masses showed no statistically significant difference between the patient group with and without diabetes.
Other potential factors affecting uptake might be acute phases in patients with chronic pan-creatitis or inflammatory reactions after inter-ventional procedures, e.g., stent-insertion or di¬lation. The reason for this possibility lies in the increased FDG uptake of activated leukocytes (mainly monocytes), which suggests the trac¬ers' value in inflammatory imaging. However, it also signifies that FDG is not tumor-specific, and the differentiation between benign inflammation and pancreatic carcinoma is difficult (Diederichs et al. 2000), not to mention the fact that more than 24% of the FDG uptake in cancer is due to the accompanying inflammatory reactions (Kubota et al. 1994).
According to our own data, endoscopic ex-aminations (i.e., ERC/ERCP or ultrasound) and especially manipulations do have an influence on FDG uptake and thus specificity. Specificity of FDG-PET was with 76.9% higher in those pa¬tients that had no ERC/ERCP prior to the PET
scan when compared to those that did (64%). Up to the present, no valid data on this issue or recommendations concerning a "safety interval" between intervention and PET scan exist.
3.3.1.4 Scanning Parameter for FDG-PET
A PET examination roughly consists of the in-travenous tracer injection, an uptake phase for tracer distribution, and the actual scan. Espe-cially the length of the uptake phase is currently under debate. Although the guideline of the European Association of Nuclear Medicine de-scribes an uptake phase of approximately 60 min as sufficient (Bombardieri et al. 2003), there is a tendency for longer uptake phases (up to 2-3 h) in more recent publications, as the glucose up¬take in malignomas usually increases with up¬take time, thus allowing for a better specificity
(Nakamoto et al. 2000).
Unfortunately, this is not true for all pa¬tients. Higashi and coworkers (2003) examined 68 patients with suspected pancreatic cancer by measuring glucose uptake 1 and 2 h after tracer injection and found 13 patients who did show a decrease in FDG uptake in the late scan, whereas 3 patients even had no noticeable uptake at all. Therefore we also adapted our scanning proto¬col according to their recommendations: static whole-body examination 1 h ater tracer injec¬tion.
1. If positive, end of examination.
2. If negative or unambiguous, a second scan of the pancreatic region is performed 2 h ater tracer injection.
3.3.1.5 Semiquantification
by Standard Uptake Values
In PET imaging, the tracer uptake in a target re-gion is measured as the standard uptake value (SUV) and reflects the radioactivity concentra¬tion within the target tissue divided by the whole body activity concentration (including tracer ex-cretion). The tissue activity concentration (cor-rected for decay) is defined by an ROI or VOI (measured in Becquerel per gram, milliliter, or cm2) whereas the whole-body activity concentra-tion is defined as injected activity (in Becquerels) divided by the patient's respective body surface area, weight, or volume (Thie 2004). Most studies usually refer to the maximal SUV, SUVmax, that represents the maximal pixel or voxel activity of the target ROI or VOI, respectively.
3.3.1.6 Visual Versus Semiquantitative Analysis
In the middle of the 1990s, the first studies on use of semiquantification by SUV for a more ob-jective differentiation of pancreatic carcinoma from pancreatitis were published (Inokuma et al. 1995; Koyama et al. 2001; Zimny et al. 1997; Nakata et al. 2001). However, in comparison to these one-time measurements a more reliable differentiation appears achievable by dynamic protocols that pay tribute to tracer kinetics (Voth
et al. 2003; Higashi et al. 2002).
In our patient collective, an SUV of 2-13 for benign pancreatic lesions (mean: 3.6) and 2-43 for malignant lesions (mean: 4.9) was measured. Consecutive receiver operating characteristic (ROC) analysis revealed a cut-off value of 3.7 for the differentiation of benign from malignant le-sions. The large overlap in SUVs for benign and malignant disease indicates the need for a criti¬cal approach to SUV utilization. Factors such as blood glucose level, body size, the length of the uptake phase, the shape and size of ROI or VOI, reconstruction parameters, and attenuation cor-rection have to be heeded (Thie 2004). This also implies that SUV thresholds determined at one institution cannot simply be transferred to other institutions unless parameters are kept the same (Keyes 1995; Nitzsche et al. 2002).
3.4 Detection of Pancreatic
Carcinoma and Differentiation from Pancreatitis
A summary of published FDG-PET studies on pancreatic cancer published form 1997 to 2005 resulted in a sensitivity of 71%-100% and a spec-ificity of 60%-100% (Table 3.1).
Although our own results showed a high sensitivity (96%) for the detection of pancreatic carcinoma when FDG-PET was analyzed visu-ally, we could not reproduce the high specifici¬ties reported in the literature. Using visual analy¬sis, specificity was 35%, whereas the use of the
ROC-derived SUV threshold of 3.7 only raised specificity to 68%. Therefore, the value of FDG-PET for the differentiation of pancreatitis and pancreatic carcinoma has to be regarded with care as semiquantitative imaging does not suf¬ficiently improve specificity and the data on de¬layed, dual-phase, or kinetic imaging are scarce or controversial.
Despite this skepticism, the high variability in sensitivity and specificity reported must also be analyzed with regards to the patient collec¬tive, tumor size/stage, and preexistent imaging information. As a consequence, the true poten¬tial of FDG-PET can only be assessed when an appropriate diagnostic algorithm on the basis of pretest likelihood is developed. A first analysis by Heinrich and coworkers (2005) showed that in case of a positive CT scan prior to PET imag¬ing, FDG-PET achieved a sensitivity of 92% and a specificity of 68% for the detection of pancre¬atic malignoma. In case of a negative CT scan, FDG-PET achieved a sensitivity of 73% while specificity increased to 86%. In contrast, patients with unambiguous findings in CT showed de-tection of cancer by FDG-PET with a sensitiv¬ity of 100% and a specificity of 68% (Fig. 3.1). Although preliminary, Heinrich and coworkers' results confirm the need for staging algorithms that pay tribute to the strengths and weaknesses of either method.
The functional imaging approach of nuclear medicine offers important information for the characterization of a tumor's pathobiology. In oncology, positron emission tomography (PET) especially has had great impact on the staging of tumor patients and the assessment of therapy. Both the development of new, tumor-specific, tracers and the introduction of by software- and hardware-driven image fusion emphasize the potential of this modality for an all-embracing diagnostic modality.
3.1 Introduction
More than 90% of malignant pancreatic tumors are constituted by ductal adenocarcinomas, which are characterized by their poor prognosis (mean survival 13-15 months in case of locally confined disease; 3-6 months in case of systemic spread) (Gold and Goldin 1998; Pakzad et al. 2006). Currently no screening test for the early detection of pancreatic carcinoma exists. Al-though some cancers are detected coincidentally, more are detected at an advanced stage due to the initial lack of clinical symptoms.
As cure by surgery is only possible at the early stages, there is a need for timely detection, pref-erably by noninvasive diagnostic modalities (Hi-
gashi et al. 2003).
Apart from accurate detection, imaging modalities must also meet the following de-mands: differentiation from inflammatory pan-creatic disease, assessment of metastatic spread, therapy control, and the detection of recurrence. Conventional imaging work-up usually consists of endoscopy with X-ray examination (endo-scopic retrograde cholangiopancreatography, ERCP), (endo-)sonography, computed tomog¬raphy (CT), magnetic resonance imaging (MRI), or a combination of them that offers anatomical information at the high spatial resolution neces-sary for the assessment of the primary structures and possible infiltration of neighboring struc¬tures.
These methods can be subdivided into in-vasive procedures such as endoscopy or endo-sonography that also allow interventions (e.g., Stent-placement) or biopsies (Ponchon and Pil-leul 2002; Harewood and Wiersema 2002) and noninvasive imaging modalities, for which spiral CT is considered the standard imaging modal¬ity with a sensitivity between 69% and 92% and a specificity of 74%-100% for the detection of pancreatic carcinoma (Freeny 2001). While the introduction of multislice CT (MSCT) greatly improved the accuracy of CT (Catalano et al. 2003), MRI has also experienced technical im-provements in recent years. This is especially true for the performance of additional MRCP and MR angiography in one imaging session (so-called "one-stop-shop"), which leads to sen¬sitivities of up to 95% and specificities up to 82% for carcinoma detection (Lopez Hanninen et al.
2002; Ishiguchi et al. 2001).
Apart from the initial diagnosis of pancreatic carcinoma, imaging modalities are faced with the problem of differentiating tumor from chronic pancreatitis. Although morphologic imaging al-lows a sufficiently reliable detection of pancreatic lesions, a specific characterization, crucial for the differentiation of benign from malignant disease is oten impossible. Moreover, both MRI and CT are limited in the accurate assessment of prog-nostic factors such as resectability or exclusion of distant metastases (Hanbidge 2002).
Both difficulties are the result of the fact that imaging modalities based on the visualization of anatomic details do not pay sufficient attention to an important aspect of tumor biology, the patho-biochemical changes associated with malignant transformation.
In contrast, the assessment of cellular charac-teristics such as metabolism and receptor expres-sion is the domain of nuclear medicine. Thus, even in structures with similar morphology, the functional imaging approach can differentiate viable tumor tissue from e.g., fibrosis due to a difference in tracer uptake. Another advantage of the usually systemically administered tracers is that whole-body examinations are routinely performed, while most conventional imaging modalities are often limited to a specific region. As the pathophysiologic changes on the cellular level occur ahead of the anatomical changes on the macroscopic level, functional imaging can also provide important information concerning the assessment of therapy and prognosis.
3.2 Introduction to Nuclear Medicine Imaging
Conventional scintigraphy is usually based on the acquisition of emitted gamma-quants summed in anterior and posterior projection (planar scin-tigraphy). In order to increase spatial resolution and to gain information on three-dimensional tracer distribution, single photon emission to¬mography (SPECT) of the region in question can be performed and visualized in axial, coronal, or sagittal slicing. Somatostatin receptor scintigra-phy in pancreatic tumors with neuroendocrine differentiation is a good example of a well-estab¬lished tumor scintigraphy that benefits greatly from SPECT and especially hybrid SPECT/CT imaging (de Herder et al. 2005; Amthauer et al. 2005). Moreover, a bone scan may be performed if osseous metastases are suspected. More experi-mental approaches, still limited to studies, aim at the direct visualization and therapy of pancreatic adenocarcinoma by labeling antibodies with di-agnostic and therapeutic nuclides respectively
(Cardillo et al. 2004).
Positron emission tomography (PET) offers an even higher spatial resolution than SPECT and, in contrast to SPECT, allows three-dimen¬sional visualization of the whole body. Moreover, the correction of photons for scatter and attenu¬ation allows a determination of the concentra¬tion of the tracer in the target region of interest (ROI) or target volume of interest (VOI). This semiquantification is usually expressed as the so-called "standardized uptake value" (SUV), which reflects tracer uptake in relation to the activity administered (corrected for decay) and the body weight or mass of the patient. The resulting SUVs can then be used for intra- and interindividual comparisons.
While the traditional synthesis of functional scintigraphy and morphologic imaging as a side-by-side analysis has already greatly improved the information output of either examination, the pinnacle of this combined imaging approach is seen in the visualization of the information from both examinations in one image. For a clinical routine, software-based retrospective image fu¬sion of separately acquired sets of data (CT, MRT, SPECT, PET) is feasible and has been validated in several studies (Amthauer et al. 2004, 2005;
Ruf et al. 2006; Lemke et al. 2004). Meanwhile,
both hybrid SPECT/CT and PET/CT systems are available that allow an almost simultaneous ac-quisition of imaging data, resulting in inherently fused images.
3.3 Positron Emission Tomography
F18-fluorodeoxyglucose positron emission to-mography (FDG-PET) is increasingly becoming an important diagnostic pillar in oncology. Con-cerning pancreatic carcinoma, it has shown its value in the initial detection, the differentiation from pancreatitis, and the preoperative exclusion of distant metastases. In contrast to conventional imaging modalities, FDG-PET also allows a reli-able detection of disease recurrence. Moreover, the assessment of glucose metabolism permits as-sumptions on response to therapy and prognosis. Finally, not only due to the use of hybrid PET/CT devices but also the development of more tumor-specific tracers, another breakthrough in the field of functional imaging is promising.
3.3.1 Methodic Fundamentals for the Use of FDG-PET in Pancreatic Carcinoma
3.3.1.1 Physical Aspects
In PET, both the physical prerequisites as well as the biochemical and radiochemical properties of the PET tracer have to be taken into account. The generation of a measurable signal depends on the one hand on the relation between systemic spa¬tial resolution and tumor size and on the other hand on the relation between tracer uptake and tumor metabolism.
In essence, spatial resolution is determined by the positron energy of the nuclide utilized (ef-fective positron range), the diameter of the PET scanner (non-colinearity), and the detector itself (material and size of a single detector-element). In the case of the most commonly used PET nu-clide, fluor-18, current clinical scanners achieve a spatial resolution of approximately 4 mm. In principle, smaller structures can be detected pro-vided that the tracer accumulation is sufficient to overcome the resulting partial volume effect (Cherry et al. 2003).
3.3.1.2 Mechanisms of Cellular Glucose Uptake
The use of radioactive glucose for tumor imag¬ing is based on the observations of Otto Warburg et al. (1924) who noted an increased glycolysis in malignant cells. In the case of FDG, labeling is performed by replacing the hydroxyl group at the second carbon atom with the positron emit¬ter F18. Although FDG also experiences cellular uptake via glucose transporters and consecutive phosphorylation, in contrast to regular glucose it is not subject to the further path of glycoly-sis. The result is an intracellular accumulation of the tracer, the so-called "metabolic trapping." As this accumulation is proportional to the glu¬cose intake of the target tissue, metabolically ac¬tive tumor tissue can be visualized. However, it has to be noted that, depending on the tumor entity, varying degrees of enzymatic activity and transport molecules exist, ultimately influencing cellular glucose accumulation (Arora et al. 1992; Smith 1999; Smith 2000). Concerning pancreatic cancer cells, an increased expression of trans-membranous transport proteins such as GLUT-1 or the increased enzymatic activity of hexokinase have been identified as influential factors (Reske et al. 1997; Higashi et al. 1998, 2002; Pessin and Bell 1992).
As the PET signal is generated by the activity accumulated within a voxel of 4 mm edge length, it must be noted that the above-mentioned par¬tial volume effects not only affect structures smaller than 4 mm, but also that the visualiza¬tion of larger structures is basically the result of the respective partial volume effects of neighbor¬ing voxels. This fact emphasizes that apart from the glucose avidity of a single cell, cellularity, i.e., the cellular content per volume unit, is of great importance for tumor imaging. As pancreatic carcinoma is often accompanied by desmoplas-tic reactions, the bad ratio of cellular content and extracellular matrix may cause a limited sensitiv¬ity of PET (Higashi et al. 1998). One study re¬ported several cases of tumors as large as 6 cm that did not show an elevated glucose metabo¬lism (Higashi et al. 2003). The low cellularity as a cause of nondetection is especially plausible for the scirrhous type of adenocarcinoma, but conflicting reports exist with regards to cystic tumors (Kasperk et al. 2001; Berger et al. 2004; Sperti et al. 2005)
3.3.1.3 Clinical Factors
Hyperglycemia prior to the FDG examination has been reported to negatively affect tracer up¬take (Diederichs et al. 1998). Although the ac¬tual impact of hyperglycemia is still the subject of debate, a diabetic state has to be expected in a large number of pancreatic cancer patients, either caused by the tumor itself or due to preexisting chronic pancreatitis. The results of our own analy¬sis of 174 patients with pancreatic masses showed no statistically significant difference between the patient group with and without diabetes.
Other potential factors affecting uptake might be acute phases in patients with chronic pan-creatitis or inflammatory reactions after inter-ventional procedures, e.g., stent-insertion or di¬lation. The reason for this possibility lies in the increased FDG uptake of activated leukocytes (mainly monocytes), which suggests the trac¬ers' value in inflammatory imaging. However, it also signifies that FDG is not tumor-specific, and the differentiation between benign inflammation and pancreatic carcinoma is difficult (Diederichs et al. 2000), not to mention the fact that more than 24% of the FDG uptake in cancer is due to the accompanying inflammatory reactions (Kubota et al. 1994).
According to our own data, endoscopic ex-aminations (i.e., ERC/ERCP or ultrasound) and especially manipulations do have an influence on FDG uptake and thus specificity. Specificity of FDG-PET was with 76.9% higher in those pa¬tients that had no ERC/ERCP prior to the PET
scan when compared to those that did (64%). Up to the present, no valid data on this issue or recommendations concerning a "safety interval" between intervention and PET scan exist.
3.3.1.4 Scanning Parameter for FDG-PET
A PET examination roughly consists of the in-travenous tracer injection, an uptake phase for tracer distribution, and the actual scan. Espe-cially the length of the uptake phase is currently under debate. Although the guideline of the European Association of Nuclear Medicine de-scribes an uptake phase of approximately 60 min as sufficient (Bombardieri et al. 2003), there is a tendency for longer uptake phases (up to 2-3 h) in more recent publications, as the glucose up¬take in malignomas usually increases with up¬take time, thus allowing for a better specificity
(Nakamoto et al. 2000).
Unfortunately, this is not true for all pa¬tients. Higashi and coworkers (2003) examined 68 patients with suspected pancreatic cancer by measuring glucose uptake 1 and 2 h after tracer injection and found 13 patients who did show a decrease in FDG uptake in the late scan, whereas 3 patients even had no noticeable uptake at all. Therefore we also adapted our scanning proto¬col according to their recommendations: static whole-body examination 1 h ater tracer injec¬tion.
1. If positive, end of examination.
2. If negative or unambiguous, a second scan of the pancreatic region is performed 2 h ater tracer injection.
3.3.1.5 Semiquantification
by Standard Uptake Values
In PET imaging, the tracer uptake in a target re-gion is measured as the standard uptake value (SUV) and reflects the radioactivity concentra¬tion within the target tissue divided by the whole body activity concentration (including tracer ex-cretion). The tissue activity concentration (cor-rected for decay) is defined by an ROI or VOI (measured in Becquerel per gram, milliliter, or cm2) whereas the whole-body activity concentra-tion is defined as injected activity (in Becquerels) divided by the patient's respective body surface area, weight, or volume (Thie 2004). Most studies usually refer to the maximal SUV, SUVmax, that represents the maximal pixel or voxel activity of the target ROI or VOI, respectively.
3.3.1.6 Visual Versus Semiquantitative Analysis
In the middle of the 1990s, the first studies on use of semiquantification by SUV for a more ob-jective differentiation of pancreatic carcinoma from pancreatitis were published (Inokuma et al. 1995; Koyama et al. 2001; Zimny et al. 1997; Nakata et al. 2001). However, in comparison to these one-time measurements a more reliable differentiation appears achievable by dynamic protocols that pay tribute to tracer kinetics (Voth
et al. 2003; Higashi et al. 2002).
In our patient collective, an SUV of 2-13 for benign pancreatic lesions (mean: 3.6) and 2-43 for malignant lesions (mean: 4.9) was measured. Consecutive receiver operating characteristic (ROC) analysis revealed a cut-off value of 3.7 for the differentiation of benign from malignant le-sions. The large overlap in SUVs for benign and malignant disease indicates the need for a criti¬cal approach to SUV utilization. Factors such as blood glucose level, body size, the length of the uptake phase, the shape and size of ROI or VOI, reconstruction parameters, and attenuation cor-rection have to be heeded (Thie 2004). This also implies that SUV thresholds determined at one institution cannot simply be transferred to other institutions unless parameters are kept the same (Keyes 1995; Nitzsche et al. 2002).
3.4 Detection of Pancreatic
Carcinoma and Differentiation from Pancreatitis
A summary of published FDG-PET studies on pancreatic cancer published form 1997 to 2005 resulted in a sensitivity of 71%-100% and a spec-ificity of 60%-100% (Table 3.1).
Although our own results showed a high sensitivity (96%) for the detection of pancreatic carcinoma when FDG-PET was analyzed visu-ally, we could not reproduce the high specifici¬ties reported in the literature. Using visual analy¬sis, specificity was 35%, whereas the use of the
ROC-derived SUV threshold of 3.7 only raised specificity to 68%. Therefore, the value of FDG-PET for the differentiation of pancreatitis and pancreatic carcinoma has to be regarded with care as semiquantitative imaging does not suf¬ficiently improve specificity and the data on de¬layed, dual-phase, or kinetic imaging are scarce or controversial.
Despite this skepticism, the high variability in sensitivity and specificity reported must also be analyzed with regards to the patient collec¬tive, tumor size/stage, and preexistent imaging information. As a consequence, the true poten¬tial of FDG-PET can only be assessed when an appropriate diagnostic algorithm on the basis of pretest likelihood is developed. A first analysis by Heinrich and coworkers (2005) showed that in case of a positive CT scan prior to PET imag¬ing, FDG-PET achieved a sensitivity of 92% and a specificity of 68% for the detection of pancre¬atic malignoma. In case of a negative CT scan, FDG-PET achieved a sensitivity of 73% while specificity increased to 86%. In contrast, patients with unambiguous findings in CT showed de-tection of cancer by FDG-PET with a sensitiv¬ity of 100% and a specificity of 68% (Fig. 3.1). Although preliminary, Heinrich and coworkers' results confirm the need for staging algorithms that pay tribute to the strengths and weaknesses of either method.
Комментариев нет:
Отправить комментарий