|Home | About | Journals | Submit | Contact Us | Français|
Pretreatment multimodality imaging can provide useful anatomical and functional data about tumors, including perfusion and possibly hypoxia status. The purpose of our study was to assess non-invasively the tumor microenvironment of neck nodal metastases in patients with head and neck (HN) cancer by investigating the relationship between tumor perfusion measured using Dynamic Contrast Enhanced MRI (DCE-MRI) and hypoxia measured by 18F-fluoromisonidazole (18F-FMISO) PET.
Thirteen newly diagnosed HN cancer patients with metastatic neck nodes underwent DCE-MRI and 18F-FMISO PET imaging prior to chemotherapy and radiation therapy. The matched regions of interests from both modalities were analyzed. To examine the correlations between DCE-MRI parameters and standard uptake value (SUV) measurements from 18F-FMISO PET, the non-parametric Spearman correlation coefficient was calculated. Furthermore, DCE-MRI parameters were compared between nodes with 18F-FMISO uptake and nodes with no 18F-FMISO uptake using Mann-Whitney U tests.
For the 13 patients, a total of 18 nodes were analyzed. The nodal size strongly correlated with the 18F-FMISO SUV (ρ=0.74, p<0.001). There was a strong negative correlation between the median kep (ρ=−0.58, p=0.042) and the 18F-FMISO SUV. Hypoxic nodes (moderate to severe 18F-FMISO uptake) had significantly lower median Ktrans (p=0.049) and median kep (p=0.027) values than did non-hypoxic nodes (no 18F-FMISO uptake).
This initial evaluation of the preliminary results support the hypothesis that in metastatic neck lymph nodes, hypoxic nodes are poorly perfused (i.e., have significantly lower kep and Ktrans values) compared to non-hypoxic nodes.
The use of multimodality (MM) imaging is rapidly gaining acceptance in oncology for the study of tumor microenvironment characteristics, including tumor vascular/perfusion status and tumor cell hypoxia, which have been shown to be related to treatment outcomes (1). Multimodality imaging allows the acquisition of both functional and anatomical images by exploiting the unique strengths of different imaging techniques (1). The present study focuses on the application of MM imaging techniques using gadopentetate dimeglumine (Gd-DTPA)-based dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) and 18F-fluoromisonidazole (FMISO) positron emission tomography (PET) in studying neck nodal metastases.
DCE-MRI is a useful noninvasive method for characterizing the pathophysiological microenvironment of tumors (2). DCE-MRI involves assessing changes in signal intensity over time. With proper compartmental modeling, using models like the Brix-Hoffmann model (3), the modified analysis by Port et al. (4), or the Tofts model (5), the data may yield results on tumor-vessel permeability, tumor perfusion, and extracellular-extravascular volume fraction, i.e., data related to the tumor microenvironment (6). On the other hand, hypoxia imaging has been most extensively studied with 18F-FMISO PET in the clinical setting (7, 8). This fluorinated radiotracer was developed from the nitroimidazole compound misonidazole as a hypoxic cell radiosensitizer (9). It is thought to be a surrogate imaging marker for hypoxia (7, 8, 10).
DCE-MRI and 18F-FMISO PET are typically acquired and assessed separately. In a recent study of an animal tumor model, Cho et al. (11) found a negative correlation between perfusion as assessed by DCE-MRI and hypoxia as measured by late 18F-FMISO PET uptake, hence providing the evidence for the hypothesis that poorly perfused tumors are hypoxic (12). The study also established the utility of combined MRI and PET measurements for imaging the tumor microenvironment and identifying regions of tumor hypoxia and well-perfused tissue (11).
Clinical studies in head and neck (HN) cancers have assessed perfusion or hypoxia separately (7, 8, 13–18). Hoskin et al. performed DCE-MRI on 13 HN cancer patients before and after radiotherapy and suggested that treated patients in whom tumors have high MR signal enhancement might be considered for dose escalation (14). Rasey et al. found marked intra- and inter-tumor variability of hypoxia and emphasized that complete, non-invasive hypoxia measurements in individual tumors are important for patient-specific treatment planning (8). Recently, Lee et al., using the same radiotracer, illustrated the feasibility of dose escalation of the gross tumor volume in HN cancer patients (7).
Several investigators have reported tumor oxygenation (using invasive Eppendorf pO2 measurements) in HN cancers to be associated with poor outcome (19, 20). Nordsmark et al. performed pre-treatment measurements of tumor oxygen tension in a multicentric trial with 397 HN cancer patients and showed that tumor hypoxia is associated with a poor prognosis (20). Thus, the detection and quantification of hypoxia might provide a prerequisite for the clinical implementation of hypoxia-directed therapies that could potentially improve outcomes in HN cancer patients (21). Although often considered the gold standard, invasive Eppendorf pO2 histography has disadvantages as it is limited by sampling, unable to differentiate hypoxic and necrotic tissue, and cannot access all tumors. In contrast, 18F-FMISO PET is only sensitive to the presence of hypoxia in viable cells, can cover the entire region of interest and is well tolerated by patients (22). The imaging techniques Gd-DTPA-based DCE-MRI and 18F-FMISO PET have been proposed separately as promising tools for predicting treatment response for HN cancer (7, 8, 14, 17). As yet, however, no clinical studies have compared data from these modalities for the assessment of the tumor microenvironment. The purpose of our study was to assess non-invasively the tumor microenvironment of neck nodal metastases in patients with HN cancer by investigating the relationship between tumor perfusion measured using DCE-MRI and hypoxia measured by 18F-FMISO PET.
The institutional review board (IRB) granted a waiver of informed consent for this retrospective study that included thirteen patients with metastatic lymph nodes who underwent 18F-FMISO PET (IRB No. 04-070) and DCE-MRI as part of clinical MRI. Our study was compliant with the Health Insurance Portability and Accountability Act. PET examinations were performed 6±9 (mean ± SD) and 3±9 days after MRI for 18F-FMISO and 18F-fluorodeoxyglucose (18F-FDG), respectively in these 13 HN cancer patients with neck nodal metastasis before beginning chemo-radiation therapy. The imaging details for MRI and PET/CT examinations have been described previously (7, 23). All patients underwent needle biopsy at the primary tumor location (14±11 days before MRI). To ensure that the tumor microenvironment would be unchanged, biopsies of the nodes were not done. Patient characteristics are listed in Table 1.
For simplicity, we have referred to 18F-FMISO as FMISO and18F-FDG as FDG for the rest of this paper.
MRI data were acquired on a 1.5 Tesla G.E. Excite scanner (General Electric, Milwaukee, WI) with a 4-channel neurovascular phased-array coil for signal reception and a body coil for transmission. The study consisted of MRI covering the entire neck (23). The MR acquisition parameters for the neck survey were as follows: rapid scout images, multiplanar (axial, coronal and sagittal) T2-weighted, fat-suppressed, fast-spin echo images (TR = 5000 ms, TE = 102 ms, averages = 2, slice thickness = 5.0 mm and gap = 2.5 mm), and multi-planar T1-weighted images (TR = 675 ms, TE = 8 ms, averages = 2, slice thickness = 5.0 mm and gap = 2.5 mm). Standard T1- and T2-weighted imaging were followed by proton density MR imaging acquired (for the purpose of determining the longitudinal relaxation rate constant R1 for each DCE-MRI data point) in the axial plane (TR = 350 ms, TE = 2 ms with a 30° flip angle (α), 2 excitations, 15.63-kHz receive bandwidth, field of view (FOV) = 18–20cm, slice thickness = 5–6 mm, zero gap and a 256×128 matrix). This was followed by axial T1-weighted DCE-MR imaging (described below) and then post-contrast T1-weighted imaging in the axial and coronal planes.
Dynamic perfusion studies of the nodes were acquired using a fast multi-phase spoiled gradient echo sequence. Antecubital vein catheters delivered a bolus of 0.1 mmol/kg Gd-DTPA (Magnevist; Berlex Laboratories, USA) at 2 cc/s, followed by saline flush using an MR-compatible programmable power injector (Spectris; Medrad, Indianola, PA, USA). The entire node was covered contiguously with 5–7 mm thick slices, zero gap, yielding 3–6 slices with 3.75–7.5 sec temporal resolution. Acquisition parameters were similar to those described above for proton density imaging except that the TR = 9 ms and 40–80 time course data points were collected. For both proton density images and DCE-MRI the 256 × 128 matrix was zero filled to 256 × 256 during image reconstruction.
All PET scans were performed on a GE Discovery LS PET/CT scanner (GE Medical Systems, Waukesha, WI) consisting of an Advance NXi PET and a LightSpeed CT unit (7). For the PET studies, patients were scanned in the supine position and immobilized using an immobilization device, an Aquaplast mask (Orfit Industries, Wijnegem, Belgium) that extended to the shoulders (7). Markers were used on the flat insert to ensure proper patient positioning (7).
Before the FDG PET examination, patients fasted for at least 6 hours, but liberal water intake was allowed. An activity of approximately 15 mCi of FDG was then injected intravenously. After a 45- to 60- minute uptake period, a PET/CT study was acquired on the GE Discovery LS PET/CT with the patient in the same treatment position as described by Lee et al. (7) Fluorine-18–labeled fluoride was produced by the Memorial Sloan-Kettering Cancer Center/Cornell Cyclotron Radiochemistry Core Facility by proton irradiation of an enriched 18O water target in a small-volume niobium chamber with an approximate specific activity of 1–2 Ci/μmol at the end of the bombardment time and tested for quality assurance. The details for PET imaging to assess tumor uptake and biodistribution are described by Lee et al. (7). Blood samples were obtained immediately before and after the PET/CT session. No fasting state was required for the patients. FMISO was injected into the patients approximately 120–150 minutes before scanning. Each of the FMISO studies covered two PET axial FOVs at the tumor position, at 8 min/FOV. The average duration for all patients between FMISO injection and PET/CT acquisition was 162 min. The FMISO emission data were corrected for attenuation, scatter, and random counts and subsequently reconstructed (7).
MRI data was analyzed with IDL 5.4 (Research Systems Inc., Boulder Co). For the tumor tissue time course data, region of interests (ROIs) were manually drawn by an experienced neuroradiologist (> 10 yrs of experience) outlining the contrast-enhanced neck nodes for signal intensity measurements. All the slices containing the tumor were outlined and analyzed. The total number of pixels within the entire ROI was converted into the tumor volume (mm3) in MatLab (The MathWorks). Out of 13 patients, 4 patients had more than one node (ROI). The exact location of each ROI per patient was noted and tabulated by the neuro radiologist to facilitate the matching of the ROIs later between the MRI and PET studies.
Quantitative DCE-MRI analyses of the tumor tissue time course data was performed using the two-compartment Tofts model in all ROIs (5). A population-based arterial input function was used (23) that was derived from the carotid arteries in head and neck patients. This approach included a phantom study to generate a calibration curve of signal intensity ratio of T1-weighted image over proton density image versus T1 (23). With this calibration curve, it was possible to convert signal intensities into Gd-DTPA concentrations. The model fitted the tissue contrast agent concentration, Ct(t), time course for the extraction of the Ktrans and ve parameters from the whole tumor (multiple slices), as shown in the following Kety-Schmidt method-related equation:
The analyses yielded the pixel Ktrans (the volume transfer constant between the plasma and the extravascular extracellular space (EES) in min−1), ve (the volume fraction of the EES which is dimensionless and also referred to as interstitial space), and kep (the rate constant describing the contrast transfer between the EES and plasma in min−1, which equals the ratio Ktrans/ve). Quantitative DCE-MRI analyses of the tumor tissue time course data was done in all the patients per ROI on a pixel-by-pixel basis. A histogram analysis was performed on all pixels within the ROI, which yielded median, standard deviation, and skewness of the distribution of all pixels. Histograms were normalized to the total number of tumor voxels to allow direct comparison between patients. The median is a number separating the higher half of the population from the lower half. The standard deviation describes the width of the distribution, and the skewness characterizes the asymmetry of the distribution.
All data were reviewed on a standard clinical workstation (GE PACS with AW extension). PET images were first reviewed in 3 orthogonal planes (transaxial, coronal, sagittal) and a maximum-intensity projection image. Afterward, the CT, PET, and PET/CT fusion images were displayed simultaneously. One board-certified nuclear medicine physician with > 10 years of experience in PET reviewed all PET/CT studies. The nuclear medicine physician was provided with the excel sheet that had the MRI location for each node (ROI) studied and he matched the ROI’s from the MR images to the PET/CT images. Matched ROIs were analyzed based on visual and semiquantitative analysis, using the attenuation-corrected PET emission images. For semi-quantitative analysis, regions of interest were placed over the areas of focal FDG uptake in the neck nodal metastases. The intensity of FDG uptake in these regions was measured using the standardized uptake value (SUV), normalized to body weight. The maximum-pixel SUV was recorded. The imaging data initially available in units of microcuries per milliliter per voxel were decay corrected to the time of injection and converted into SUV units. For FMISO, semi-quantitative analysis included calculation of tumor-to-blood ratios for hypoxic tracer uptake and conversion to SUV measurements. Additionally, FMISO uptake by the tumor was graded using a dichotomous score: no uptake (score 0) or moderate-severe uptake (score 1).
To examine the correlations between DCE-MRI parameters and SUV measurements from FMISO PET, the non-parametric Spearman correlation coefficient (ρ) was calculated. The same Spearman test was used to examine the correlation between SUVs of FMISO and FDG PET. The correlations were interpreted using the guidelines from Cohen et al (24), with absolute correlations of <0.3 considered weak, 0.3–0.5 considered moderate, and 0.5–1.0 considered strong. This study is aimed at relating DCE-MRI parameters with PET (as mentioned above) but not at assessing associations of parameters obtained from individual examinations with each other (such as all individual DCE-MRI parameters). DCE-MRI parameters were compared between nodes with FMISO uptake and nodes with no FMISO uptake using 2-sided non-parametric Mann-Whitney U tests. For all statistical tests (p<0.05) denotes significance.
All 13 patients were untreated at the time of imaging, and all had biopsy proven squamous cell carcinoma. For the 13 patients, a total of 18 lymph nodes were analyzed, as 4 patients had more than one metastatic node (Table 1). The 18FDG SUV for the 18 lymph nodes was 8.8±5.7 (mean±SD). The nodal size measured on MRI strongly correlated with the FMISO SUV (ρ= 0.74, p< 0.001) but not with DCE-MRI parameters, i.e median Ktrans, ve and kep (p> 0.5 for all three parameters). Table 2 shows the correlations between FMISO SUV and DCE-MRI parameters. There was a strong negative correlation (ρ= −0.58, p= 0.042) between the median kep and the FMISO SUV. Additionally, the association between the FDG SUV and FMISO SUV was statistically significant (p< 0.01), with a strong Spearman correlation of ρ= 0.70.
Figures 1 and and22 show MRI and PET images of a patient with a large hypoxic node and a patient with a small non-hypoxic node, respectively. The hypoxic node in 1G shows substantial FMISO uptake, whereas the non-hypoxic node in 2G displays FMISO distribution comparable to the background noise (i.e., no substantial uptake). Furthermore, the slope of the Gd-DTPA uptake is steeper in the non-hypoxic node (2F) than in the hypoxic node (1F). Table 3 allows comparison of the lymph nodes’ hypoxic characteristics (derived from FMISO PET exams) with their quantitative DCE-MRI parameters. A Mann-Whitney U test showed that hypoxic lymph nodes were significantly larger in size (mean±SD: 148±119 mm3) than non-hypoxic nodes (35±52 mm3). It also showed that hypoxic nodes had significantly lower median Ktrans (p= 0.049) and median kep (p= 0.027) values than did non-hypoxic nodes. Figure 3 displays the average distribution plots of the calculated Ktrans (3A) and kep (3B) parameters for the hypoxic (black bars) and non-hypoxic nodes (white bars). The histograms depicting hypoxic nodes show significant shifts to lower Ktrans (3A) and kep (3B) values, indicative of lower perfusion in hypoxic nodes. Furthermore, although the kep data points are spread out more to the right of the mean than to the left for both the hypoxic and the non-hypoxic nodes, this asymmetry, as determined by the kep skewness, is significantly larger in the hypoxic nodes (p= 0.021). Finally, Figure 4 depicts box plots of the average Ktrans (4A) and kep (4B) values for hypoxic and non-hypoxic nodes. For both Ktrans and kep, hypoxic nodes have significantly smaller values (p< 0.05), as indicated by a Mann-Whitney U test.
Our study investigated whether pre-therapy DCE-MRI parameters correlate with hypoxia as measured by FMISO PET in neck nodal metastases in 13 newly diagnosed HN cancer patients. The results of our study support the hypothesis that hypoxic nodes are poorly perfused compared to nodes without hypoxia (12). A negative correlation between FMISO uptake (measure of hypoxia) and the median kep perfusion value was observed, and hypoxic nodes had lower median kep and median Ktrans values than did non-hypoxic nodes. Furthermore, hypoxic nodes had a more asymmetric distribution of kep values than did non-hypoxic nodes. Finally, tumor size strongly correlated with FMISO uptake, which is in agreement with the results of a previous study by Dunst et al. (25), which showed a strong correlation between total tumor volume and hypoxic volume in 125 patients with HN squamous cell cancer.
Tumor progression requires a vascular network adequate to meet the metabolic needs of growing tumor tissue. As local tissue perfusion decreases, hypoxia develops because of an insufficient vascular supply of oxygen to the growing tumor mass. It has been shown that the presence of tumor hypoxia is an established vital factor in tumor progression and in resistance of tumors to therapy (26). Well-oxygenated cells are more sensitive to the cytotoxic effects of ionizing radiation compared with poorly oxygenated cells (12). Hypoxia can also cause genome changes that select cells that are resistant to apoptotic signals, thus favoring the proliferation of resistant cell populations and promoting the development of cells with more aggressive phenotypes (27). Therefore, hypoxia in tumor tissue is an important prognostic indicator of chemotherapy and/or radiation therapy response (28).
To the best of our knowledge, ours is the first clinical study in patients with metastatic neck lymph nodes which indicates that noninvasive DCE-MRI and FMISO PET measurements of tumor perfusion and hypoxia, respectively, are inversely correlated. Several studies in the literature are in agreement with our results, although the techniques and subjects used in these studies differ, in some cases substantially. Our group recently published a study comparing perfusion and hypoxia parameters derived from DCE-MRI studies and from early FMISO uptake PET studies, respectively, in a rat prostate tumor model (11). The results suggested that tumor vasculature was a major determinant of early FMISO uptake. Furthermore, a negative correlation between the DCE-derived perfusion and late FMISO uptake was observed. These preclinical results are in agreement with the results of the present clinical study.
Scholbach et al. (29) recently performed a clinical study, in which the metastatic lymph nodes of 24 HN cancer patients were assessed for tissue oxygenation and perfusion using direct polarography and ultrasound, respectively. Although the assessment techniques used were inherently different, the end result was remarkably similar: an inverse relationship between hypoxia and tissue perfusion was found (29).
Our study also found that FMISO uptake correlated with FDG uptake, suggesting that in pretreatment HN metastatic nodes, hypoxic conditions leads to increased glucose uptake. These results are in accordance with other previous clinical HN studies (16, 30, 31). However, Rajendran et al. (16) examined HN cancers, soft tissue sarcomas, breast cancer and glioblastoma multifore and found that highly metabolic tumors are not necessarily hypoxic. Therefore, FDG uptake should not be considered a surrogate marker for hypoxia (16). This will be the subject of future studies.
Firstly, the major limitation of our study was that a pixel by pixel analysis of DCE-MRI and FMISO PET images was not performed. However, it is important to note that DCE-MRI was performed as part of a diagnostic MRI examination yielding high resolution MR images and not as part of a radiation treatment planning study which uses exact matching with the PET study but provides lower resolution images. PET studies utilized an Aquaplast immobilization mask and a specific head rest. This led to different patient orientations during the two imaging (MRI and PET) examinations. For a pixel-by-pixel comparison, the factors that affect patient positioning would need to match each other exactly for the MRI and the FMISO PET examinations. These issues of image co-registration would be resolved best by the use of combined MRI/PET systems, which are currently entering the clinical arena (32). Secondly, it has previously been shown that there are potential limitations with single time-point FMISO imaging, which have led to the proposal of a dynamic scanning approach (11, 33). In such an approach, the kinetic information of the tracer uptake and retention behavior in each PET image voxel can be analyzed using a mathematical compartment model, similar to the ones used for DCE-MRI, to generate parametric maps of putative hypoxia tracer trapping (33). Thirdly, FMISO PET images typically have low signal-to-noise characteristics as uptake of FMISO in tumors is relatively low. Therefore, the contrast between healthy tissue and tumors is typically low (34).
For future studies, multimodality imaging using DCE-MRI and 18F-FMISO PET should ensure that patient positioning is the same as in PET, to enable a pixel-by-pixel co-registration that will help in radiation treatment planning.
Our preliminary results support the hypothesis that in metastatic neck lymph nodes, hypoxic nodes are poorly perfused (i.e., have significantly lower kep and Ktrans values) compared to non-hypoxic nodes.
This study was conducted with support from National Cancer Institute/National Institutes of Health (grant number 1 R01 CA115895). We thank Dr Sadek A. Nehmeh for his help in FMISO studies.
Conflict of Interest Notification: All authors in this manuscript have no conflict of interest regarding the study.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.