Whole-Body Dual-Modality PET/CT and Whole-Body MRI for Tumor Staging in Oncology FREE

In malignant diseases, therapeutic options as well as patients' prognoses strongly depend on the tumor stage.¹ Thus, accurate tumor staging encompassing the entire body is of essential importance. Based on its high spatial resolution and soft tissue contrast, magnetic resonance imaging (MRI) seems well suited for tumor staging. Its advantages over computed tomography (CT) and other imaging procedures have been determined for detection of parenchymal and osseous lesions.^{2 - 4} However, lengthy acquisition times in combination with limits of contrast usage have required several separate imaging sessions for whole-body coverage using MRI. Advances in scanner technology and the introduction of moving patient platforms with integrated surface-coil technology have recently enabled whole-body MRI within a single session.^{5 - 6}

In contrast to MRI, which is mainly focused on the assessment of morphological characteristics of tissue, glucose analog [18F]-fluorodeoxyglucose–positron emission tomography (FDG-PET) provides functional information on tumor metabolism. The functional data available in whole-body scans complement morphological imaging for staging different malignancies.^{7 - 9} The limitations of PET, which are associated with poor anatomic information, have recently been overcome with the introduction of dual-modality PET/CT imaging. The dual-modality approach provides accurately fused functional PET and morphological CT data in a single examination.¹⁰ Initial studies evaluating this new imaging tool report promising results in staging oncological diseases when compared with PET alone and CT alone.^{11 - 14}

The aim of this study was to determine the staging accuracies of whole-body FDG-PET/CT and whole-body MRI for different malignant diseases and to compare these 2 new imaging tools. Histological and clinical follow-up served as the standards of reference.

From December 2001 through October 2002, 115 patients with different histologically proven malignancies were referred for tumor staging with whole-body FDG-PET/CT and whole-body MRI. Data collection was performed prospectively in consecutive patients. PET/CT and MRI were performed within a 48-hour time interval. Seventeen patients fulfilled 1 of the previously defined exclusion criteria and could not be included in the study; 8 patients had a history of severe allergic reactions to intravenous contrast agents and 9 patients had contraindications for MRI such as pacemakers and other metallic implants as well as claustrophobia. The study was performed in full accordance with guidelines issued by the university hospital institutional review board. Written informed consent was obtained from all patients prior to performing the imaging procedures.

Dual-modality PET/CT imaging was performed using a biograph (Siemens Medical Solutions, Hoffman Estates, Ill). The system provides separate CT and PET data sets, which can be accurately fused on a computer workstation.

Whole-body CT (130 mA, 130 kV, 5-mm sections, 8-mm table feed, 2.4-mm incremental reconstruction) covered a region ranging from the head to the upper thighs. A dose of 140 mL of an iodinated contrast agent (Xenetix 300, Guerbet GmbH, Sulzbach, Germany) was administered intravenously and small bowel distension was ensured by administration of 1000 mL of glucose-free barium (1.5 g of barium sulfate/100 mL; Micropaque CT, Guerbet GmbH, Sulzbach, Germany) according to a standardized protocol.¹⁵ A limited breath-hold technique was used to avoid motion-induced artifacts in the area of the diaphragm.¹⁶

The PET component of the combined imaging system has an axial field of view of 15.5 cm (per bed position) with an in-plane spatial resolution of 4.6 mm. PET images were acquired 60 minutes following the administration of 350 MBq of FDG covering the same field of view as with CT. The acquisition time of PET was adapted according to the patients' weight. The time was set to 3 minutes for patients weighing up to 65 kg; to 4 minutes for patients weighing 65 to 85 kg; and to 5 minutes for patients weighing more than 85 kg. Images were scatter corrected and iteratively reconstructed. Image reconstruction was performed with and without PET attenuation correction. PET attenuation correction was based on the CT data. Before the injection of the radioactive tracer, a blood sample was taken to ensure blood glucose levels were within the normal range.

The MRI was performed on a 1.5-T Sonata System (Siemens Medical Solutions, Erlangen, Germany) using a rolling table platform (Body SURF, MR-Innovation, Essen, Germany).¹⁷ The axial scanning range was set to cover a field of view from head to thigh and hence corresponded well to the region examined with PET/CT. First, the chest and abdomen were scanned with noncontrast-enhanced T1-weighted (repetition time of 124 ms and echo time of 1.8 ms) and T2-weighted sequences (repetition time of 1200 ms and echo time of 60 ms) using a section width of 7 mm. Subsequently, a paramagnetic contrast agent (Magnevist, Schering, Germany) was administered intravenously at 3 mL/s with a dose of 0.2 mmol/kg. Commencing 12 seconds after contrast application, 7 successive contrast-enhanced 3-dimensional data sets (repetition time of 3.0 ms and echo time of 1.2 ms, flip angle of 12°) were acquired in the following order: abdomen (arterial phase), chest, abdomen (portal phase), pelvis, thighs, head, and abdomen (late venous phase).

PET/CT images were evaluated by a radiologist and a nuclear medicine physician (G.A. and L.S.F.), while MRI was read by 2 radiologists (F.M.V. and S.G.R.), who achieved consensus. The evaluating physicians were blinded to the results of the other imaging procedure. Both reader teams were supplied with the same clinical information about each patient.

Assessment of the primary tumor, of lymph node metastases, and distant metastases with PET/CT was based on qualitative and quantitative analyses. PET/CT data were evaluated qualitatively for regions of focally increased glucose metabolism and quantitatively by maximal standardized uptake values. Glucose uptake above levels of the surrounding tissue on qualitative analysis and a standardized uptake value higher than 2.5 indicated malignancy.^{18 - 19} Lymph nodes were graded as malignant or benign based on these functional criteria, independent of their size. PET/CT images were assessed with and without attenuation correction of the PET data.

Determination of malignancy on MRI was based on assessment of tumor morphological characteristics as well as the pattern of contrast enhancement. Lymph nodes were characterized to harbor malignancy based on their size. Region-specific size criteria were applied when assessing lymph nodes for malignancy. Determination of lymph node size was based on measurement of the small axis diameter. Region-specific cut-off values were defined according to previous studies.^{20 - 23} Central necrosis was considered a sign of malignant tumor spread independent of lymph node size.

Evaluation of malignant diseases was performed according to the American Joint Committee on Cancer (AJCC) staging classification.¹ The 2 imaging modalities were compared for accurate assessment of the overall primary tumor, regional lymph nodes, and distant metastasis (TNM) stage. Correct assessment of the T-stage was evaluated separately. The ability of the 2 imaging procedures to differentiate between N-positive and M-positive and N-negative and M-negative cases was assessed. The number of distant metastases was determined and recorded on an organ-to-organ basis.

Both histology and clinical-radiological follow-up served as standards of reference. Malignant disease was confirmed by histopathological verification in all patients. Tumor resection with pathological T-stage verification was performed in 46 patients; comparison of PET/CT and MRI for accurate assessment of T-stage was limited to these patients. Pathological N-stage verification by complete lymph node sampling was available in 43 patients, while M-stage was pathologically verified in 14 patients. In patients without histopathological N-stage or M-stage verification, clinical follow-up served as the standard of reference for N and M stages. The mean clinical follow-up time for all patients was 273 days (range, 75-515 days). The follow-up examination comprised all clinical information available after performance of the 2 imaging procedures, including clinical examinations, laboratory tests, radiological follow-up examinations (CT, MRI, and/or PET/CT), as well as histopathology performed later. The data of the reference standard were collected by a physician unaware of the results of the 2 imaging procedures.

The primary end point of the study was the correct classification of the TNM tumor stage using whole-body PET/CT and whole-body MRI. The study was designed to detect a difference in proportions of 0.15 (0.9 for PET/CT and 0.75 for MRI) with a proportion of discordant pairs of 0.3. To identify such assumed proportions with a power of 0.8 and a significance level of .05 using a 2-sided McNemar test, the study required a minimum of 98 patients.

Secondary end-point differences in diagnostic performance between the 2 imaging procedures were tested for significance by the McNemar test (exact) with a significance level of .05. Sensitivities, specificities, negative predictive values, positive predictive values, and accuracies (with exact 95% confidence intervals [CIs]) for both modalities were determined for detection of metastases to lymph nodes as well as for distant metastases. Statistical analyses were performed with StatXact (Version 5.0, CYTEL Software Corp, Cambridge, Mass) and SAS statistical software (Version 8.2, SAS Institute Inc, Cary, NC).

Ninety-eight patients with a mean age of 58 years (range, 27-94 years) were included in the study. Sixty-three of the patients were male and 35 were female. Eighty-two of these patients were referred for primary tumor staging, while 16 patients underwent staging for suspected tumor recurrence. The malignant diseases included bronchial carcinomas (n = 29), cancers of unknown primary site (n = 12), head and neck tumors (n = 13), melanomas of the uvea (n = 13), genitourinary tumors (n = 8), tumors of the gastrointestinal tract (n = 6), thyroid tumors (n = 6), pleural mesotheliomas (n = 6), liver tumors (n = 3), and bone tumors (n = 2). Table 1 shows the distribution of the severity of the diseases based on tumor histological evaluation and clinical follow-up.

Both imaging procedures were performed without adverse effects in all patients. Examination times amounted to a mean of 26 minutes (20-34 minutes) for MRI acquisition and 27 minutes (22-39 minutes) for acquisition of PET/CT.

PET/CT correctly characterized tumor stage in 37 (80%; 95% CI, 66%-91%) of 46 patients. T-stage was overstaged in 4 patients and understaged in 5 patients. Magnetic resonance imaging was found to be accurate in 24 (52%; 95% CI, 37%-67%) of 46 patients. Six patients were overstaged and 16 patients were understaged (Figure 1). Differences between the 2 imaging modalities were statistically significant (P<.001).

Regional lymph node involvement was correctly characterized as N-positive or N-negative with PET/CT in 91 (93%; 95% CI, 86%-97%) of 98 patients. N-stage was overstaged (false-positive) in 5 patients and understaged (false-negative) in 2 patients. Analysis of the MRI data permitted correct N-staging in 77 (79%; 95% CI, 69%-86%) of 98 patients. Thirteen patients were overstaged, while 8 patients were understaged (Figure 2). Accuracy differences between whole-body PET/CT and MRI could be detected (P = .001). Table 2 summarizes sensitivities, specificities, positive predictive values, negative predictive values, and accuracies for N-staging.

PET/CT accurately differentiated between M0 and M1 disease in 92 (94%; 95% CI, 87%-98%) of 98 patients. Metastases were overstaged in 3 patients and understaged in 3 patients. Magnetic resonance imaging correctly staged M-disease in 91 (93%; 95% CI, 86%-97%) of 98 patients, while distant metastases were overstaged in 3 and understaged in 4 patients (accuracy difference: P>.99). PET/CT proved more reliable in the detection of pulmonary metastases by revealing 170 lesions in 25 patients (true-positives) compared with only 139 lesions seen in 23 patients with MRI. However, MRI was more accurate in evaluating the liver and bones for the presence of metastases. Thus, MRI identified 132 liver metastases in 13 patients (true-positives) compared with only 117 liver lesions in 12 patients detected with PET/CT. Eighty bone metastases were detected with MRI in 11 patients (true-positives), while PET/CT revealed 75 lesions in 8 patients (Figure 3). According to the standard of reference, 20 patients had distant metastases to more than 1 organ. Sensitivities, specificities positive predictive values, negative predictive values, and accuracies of MRI and PET/CT for M-staging are shown in Table 3. Correlation of imaging results for M-staging with the standard of reference are shown in Table 4.

Of 98 patients, TNM was correctly determined by PET/CT in 75 (77%; 95% CI, 67%-85%) and by whole-body MRI in 53 (54%; 95% CI, 44%-64%). Accuracy differences between whole-body PET/CT and MRI were statistically significant (P<.001). Malignant disease was overstaged in 11 and understaged in 12 patients with PET/CT. Malignant disease was overstaged in 19 patients by MRI and understaged in 26 patients. The number of false staging results for lung cancer was 5 with PET/CT and 10 with MRI; carcinoma of unknown primary site, 2 and 4; head and neck tumors, 6 and 9; melanoma of the uvea, 5 and 10; genitourinary tumors, 2 and 3; gastrointestinal tract, 0 and 3; thyroid gland, 1 and 2; pleura, 1 and 2; and liver, 1 and 2. Compared with MRI, PET/CT had a direct impact on patient management in 12 patients. Patient management was altered from curative to palliative in 4 patients; palliative to curative in 5; curative to neoadjuvant in 1; and extended to limited surgery in 2. In 2 patients, whole-body MRI corrected the tumor stage compared with PET/CT, changing these patients' management from curative to palliative and palliative to curative.

This study highlights the potential associated with the emergence of single-examination whole-body diagnostic imaging strategies. The feasibility and diagnostic accuracy of whole-body dual-modality PET/CT and whole-body MRI has been established. They replace the classic multimodality, multiexamination strategies inherent to most established staging protocols. Reflecting the more precise definition of the T-stage and N-stage status, staging of malignancies was considerably more accurate when based on whole-body PET/CT imaging compared with whole-body MRI. Based on our data, FDG-PET/CT can be recommended as a first-line modality for whole-body tumor staging.

The concept of whole-body coverage within a single diagnostic imaging examination for tumor staging is not new. Both whole-body PET and whole-body MRI have been technically possible for many years.⁵ However, lengthy data acquisition times of up to 90 minutes have limited their clinical utility. Recent technical innovations have contributed to overcoming this limitation. By basing attenuation correction on the CT data, dual-modality PET/CT obviates the need for lengthy transmission scanning with an external radiation source, thereby reducing examination times by about 30% compared with PET imaging alone.¹⁰ Whole-body coverage with MRI is based on vastly accelerated data acquisition due to faster magnetic resonance gradient technology in combination with rolling table platforms.^{6 ,17} Whole-body imaging is achieving clinical relevance by providing tumor staging in a single approach.^{5 ,24 - 25}

Whole-body CT imaging represents another option for tumor staging in a single session. However, CT alone has been found to be inferior to MRI for assessing parenchymal organs as well as osseous structures.^{26 - 27} Furthermore, previous studies have demonstrated reduced sensitivities and specificities in staging lymph nodes when compared with PET imaging.^{28 - 29} Thus, whole-body CT alone was not included in our comparison.

T-staging was more reliable with FDG-PET/CT compared with MRI. The inferior performance of MRI reflects the more limited ability to differentiate between viable tumor and adjacent structures when imaging is based on morphological characteristics alone.^{26 ,29} Results from this study correspond well to other studies, which have demonstrated accurate T-staging in only 54% with MRI.³⁰ By integrating function and morphological imaging, PET/CT outperforms MRI. Preliminary data have revealed a similar benefit of dual-modality PET/CT compared with PET or CT alone when assessing the T-stage.^{11 ,13} Similar to T-staging, FDG-PET/CT proved superior to whole-body MRI in regional lymph node staging. PET-based functional analysis is known to be more accurate in assessing lymphatic spread compared with morphological imaging based predominantly on size criteria.^{28 ,31} Limitations associated with size criteria are well-known. They are based on false-positive findings in lymph nodes enlarged due to inflammation, or false-negative findings in normal-sized lymph nodes harboring micrometastases.^{32 - 33} Sensitivity and specificity values of 79% and 78%, respectively, lie well within the range of previously published data for MRI detection of malignant lymph nodes.^{34 - 35} The introduction of new contrast agents based on ultrasmall superparamagnetic iron oxides (currently undergoing phase 3 clinical evaluation) may enhance the performance of MRI. Initial results report sensitivities ranging between 64% and 100% and specificities of between 50% and 94%.^{36 - 37}

PET/CT and MRI were similarly accurate when assessing M-stage. Magnetic resonance imaging has been known to be less sensitive than CT when assessing the lung for pulmonary lesions. Furthermore, recent findings by Pastorino et al³⁸ have shown an increase in diagnostic accuracy when adding PET imaging to indeterminate CT findings in patients with pulmonary nodules of unknown dignity. Additional availability of the PET data contributed to differentiating benign from malignant lesions in 43% of patients with equivocal findings on CT. Based on this synergetic property of CT and PET, superiority of PET/CT over MRI can be expected when evaluating metastases to the lung. On the other hand, MRI proved more reliable when assessing the liver for metastases. This finding is in contrast to other studies comparing PET imaging with MRI.^{9 ,39} The higher performance of MRI in detecting liver lesions may be explained by the optimized MRI protocol providing unenhanced as well as contrast-enhanced coverage of the liver in 3 different phases. Furthermore, hepatic metastases missed on PET/CT were all found to be below 1 cm in diameter. Increased glucose metabolism in smaller lesions may remain undetected on PET due to partial volume effects caused by respiratory motion.⁴⁰ Not even the integration of diagnostic CT data into the dual-modality concept could compensate for this limitation. These findings are corroborated by other published data, which illustrated CT imaging to be less sensitive and specific than MRI when assessing the liver for metastases.²⁷

Differences between MRI and PET regarding the detection of bone metastases have been controversial. While some researchers have demonstrated an advantage of PET over MRI,⁴¹ others report the opposite.⁴² The type of malignancy may serve as an explanation for this controversy as some tumors are frequently found to be FDG-PET negative. In our study, however, PET/CT failed to detect bone metastases in 2 patients with bronchial carcinoma and 1 patient with pleural mesothelioma. In all 3 patients, the primary lesion was PET-positive. When considering the overall TNM stage, PET/CT incorrectly determined the TNM stage compared with the standard of reference in 23 patients. In only 6 of these patients, the primary tumor was FDG-PET negative, while the other tumors were FDG-PET positive (mean standardized uptake value: 8.4). Thus, incorrect staging results in the majority of patients cannot be attributed to tumor histological characteristics (FDG-PET–negative tumor cells). Insensitivity to micrometastases may be improved by increasing scanner resolution as well as implementation of new, more specific radioactive tracers.

There are some technical limitations to image fusion in combined PET/CT. Limitations are based on differences in respiration state between the CT and PET components as well as effects of contrast agents on the CT-based PET attenuation correction. Osman et al⁴³ have found inaccurate localization of lesions close to the diaphragm caused by respiration-induced CT and PET misregistration. In this study, however, the CT component of the combined PET/CT was acquired with patients breathing shallowly, which caused respiration mismatch between CT and PET. Based on the use of a special breathing protocol,¹⁶ respiration-induced misregistration was minimized in the current imaging protocol. Other clinicians have reported artefacts on PET images due to application of CT contrast agents.^{44 - 46} However, these artefacts are rare and considered to have only a minor impact on PET image quality and image assessment. Artefacts can be avoided by performing the CT without contrast agents, but nonenhanced CT images cannot be considered fully diagnostic. Because artefacts only appear on attenuation-corrected PET images, PET/CT images were assessed with and without PET attenuation correction in the current study.

Of course, decision on either 1 of the 2 imaging modalities will be influenced by local availability of the 2 imaging procedures. Up-to-date MRI systems have been more widely available than PET/CT scanners. However, high-resolution whole-body MRI requires movable table platforms, which are still not commercially available from all manufacturers. Considering the cost of the examinations and time requirements associated with image acquisition, there were no major differences between the 2 imaging procedures. In the current hospital setting, approximately $1900 is charged for a whole-body PET-CT and $1850 for a whole-body MRI. However, these numbers may vary substantially from country to country.

The most crucial aspect of clinical tumor staging relates to the staging impact on patient management. Compared with whole-body MRI, the therapy regimen was altered in a substantially larger number of patients when staging analysis was based on the PET/CT data. Therefore, FDG-PET/CT can be recommended as a first-line tool for whole-body tumor staging of different oncological diseases. However, MRI with a whole-body field of view may complement the PET/CT data if detection of liver and osseous metastases is deemed crucial. In patients with known history of adverse reactions to iodine-based intravenous contrast agents as well as patients with impaired renal function, a contrast-enhanced whole-body MRI might serve as an alternative to an unenhanced PET/CT.