The clinical applications of 18F-2-fluoro-2-deoxy-d-glucose (FDG) positron emission tomography (PET) are continually expanding, especially in the field of oncology. In colorectal cancer (CRC), the diverse uses of PET include initial diagnosis, staging, restaging, and assessment of the therapeutic response [1–3]. PET has also been reported to offer advantages over conventional, anatomically based morphologic modalities for detecting recurrent CRC and metastatic disease, because of its capacity to provide a functional image and evidence of tumor behavior [4, 5]. This is based on the knowledge that enhanced glucose uptake is one of the major metabolic changes characteristic of malignant tumors. Clinically, FDG is the most commonly used positron-emitting radiotracer. It is metabolized similarly to glucose, being transported into the cell, but once enzymatically phosphorylated, FDG-6-phosphate is metabolically trapped in tumor cells. Thus, tumors demonstrate increased emission of positron from FDG and can be distinguished on PET scan images by areas of increased tracer activity [6].

Clinically, variable FDG uptake, semiquantitated as the standardized uptake value (SUV), has been seen on PET scans of tumors from the same origin, including CRC. Much research has been done on the differences in FDG uptake among tumors and the mechanism of this uptake. Emerging evidence indicates that the factors affecting FDG uptake are complicated because the specific biological characteristics of tumors determine the degree of glucose metabolism [7–9]. Most factors affecting FDG uptake, such as hypoxia and cell density, are thought to be associated with changes in glycolysis-related protein expression [10, 11]. The expression of glucose transporter proteins, especially GLUT-1, which is directly involved in FDG uptake, is thought to determine the levels of FDG uptake in cancer cells [12–14].

In the present study, we examined the association between FDG uptake and clinicopathologic features and GLUT-1 expression in patients with CRC. We investigated the factors determining FDG uptake and attempt to explain the SUV variability measured by PET.

Patients and methods

Patients

The subjects of this study were 20 patients with histologically proven colorectal adenocarcinoma treated at the Department of Surgery and Clinical Oncology, Graduate Medical School, Osaka University, between 2000 and 2001. The disease was graded into clinical stages 0–IV, according to the criteria of the International Union Against Cancer. Synchronous liver metastasis was detected in 1 patient. All patients underwent FGD PET before surgery. This study was approved by the institutional review board, and written informed consent was obtained from the patients before inclusion. Patients with underlying inflammatory bowel disease and diabetes were excluded because of the potential diagnostic overlap in PET findings. Patient characteristics and tumor status are shown in Table 1. None of the patients were given preoperative chemotherapy or irradiation.

Table 1 Patient characteristics and tumor status
Full size table

FDG-PET procedures

After fasting for at least 4 h, the patients had blood collected for serum glucose measurement, and were then given approximately 370 MBq of FDG IV. All patients were normoglycemic (blood glucose level, 101.4±13.6 mg/dL−1). Simultaneous emission–transmission PET scans were acquired 1 h after FDG injection with a dedicated PET scanner (Headtome V/SET 2400W, Shimadzu Corporation, Kyoto, Japan) and a rotating external transmission source (68Ge-68Ga rod source). This system has 32 rings and simultaneously produces 63 slices 3.125 mm thick along a 20-cm longitudinal field. The intrinsic resolution was 3.7 mm full width at half maximum, and the sensitivity of the device was 7.3 cps/Bq cm−3. The bladder was continuously flushed with 1000 mL of saline via a triple-lumen catheter [15, 16].

Image analysis

Images were reconstructed with an iterative median root prior to a reconstruction algorithm (mask size 3×3, β 0.3, subsets 24, iteration 1). For visual analysis, 2 experienced nuclear medicine physicians blinded to the clinical data of the patients read the FDG images on a high-resolution computer screen. Areas with focally increased FDG uptake compared with the surrounding tissue were read as positive. The final diagnosis was made by consensus between the 2 observers. Subsequently, all positive sites were compared with the known location of the tumor according to the available morphologic imaging data (endoscopy, computed tomography). For semiquantitative analysis of the FDG uptake, irregular regions of interest (ROIs) were semi-automatically placed over the transaxial slices of the PET images. In each consecutive slice, the ROI was adjusted manually to clearly reveal the boundary enclosing all pixels with an SUV ≥2.0, being the threshold between malignant and benign colorectal tumors in our university. We measured the mean counts per pixel within all the ROIs. SUVmean, normalized for the body weight, was calculated as the formula: SUVmean=(PET counts×calibration factor)/(injected dose/body weight). The effects of partial volume were corrected by recovery coefficients derived from phantom studies simulating lesions of various sizes and uptake values.

Immunohistochemical staining for GLUT-1 expression

The resected surgical specimens, consisting of 20 CRCs and 1 liver metastatic focus, were fixed in formalin, then processed through graded ethanol, and embedded in paraffin. Sections 4 μm thick were mounted on Capillary Gap Plus microscope slides (BioTek Solutions, Santa Barbara, CA), which were deparaffinized in xylene and rehydrated. Immunostaining was done on a TechMate Horizon automated staining system (DAKO, Glostrup, Denmark), as described previously [17]. Briefly, sections were subjected to heat antigen retrieval in 10 mmol citrate buffer (pH 6.0); then, after blocking with 1% bovine serum albumin, they were incubated with anti–GLUT-1 polyclonal antibody (DAKO, Carpinteria, CA) overnight at a dilution of 1:500. After washing, the sections were incubated with biotinylated goat anti-rabbit secondary antibody, and then subjected to the streptavidin-biotin complex method (Histofine SAB-PO kit; Nichirei, Tokyo, Japan). We used 3, 3-diaminobenzidine plus H2O2 for visualization of the signals in brown. Sections were then counterstained with hematoxylin. For a negative control, sections were incubated with normal rabbit IgG (Vector Laboratories, Burlingame, CA) or phosphate-buffered saline instead of the primary antibody, as a substitute for the primary antibody, to exclude false-positive responses from nonspecific binding to IgG or from the secondary antibody.

Fig. 1
figure 1

A coronal slice of sigmoid colonic cancer shows high accumulation of 18F-2-fluoro-2-deoxy-d-glucose (FDG) (arrow)

Full size image

Intensity of GLUT-1 staining

The intensity of GLUT-1 immunoreactivity was categorized as none, faint, or strong by averaging the reaction intensity compared with that of tissue erythrocytes, which were used as an internal control [12]. We randomly selected 10 visual fields and counted the number of strongly immunoreactive tumor cells and all tumor cells per square millimeter under a microscope at 200× magnification. For statistical analysis, we used the number of strongly stained tumor cells, representing the intensity of GLUT-1 expression. Immunohistochemical analysis for anti–GLUT-1 antibody was independently performed 3 times by 2 experienced pathologists who were unaware of the SUVs and clinical data.

Statistical analysis

Statistical analysis was performed using StatView J-5.0 software (Abacus Concepts, Berkeley, CA). Data are expressed as mean values±standard deviations. Associations between discrete variables were assessed using Fisher's exact test. Mean values were compared using the Student's t-test. Bonferroni's correction was applied for multiple comparisons. Correlation significance was assessed using Pearson's correlation coefficient test. P<.05 was considered to indicate that a given correlation was significant.

Results

PET findings

FDG-PET showed increased focal activity in all of the 20 primary CRCs and in the liver metastatic focus. Figure 1 shows the intensive accumulation of FDG in the sigmoid colon in a typical PET scan of a patient with CRC. The central part of the primary CRC tumor (Fig. 2a) and liver metastasis (Fig. 2b) had a higher SUV than the peripheral areas. Regional lymph node involvement in 7 patients was not detected by PET.

Fig. 2
figure 2figure 2

The standardized uptake values (SUVs) increased progressively from the periphery to the central part in both the primary colorectal carcinoma (a) and in liver metastatic lesions (b). As the 18F-2-fluoro-2-deoxy-D-glucose (FDG) uptake increased, the color indicating tracer uptake value changed from green (periphery) to yellow (intermediate part), and to red (center of the focus) in the Shimazu color scale system

Full size image

Intensity of GLUT-1 expression

Intravascular red cells, which were stained strongly in all tissue sections, served as internal controls. Immunoreactivity for GLUT-1 was localized in the cytoplasm of cancer cells and staining intensity was categorized as none, faint, or strong (Fig. 3). GLUT-1 immunostaining revealed positive (faint and strong) staining in 17 (85%) of the 20 primary lesions and in the liver metastasis. The intensity of GLUT-1 staining was stronger in the central part of the metastatic focus (Fig. 4).

Fig. 3
figure 3

The intensity of GLUT-1 immunoreactivity was categorized as none, faint, or strong by averaging the reaction intensity compared with that of tissue erythrocytes (arrow), used as an internal control. Original magnification ×100

Full size image
Fig. 4
figure 4

(a) Liver metastatic focus and normal liver tissue. The intensity of GLUT-1was stronger in the intermediate part (b) than in the peripheral part (c) of the lesion. Original magnification: a ×20; b, c ×100

Full size image

SUV and clinicopathologic characteristics

Based on the cutoff value of 8.0 (3.1–11.9), patients were divided into a high SUV group (SUV>8.0; n=8) or a low SUV group (SUV < 8.0, n=12). The SUVs were correlated with tumor size (P=.0004) and invasive grade (P=.042; Table 2).

Table 2 Relationship between standardized uptake value (SUV) and clinicopathologic parameters
Full size table

SUV and GLUT-1 expression

A strong correlation was found between SUV and the cellularity (per mm2) of cancer cells displaying strong GLUT-1 staining (Fig. 5).

Fig. 5
figure 5

Relationship between the standardized uptake value (SUV) and expression of GLUT-1 in colorectal carcinoma. A positive correlation was found between the SUV and the number of cells stained strongly with GLUT-1 among the 20 samples of colorectal carcinoma examined

Full size image

Discussion

The use of SUV as a semiquantitative index of regional tracer uptake in PET imaging is practical and widespread. However, SUV measurements are subject to partial-volume effects, duration of tracer uptake, blood glucose level at the time of tracer injection, and ROI effects. To account for these effects, we standardized the measurement of SUVs from 60 min after tracer administration, applied recovery coefficients, and excluded patients with diabetes. All of the patients had a blood glucose level <150 mg/dL−1 before tracer injection. Instead of the SUVmax, we used the SUVmean, which sampled multiple points within the lesion. A threshold method made the ROI selection reliable.

The findings of this study confirmed the high sensitivity of PET scanning in assisting with the detection of CRC. All of the primary carcinomas as well as the liver metastatic focus were seen on PET scans. Previous studies have also shown that PET can accurately identify the site of the primary tumor, with sensitivities ranging from 90–100% [18]. The overall sensitivity for detecting liver metastasis is as high as 99% [19].

A better understanding of the association between FDG uptake and clinicopathologic features will provide more information about the tumor and lead to a clearer interpretation of PET imaging. We found a significant correlation between several tumor characteristics and FDG uptake, which may explain the variability in PET scanning results among patients with CRC. Contrary to reports on gastric cancer [20, 21], we found no association between the SUV and regional lymph nodes metastases or TNM stage, in accordance with previous reports on CRC. In a small series of 48 patients with known or suspected primary CRC, Abdel-Nabi et al. [18] found that FDG-PET imaging identified all primary carcinomas, but it was ineffective for detecting local lymph node involvement. N-staging requires numeric assessment of pericolic and mesenteric nodes; however, the regional lymph nodes in colorectal carcinoma are often small and in close proximity to the primary tumor mass. Furthermore, the pericolic nodes often contain micrometastatic cancer cells, recognized only at the time of histopathologic evaluation. For these reasons, the sensitivity of PET for regional lymph node metastases from CRC was found to be only 29% [18].

In the present study, SUV was significantly related to tumor size (P=.0004) and depth of invasion (P=.042), which is in agreement with most clinical studies on various malignancies [7, 12, 20–22]. We hypothesize that this is because the number of tumor cells in the mass augments in parallel with an increase in the tumor size. According to a study on breast cancer, FDG uptake was affected by cellularity [9], although in general, the number of viable tumor cells per gram of tissue correlates linearly with the mass size. Therefore, the need for glucose, which determines FDG uptake or SUV, correlates with the tumor size; however, there is 1 exception. In the present study, a mucinous adenocarcinoma had the lowest SUV (3.1), even though its diameter was not the smallest. It has been suggested that the sensitivity of FDG-PET for detecting mucinous adenocarcinoma is lower than that for detecting nonmucinous adenocarcinoma [20, 21]. In a retrospective review of 22 patients with mucinous carcinoma, FDG-PET showed the mucinous carcinoma in only 13 (59%). The authors of that study found that the tumor cellularity and the amount of mucin within the tumor mass were predictive of FDG-PET results [23]. The lack of expression of GLUT-1 may also account for the low sensitivity, as indicated in our study. According to another study, the sensitivity of FDG-PET for detecting mucinous colorectal adenocarcinomas ranged from 41–58% versus 92% for nonmucinous CRCs, probably because of their relative hypocellularity [24]. Moreover, because clinical studies have shown that the presence of mucin in CRCs is associated with poorer survival rates [25], PET might be of limited value in monitoring recurrence and metastasis in these patients.

FDG uptake is positively correlated with tumor invasion in gastrointestinal carcinoma, including esophageal and gastric cancers [20, 21, 26]. Mochiki et al. [20] found that the size and depth of invasion of the primary tumor were significantly correlated with the SUV in gastric cancer, and that T3 and T4 tumors were detected more frequently than T1 tumors. Furthermore, PET scans did not show the primary tumor in 21 of 85 patients with gastric cancer, 15 (71.4%) of whom had T1 tumors that invaded the submucosa [20]. Furudoi et al. [27] found that depth of invasion was also associated with GLUT-1 in CRC. This suggests that as the invasion becomes deeper, the resulting GLUT-1 overexpression causes the SUV to increase.

In the present study, SUV was significantly related to the intensity of GLUT-1 expression (r=0.619, P=.003) and there was a strong relationship between GLUT-1 intensity and the SUV. It is reasonable to assume that glucose consumption, as calculated by the SUV in FDG-PET, predicts the level of GLUT-1 transporters in CRC. Like other cancers, CRC usually expresses higher levels of glucose transporter proteins to satisfy the demand for energy for rapidly increasing proliferation. Among the GLUT isoforms, the basic glucose transporter (GLUT-1) is thought to play a major role in glucose uptake by many tumors [27]. The positive relationship between SUV and GLUT-1 expression has been confirmed in vitro by PET previously [28, 29], and the same correlation has been proven by PET clinically in many malignancies [7, 9, 12, 13, 22].

We noticed that more FDG accumulated in the intermediate and central part of the lesions, in parallel with higher GLUT-1 expression. Studies have shown that hypoxia upregulates GLUT-1 expression in vitro [29]. In the expanding tumor mass, which is generally characterized by a limited O2 supply and a high glucose consumption rate, a series of hypoxia-induced proteomic and genomic changes activate angiogenesis, anaerobic metabolism, and other processes that enable tumor cells to survive or escape their oxygen-deficient environment [30]. The hypoxia-inducible factor 1 (HIF-1), which is overexpressed in response to hypoxia, is the primary transcription factor mediating several physiologic and biological changes, including GLUT-1 overexpression [31]. An aggravated hypoxic state increases GLUT-1 expression, which would result in increased FDG uptake, seen as the SUV from the periphery to the intermediate and central part of the lesion in the present study. We must bear in mind that the microenvironment around the tumor is complicated and that GLUT-1 is not the only hypoxia-associated factor that influences FDG uptake. Other factors, such as hexokinase and various isoforms of the GLUT family also play important roles in determining FDG uptake and SUVs [22, 29].

Based on our findings of the association of FDG uptake with tumor size, depth of invasion, and GLUT-1 overexpression, we propose the following biological sequence: As the tumor grows larger and invades deeper, the intermediate and central part of the tumor, which contains more viable tumor cells, becomes more hypoxic. The hypoxic condition stimulates GLUT-1 overexpression and induces high FDG uptake. Because depth of invasion and GLUT-1 are prognostic markers of CRC, the SUVs determined by PET scans may offer important biological information about CRC.