The usefulness of 18F-fluorodeoxyglucose positron emission tomography (18F-FDG-PET) and a comparison of 18F-FDG-pet with 67gallium scintigraphy in the evaluation of lymphoma

Relation to histologic subtypes based on the World Health Organization classification




Although studies comparing conventional imaging modalities with 18F-fluorodeoxyglucose positron emission tomography (18F-FDG-PET) for the detection of lymphoma and although the relations between 18F-FDG-PET and histologic types were reported previously, most studies were not systematic and involved relatively small numbers of patients.


Two hundred fifty-five patients with lymphoma had their disease staged using 18F-FDG-PET, and 191 of those patients also were assessed using gallium-67 scintigraphy (67Ga). Disease sites were identified on a site-by-site basis using computed tomography scans and/or magnetic resonance imaging. The results of these conventional imaging modalities were compared with the results from 8F-FDG-PET and 67Ga, and correlations between the imaging results and pathologic diagnoses were evaluated by using the World Health Organization classification system.


Of 913 disease sites in 255 patients, 18F-FDG-PET identified >97% of disease sites of Hodgkin lymphoma (HL) and aggressive and highly aggressive non-Hodgkin lymphoma. For indolent lymphoma, the detection rate of 18F-FDG-PET was 91% for follicular lymphoma (FL); 82% for extranodal marginal zone B-cell lymphoma of mucosa-associated lymphoid tissue, irrespective of plasmacytic differentiation; and approximately 50% for small lymphocytic lymphoma (SLL) and splenic marginal zone lymphoma (SMZL). The results from 67Ga were similar to those from 18F-FDG-PET for most histologic subtypes. However, the sensitivity of 67Ga was unexpectedly poor for FL, for mantle cell lymphoma (MCL), and for the nasal type of natural killer/T-cell lymphoma (NK/T-nasal), ranging from 30% to 38%.


18F-FDG-PET was useful for all histologic subtypes of lymphoma other than SLL and SMZL. Compared with 67Ga, the authors strongly recommend the use of 18F-FDG-PET in patients with FL, MCL, and NK-nasal. Cancer 2007. © 2007 American Cancer Society.

Positron emission tomography (PET) using the glucose analog 18F-fluorodeoxyglucose (18F-FDG) (18F-FDG-PET) recently has emerged as an important tool for the management of malignant disease, including malignant lymphoma.1–4 This technique is based on the principle that most malignant cells have higher rates of glucose uptake and metabolism than normal tissues, resulting in accumulation of 18F-FDG in malignant cells.518F-FDG-PET provides information that is not obtainable with other imaging modalities. However, gallium-67 scintigraphy (67Ga) remains the most widely used nuclear tracer for evaluating lymphoma.6–9 Although the mechanism of uptake of 67Ga by lymphoma cells is not understood fully, its accumulation appears to depend in part on the density of the transferrin receptor (CD71) on tumor cells and the acidic condition of tumor tissues associated with anaerobic metabolism.7–9 The major limitations of 67Ga include relatively low sensitivity and accuracy, particularly for abdominal and extranodal sites.4, 10

One of the important roles of 18F-FDG-PET and 67Ga is initial staging and monitoring of residual disease and recurrence after treatment. 18F-FDG-PET has several advantages over 67Ga, including same-day results and high resolution.1, 2 There have been several reports comparing the usefulness of 18F-FDG-PET with conventional imaging modalities, such as computed tomography (CT) and 67Ga for detecting tumors,10–14 and examining the relations between 18F-FDG-PET and histologic subtypes based on the World Health Organization (WHO) classification system.15–18 However, those studies were not systematic, and they involved relatively small numbers of patients.9–13 Comparisons between 18F-FDG-PET and conventional imaging modalities using large numbers of patients would help to further our understanding of the biologic differences between lymphoma cell types, establish guidelines for the selection of appropriate clinical management, and verify the clinical relevance of lymphoma classification systems.1, 2, 6

The distinction between indolent and aggressive lymphomas is important clinically, because the choice of treatment is influenced by histologic subtype. In some studies that examined the role of 18F-FDG-PET in distinguishing between histologic subtypes, the avidity of lymphoma lesions for 18F-FDG correlated well with tumor aggressiveness,2, 19 and the likelihood of aggressive disease paralleled the increases in the standardized uptake value (SUV).19, 20 However, because those studies primarily involved patients with diffuse large B-cell lymphoma (DLBCL), it is not known whether the findings are applicable to other histologic subtypes. The objectives of the current study were 1) to investigate retrospectively the correlation the between sensitivity of 18F-FDG-PET and pathologic diagnoses based on the WHO classification system, 2) to assess and compare the sensitivity of 18F-FDG-PET and 67Ga with respect to their ability to identify disease involvement in each histologic subtype of lymphoma, and 3) to compare the SUV between the different histologic subtypes of lymphoma.



Between April 1998 and August 2006, 255 patients with malignant lymphoma who were treated at Gunma University, including 222 patients with newly diagnosed lymphoma and 33 patients with recurrent disease, had 18F-FDG-PET scans. All these patients had CT scans for initial imaging and posttherapy evaluation. Sixty-two patients underwent magnetic resonance imaging (MRI), and 191 patients received 67Ga imaging (Table 1). All these examinations were performed within 3 weeks before the initiation of treatment. Informed consent was obtained from all patients prior to 18F-FDG-PET scanning. Pathologic specimens were reviewed by at least 2 independent pathologists; if there was any disagreement, then a diagnosis was reached by consensus. Patients were excluded if they had received therapy for lymphoma within 6 months of the 18F-FDG-PET study.

Table 1. Number of Patients With Various Histology According to the World Health Organization Classification
HistologyNo. of patientsSUV max
FDGFDG and Ga-67MedianRange
  • SUV Max indicates maximum standardized uptake value; FDG, fluorodeoxyglucose; Ga-67, gallium-67 scintigraphy; ALCL, anaplastic large cell lymphoma; AITL, angioimmunoblastic T-cell lymphoma; NK/T-nasal, natural killer/T-cell lymphoma, nasal type; PTCL, peripheral T-cell lymphoma; DLBCL, diffuse large B-cell lymphoma; FL, follicular lymphoma; MZLs, marginal zone lymphomas; MALT, marginal zone B-cell lymphoma of mucosa-associated lymphoid tissue; SMZL; splenic marginal zone lymphoma; MCL, mantle cell lymphoma; SLL, small lymphocytic lymphoma; HL, Hodgkin lymphoma.

  • *

    Higher than MZL.

  • Higher than FL, MZL, and MCL.

  • The median SUV was 3.2 in plasma cell differentiation-positive patients and 2.6 in plasma cell differentiation-negative patients.

MZLs  3 
Subcutaneous panniculitis-like T113.4 

18F-FDG-PET Imaging

18F-FDG was prepared using the cyclotron and the automated synthetic apparatus in our hospital, as reported previously.21 The PET images were obtained using a PET scanner (SET2400W; Shimadzu, Kyoto, Japan) with a 59.5-cm transaxial field of view and a 20-cm axial field of view. The spatial resolution was 4.2 mm full-width half-maximum (FWHM) at the center of the field of view, and the axial resolution was 5.0 mm FWHM.

A whole-body image using the simultaneous emission-transmission method with a rotating external source was initiated from 40 to 60 minutes after the injection of 275 to 370 megabecquerels (MBq)/kg FDG using the multiple bed-position technique. Four or five sections from the head to the thigh were imaged for 8 minutes per section. The patients fasted overnight before 18F-FDG-PET imaging. The FDG-PET protocols were approved by the Institutional Review Board of Gunma University.21

Attenuation-corrected transaxial images were reconstructed by using the ordered subset-expectation maximization algorithm into 128 × 128 matrices with pixel dimensions of 4.0 mm in plane and 3.125 mm axially. Finally, every 3 consecutive slices were summed to generate a 9.8-mm-thick transaxial image that was used for visual interpretation and quantitative analysis. Similarly, 9.8-mm-thick coronal images also were reconstructed from attenuation-corrected transaxial images. Two experienced nuclear medicine physicians independently evaluated all PET images. In most patients, interpretation of the imaging results was concordant between the 2 nuclear physicians. Some minor discordance (approximately 2–3% of patients; <1% of results) was resolved by discussion. Functional images of the SUV also were produced using the attenuation-corrected transaxial images, the amount of injected FDG, the body weight, and cross-calibration factors between PET and the dose calibrator.21 SUV was defined as follows: SUV = the radioactive concentration in the tissue or lesion (MBq/g)/(injected dose [MBq]/patient's body weight [g]).

Abnormal FDG uptake was defined as greater than background activity in the surrounding tissue that was unrelated to physiologic sites of tracer uptake for excretion. In the spleen, abnormal uptake was defined as FDG accumulation greater than that in the liver based on the SUV or the presence of areas of increased focal FDG uptake in the spleen. To assure reproducibility of the measurements, the maximum SUV (SUVmax) in a large lesion was used for further analysis.

Other Imaging

67Ga imaging was performed within 3 weeks of the FDG-PET study. Patients were injected with 148 MBq of 67Ga-citrate, and anterior and posterior views of the whole body were obtained 72 hours later, using dual- or single detector γ cameras. Although 370 MBq 67Ga imaging was used in many of the reported patients,10, 12 we routinely use 148 MBq or 4 mCi in clinical practice. Instead, we spent enough time to acquire sufficient counts for optimal scintigraphy. Planar whole-body and spot views of the chest and abdomen were acquired. To analyze the head, the patients primarily had single positron emission CT. Continuous CT scans or MRI imaging of the neck, thorax, abdomen, and pelvis were obtained using an intravenous contrast medium. Reconstructed images had a thickness of 5 mm in the neck and 7.5 mm in the thorax, abdomen, and pelvis. PET and 67Ga images were compared with CT and/or MRI images to determine the anatomic location of disease sites.

Data Analysis

The disease sites were determined on CT and/or MRI studies according to the International Workshop Response Criteria proposed in 1999.22 To analyze the lesion detection rate of each method, each patient's CT and/or MRI results were tabulated by lymph node region, basically according to Kostakoglu et al.10 (except for spleen and iliac lymph nodes) as follows; cervical, supraclavicular, paratracheal, anterior mediastinal, posterior mediastinal, hilar, axillary, celiac, paraaortic, mesenteric, iliac, inguinal, and extranodal sites (orbits, gastrointestinal tract, liver, bone, lung, and soft tissue). We classified the spleen as abdominal and iliac lymph nodes as pelvic to coincide with anatomic location. The involved lymph nodes were classified further into 4 anatomic groups: 1) head/neck (nasolaryngeal, cervical, submandibular, and supraclavicular lymph nodes), 2) chest (paratracheal, mediastinal, axially, and hilar lymph nodes), 3) abdomen (paraaortic, celiac, and mesenteric lymph nodes), and 4) pelvis (iliac and inguinal lymph nodes). Each CT/MRI-positive site within the same lymph node region was recorded as 1 site; thus, lymph nodes were not counted individually within each specific lymph node region. Concordance among the results from 18F-FDG-PET, 67Ga, and CT was evaluated according to histologic subtype and anatomic location. Any discord between the 18F-FDG-PET and CT results was resolved by comparing the pretherapy and posttherapy CT scans. Bone marrow involvement was not evaluated in this study, because many patients did not undergo a bone marrow examination and 18F-FDG-PET posttreatment.

The SUVs were analyzed in relation to histology. The mean and standard deviation were calculated. SUVs were compared in 5 B-cell lymphomas: Burkitt lymphoma (BL), DLBCL, follicular lymphoma (FL), marginal zone lymphomas (MZLs) (extranodal and splenic), and mantle cell lymphoma (MCL). In addition, SUVs were compared in 4 natural killer (NK)/T-cell lymphomas: anaplastic large cell lymphoma (ALCL), angioimmunoblastic T-cell lymphoma (AITL), peripheral T-cell lymphoma unspecified (PTCL-u), and extranodal NK/T-cell lymphoma nasal type (NK/T-nasal). The histologic subgroups were divided further into 4 groups to compare the SUVs in Hodgkin lymphoma (HL): 1) HL, 2) indolent B-cell lymphoma (FL and MZLs), 3) aggressive B-cell lymphoma (DLBCL), and 4) NK/T-cell lymphoma (ALCL, AITL, PTCL-u, and extranodal NK/T-nasal). Statistical analysis was performed using R Software (Available at URL: The SUVs among the various histologic subtypes were compared using the Kruskall-Wallis test accompanied by a Bonferroni test and the Scheffe test for multiple comparisons. Multisample tests for equality of proportion were performed using the Prop test followed by the pairwise Prop test. Comparisons of 2 different groups were analyzed using the Mann-Whitney U test for nonparametric data. P values <.05 were considered statistically significant.


Comparison of 18F-FDG-PET and CT Scan

Among 255 patients with lymphoma, 913 regions were interpreted as disease sites on CT scans and/or MRI evaluations, including 844 regions (92.4%) that were detected by 18F-FDG-PET. In ALCL, AITL, NK/T-nasal, BL, and MCL,18F-FDG-PET identified all disease sites (100%). Similarly, 18F-FDG-PET detected 98% of disease sites in PTCL-u, 97% of disease sites in DLBCL and HL, 91% of disease sites in FL, and 82% of disease sites in MALT. Conversely, the detection rate of 18F-FDG-PET was only 53% for SMZL and 50% for SLL, although the number of disease sites was small (Table 2). It is noteworthy that, in 3 patients with FL, 18F-FDG-PET did not detect FL of the duodenum. There was no tendency for the detection of lymphoma in particular anatomic regions by 18F-FDG-PET. When the patients with MALT were divided into 2 groups according to histologic subtype, a plasmacytic differentiation-positive group and a plasmacytic differentiation-negative group, the detection rates did not differ significantly groups (32 of 38 patients [84%] vs 57 of 71 patients [80%], respectively) (Table 2).

Table 2. Incidence and Anatomic Regions Detected by 18F-Fluorodeoxyglucose Positron Emission Tomography and Their Relation to Histologic Types According to World Health Organization Classification
HistologyNo. of patients/Total
Total no. (%)Head and neckChestAbdomenPelvisOthers
  • ALCL anaplastic large cell lymphoma; AITL, angioimmunoblastic T-cell lymphoma; NK/T-nasal, natural killer/T-cell lymphoma, nasal type; PTCL, peripheral T-cell lymphoma; DLBCL, diffuse large B-cell lymphoma; FL, follicular lymphoma; MALT, marginal zone B-cell lymphoma of mucosa-associated lymphoid tissue; SMZL; splenic marginal zone lymphoma; MCL, mantle cell lymphoma; SLL, small lymphocytic lymphoma; HL, Hodgkin lymphoma.

  • *

    Stomach, 2 of 4 patients.

  • Duodenum, 0 of 3 patients.

  • Thirty-two of 38 plasmacytic differentiation-positive patients (84%), and 57 of 71 plasmacytic differentiation-negative patients (80%).

  • §

    Stomach, 2 of 4 patients; orbita, 22 of 28 regions.

ALCL25/25 (100)9/93/33/32/28/8
AITL34/34 (100)12/128/810/102/22/2
NK/T-nasal30/30 (100)9/96/65/51/19/9
PTCL54/55 (98)20/2010/119/96/69/9
Burkitt21/21 (100)2/26/65/53/35/5
DLBCL268/276 (97)79/8141/4360/6128/2860/63*
FL175/193 (91)55/5838/4334/3940/428/11
MALT89/109 (82)28/3016/1910/115/730/42§
SMZL10/19 (53)2/31/24/93/50/0
MCL51/51 (100)23/239/96/69/94/4
SLL9/18 (50)3/64/52/40/30/0
HL73/75 (97)31/3123/2411/122/26/6
Subcutaneous panniculitis-like T5/7 (71)0/00/20/02/23/3
Total no. (%)844/913 (92.4)273/284 (96.1)165/181 (91.2)159/174 (91.4)103/112 (92.8)144/162 (88.9)

Comparison of 18F-FDG-PET and 67Ga

In 191 patients, 654 regions were identified as disease sites by CT scans and/or MRI evaluations. 18F-FDG-PET detected 592 sites (90.5%), whereas 67Ga detected 371 regions (56.7%). Basically, the regions detected by 67Ga were positive on 18F-FDG-PET evaluations (Table 3). With respect to the histologic subtypes, 67Ga detected disease sites relatively well: 89% in ALCL and BL, 88% in PTCL-u, 74% in AITL, and 68% in DLBCL. Conversely, 67Ga detected disease sites in only 13% of SLL, 30% of MCL, 31% of SMZL, 33% of FL, and 38% of NK/T-nasal. The detection rate of 67Ga was 55% in MALT and 54% in HL. Similarly, 67Ga was able to detect disease sites that had been identified positively by 18F-FDG-PET relatively well: 90% of PTCL-u, 89% of ALCL and BL, 74% of AITL, and 73% of DLBCL. However, 67Ga detected disease sites that were identified by 18F-FDG-PET in only 30% of MCL, 38% of NK/T-nasal and FL, and 50% of SMZL and SLL. Thus, in imaging lymphoma before therapy, 18F-FDG-PET had significantly higher sensitivity than 67Ga for all histologic subtypes. Unexpectedly, for 1 patient with MALT in the orbit and for 1 patient with SMZL in the spleen, 18F-FDG-PET was not able to detect the disease site detected by 67Ga; the region was adjacent to the upper portion of eyeball in the former patient, and the latter patient had disease that was accompanied by marked splenomegaly (22 × 15 × 6 cm; 1.4 kg). The sites detected by 67Ga are listed in Table 3; head/neck lesions (65.2%) were detected relatively well by 67Ga compared with abdominal and pelvic lesions (46.3% and 42.7%, respectively).

Table 3. Comparison of 18F-fluorodeoxyglucose Positron Emission Tomography and Gallium-67 Scintigraphy for Detecting Disease Sites and their Relation to the World Health Organization Classification
HistologyGa-67/CT, MRIGa-67/PET
Total (%)Head and neckChestAbdomenPelvisOthersTotal (%)Head and neckChestAbdomenPelvisOthers
  1. Ga-67 indicates gallium-67 scintigraphy; CT, computed tomography; PET, positron emission tomography; ALCL, anaplastic large cell lymphoma; AITL, angioimmunoblastic T-cell lymphoma; NK/T-nasal, natural killer/T-cell lymphoma, nasal type; PTCL, peripheral T-cell lymphoma; DLBCL, diffuse large B-cell lymphoma; FL, follicular lymphoma; MALT, marginal zone B-cell lymphoma of mucosa-associated lymphoid tissue; SMZL; splenic marginal zone lymphoma; MCL, mantle cell lymphoma; SLL, small lymphocytic lymphoma; HL, Hodgkin lymphoma.

ALCL17/19 (89)8/83/32/30/14/417/19 (89)8/83/32/30/14/4
AITL17/23 (74)8/85/62/61/11/217/23 (74)8/85/62/61/11/2
NK/T-nasal6/16 (38)2/51/40/30/03/46/16 (38)2/51/40/30/03/4
PTCL38/43 (88)15/166/84/55/58/938/42 (90)15/166/74/55/58/9
Burkitt16/18 (89)2/23/45/53/33/416/18 (89)2/23/45/53/33/4
DLBCL129/189 (68)41/5320/3028/3913/1727/50129/177 (73)41/4920/2828/3813/1727/45
FL48/145 (33)24/4510/325/264/315/1148/127 (38)24/4410/235/234/295/8
MALT51/92 (55)11/238/145/112/625/3851/78 (65)11/218/105/102/425/33
SMZL4/13 (31)0/10/12/72/40/04/6 (67)0/00/12/32/20/0
MCL8/27 (30)5/150/40/20/23/48/27 (30)5/150/40/20/23/4
SLL1/8 (13)1/40/10/20/10/01/2 (50)1/10/00/10/00/0
HL31/54 (57)16/2411/153/120/21/131/52 (60)16/2311/143/120/21/1
Subcutaneous panniculitis-like T5/7 (71)0/00/20/02/23/35/5 (100)0/00/00/02/23/3
Total no. (%)371/654 (56.7)133/204 (65.2)67/124 (54)56/121 (46.3)32/75 (42.7)83/130 (63.8)371/592 (62.7)133/192 (69.3)67/104 (64.4)56/111 (50.5)32/68 (47.1)83/117 (70.9)

Intensity of FDG Uptake

The intensity of 18F-FDG uptake for the various histologic subgroups is listed in Table 1. The median SUVmax was 12.0 in ALCL, 8.6 in AITL, 7.5 in NK-nasal, 8.8 in PTCL-u, 9.9 in BL and DLBCL, 4.8 in FL, 3.0 in MZLs (3.1 in MALT and 2.2 in SMZL), 5.1 in MCL, 2.4 in SLL, 6.6 in HL, and 3.4 in subcutaneous panniculitis-like T-cell lymphoma. In NK/T lymphomas (ALCL, AITL, NK-nasal, and PTCL-u), there was no statistically significant difference in the SUV among the subtypes. Concerning B-cell lymphomas, statistically, the SUV in BL was significantly higher than in MZL. Likewise, the SUV in DLBCL was higher than in FL, MZLs, and MCL. However, there was no difference among the other histologic subtypes. In FL, the median SUVmax of grade 1 FL, grade 2 FL, and grade 3 FL was 4.5, 4.8, and 5.4, respectively; the differences among the subtypes were not statistically significant. Likewise, in MALT, the median SUVmax in patients who had disease with or without plasmacytic differentiation was 3.2, and 2.6, respectively; the difference between the subgroups was not statistically significant (Table 1).

When the histologic subtypes were divided into 4 groups, the median SUVmax of NK/T lymphoma, aggressive B-cell lymphoma (DLBCL), indolent B-cell lymphoma (FL and MZLs), and HL was 9.4, 9.9, 3.3, and 6.6, respectively (Table 4). The 18F-FDG uptake in HL was significantly higher than that in indolent B-cell lymphoma but lower than that in aggressive B-cell lymphoma. Likewise, the 18F-FDG uptake in indolent B-cell lymphoma was significantly lower than that in NK/T lymphoma and DLBCL. However, no difference was observed between the other histologic subtypes (Table 4).

Table 4. Maximum Standardized Uptake Value in Four Types of Lymphoma
HistologyMedian SUV max
  • SUV Max indicates maximum standardized uptake value; NK, natural killer; DLBCL, diffuse large B-cell lymphoma; HL, Hodgkin lymphoma.

  • *

    Higher than indolent lymphoma but lower than DLBCL.

  • Lower than NK/T-cell lymphoma and DLBCL.

NK/T-cell lymphoma9.4
Indolent B-cell lymphoma3.3*


In this study of 255 patients that had 913 sites of disease involvement, FDG-PET detected disease sites with almost 100% accuracy in ALCL, AITL, BL, DLBCL, HL, MCL, NK-nasal, and PTCL-u. Elstrom et al. reported that FDG-PET detected disease in at least 1 site in 100% of patients with ALCL and BL but in only 40% of patients with PTCL in a limited number of patients.16 Based on that report, 18F-FDG-PET was supposed to be less sensitive in T-cell lymphoma than in B-cell lymphoma. However, it was not determined with certainty whether, in those patients, all disease sites were 18F-FDG-PET-positive. Our current results showed clearly that 18F-FDG-PET depicted the disease sites very well not only in HL, DLBCL, and BL but also in PTCL-u, ALCL, and AITL; this is evidence that lymphoma cell surface markers do not influence 18F-FDG uptake. However, because the numbers of patients with these subtypes of aggressive lymphoma were small, further studies will be needed to confirm the utility of 18F-FDG-PET in patients with these lymphoma subtypes.

To our knowledge, few studies to date have examined the sensitivity of 18F-FDG-PET in the various histologic subtypes of indolent lymphoma.17, 23, 24 In the current study, 18F-FDG-PET detected disease sites with high accuracy in FL, which was in agreement with previous reports.16, 17 However, it should be noted that 18F-FDG-PET failed to detect sites in all 3 patients who had FL in the duodenum. Hoffmann et al. reported similar results in 7 patients with primary duodenal FL.25 It would be interesting to determine whether the pathologic grade of FL is correlated with 18F-FDG uptake. Our results and those from a recent study suggest that there is no correlation between 18F-FDG uptake and patients with grade 1 or grade 2 FL. This is similar to the clinical observation that outcomes do not differ significantly between these 2 categories.17 However, this information would be important for patients with grade 3 FL, because it is believed that these patients to have a more aggressive clinical course and to need more aggressive treatment. However, neither the disease site detection rate nor the SUV differed between grade 3 and grades 1 or 2. Recently, Wohrer et al. also demonstrated similar results in their analysis of 64 patients with FL.26

Conflicting results have been reported concerning the utility of FDG-PET in MALT. In 2 of their articles, Hoffmann et al. reported poor tracer uptake in patients with MALT.27, 28 In contrast, Beal et al. reported that MALT lymphoma could be detected with 18F-FDG-PET in approximately 70% to 80% of patients and that there was no difference in detection between patients with lymph node and extranodal disease sites.18 Our study, which, to our knowledge, includes the largest number of patients to date, supports the latter observation. This is important because patients with early-stage disease are treated most effectively with radiation therapy. Results from a recent study suggested that histologic plasma cell differentiation may influence the detection rate of 18F-FDG-PET.29 However, in the current study, there was no difference in the disease site detection rate or SUVmax between patients with and without plasmacytic differentiation. However, our patients with MALT were biased, because there was a relatively high proportion of patients with orbital disease sites (28 regions in 24 patients) and a low proportion of patients with gastric disease sites (4 patients).

In contrast to these findings, 18F-FDG-PET detected only 50% of disease sites in SLL and 53% of disease sites in SMZL, although the number of patients with SLL was small because of its rarity among the Japanese population. Previous reports also have reported its limited usefulness for staging in SLL, with a sensitivity of approximately 50%.17, 23, 24 To our knowledge, there has been only 1 report regarding the utility of 18F-FDG-PET in SMZL, in which the PET scan showed homogenous but increased uptake (SUV = 4.2) in the spleen.17 Conversely, in the current study, we demonstrated that 18F-FDG-PET is limited in detecting SMZL disease sites, and the median SUV was 2.5. Low degrees of 18F-FDG uptake in patients with SMZL and SLL may correlate well with their indolent clinical courses and most likely highlight the biologic similarities of these tumor types. Because the evaluation of lymph node involvement is sometimes critical for deciding treatment in both subtypes, staging by 18F-FDG-PET alone is not recommended; combined evaluation with 18F-FDG-PET and another imaging method, such as CT scanning, is essential.

Several small studies have compared the sensitivities of 18F-FDG-PET and 67Ga10–12 and concluded that the sensitivity of 67Ga for evaluating lymphomas was dependent on the cell type; the detection rate reportedly is relatively high in HL and aggressive NHL (up to approximately 70–80%10–12 but low in indolent lymphomas (range, 41–56%).10, 30 In our study, 18F-FDG-PET was superior to 67Ga in 191 patients with 654 disease sites in all histologic subtypes. 67Ga detected disease sites relatively well (≥66% of disease sites) in ALCL, AITL, BL, PTCL-u, and DLBCL. Conversely, ≤55% of disease sites were detected in FL, MALT, and SMZL. In particular, it should be noted that the sensitivity of 67Ga in FL and SMZL was low (33% and 31%, respectively). These results are concordant with the observation that the degree of 18F-FDG or 67Ga accumulation is associated with cell proliferation and histologic grade,8, 20, 30 and they suggest that caution should be used in the staging of indolent lymphoma by 67Ga. One of the striking results of the current study was that the sensitivity of 67Ga in MCL and NK-nasal was extremely low, whereas 18F-FDG-PET detected disease sites in these subtypes with high sensitivity, because these 2 subtypes are aggressive clinically and are resistant to conventional chemotherapy (combined cyclophosphamide, doxorubicin, vincristine, and prednisone).31–34 There have been few studies on the sensitivity of 67Ga in these subtypes. The 1 reported patient with MCL demonstrated extremely low 67Ga uptake.8 Concerning MCL that previously was categorized as low-grade lymphoma,31, 32 this result may reflect biologic features in these subtypes and demonstrates the utility of 18F-FDG-PET for these subtypes. Currently, however, the precise mechanism in these lymphoma subtypes is uncertain. Because the sensitivity of 18F-FDG-PET was 100% in these subtypes, 18F-FDG-PET is recommended strongly for patients with lymphoma who have these subtypes.

Some reports have suggested that 18F-FDG uptake is correlated with the histologic grade of malignancy.2, 19 However, in those reports, the numbers of patients were relatively small, and the histologic subtype of aggressive lymphoma that was reported tended to be DLBCL.19 The current study, we clearly observed that 18F-FDG uptake was higher in aggressive and highly aggressive lymphomas irrespective of tumor surface markers, which may reflect the proliferative activity of these subtypes.

With regard to indolent lymphoma and MCL, 18F-FDG uptake was low with a median SUVmax ranging from 2.2 to 5.1. Wohrer et al26 and Beal et al.18 reported that the median SUV in grade 1 or 2 FL and MALT was 5.7 and 5.5, respectively. Karam et al. also reported similar results in their analysis of 47 patients who had indolent lymphoma.17 Those results support the view that an SUV ≤6 most likely is associated with indolent lymphoma.35 In addition, monitoring the SUV in patients with indolent lymphomas would be helpful for the early detection of transformation to a more aggressive subtype, because the SUV reportedly increases when transformation occurs.19 These results seem to suggest a potential role for 18F-FDG-PET in the diagnosis of lymphoma. However, because there are some exceptions in each histologic type, the discrimination of histologic subtypes by using SUV alone may be limited. Thus, biopsy continues to be the most important procedure in establishing an unequivocal diagnosis. However, if a discrepancy occurs between the initial diagnosis and the clinical course, then SUV data would be helpful to the histopathologist for reviewing the diagnosis.

In conclusion, 18F-FDG-PET is now used to evaluate treatment efficacy in patients with lymphoma.19, 36, 37 However, information regarding the sensitivity of 18F-FDG-PET for each histologic subtype is necessary to evaluate treatment efficacy. In the current study, we demonstrated the utility of 18F-FDG-PET in all histologic lymphoma types except SMZL and SLL. Compared with 67Ga, 18F-FDG-PET is useful primarily in FL, MCL, and NK/T-nasal. However, combined evaluation by 18F-FDG-PET and CT scan and/or MRI is essential.


We thank Drs. Hiroyuki Irisawa, Hideki Uchiumi, Yoshiyuki Suzuki, and Masamitsu Karasawa for useful discussions and Drs. Minato Nakazawa and Kohtaro Toyama for help with statistical analyses.