Abstract
Objective:
In this study, we aimed to investigate the value of 18F-fluorodeoxyglucose positron emission tomography/computed tomography (18F-FDG PET/CT) in the differential diagnosis of nodal diffuse large B-cell lymphoma (DLBCL) and follicular lymphoma (FL) and to detect early FL transformation.
Methods:
We retrospectively enrolled 141 and 56 patients with pathologically confirmed DLBCL and FL, respectively, who underwent 18F-FDG PET/CT before treatment. The most metabolically active lymph nodes in each patient were selected to compare the imaging characteristics (including shape, size, density, adjacent tissue structures, and metabolism) between DLBCL and FL. Logistic regression was performed to identify the independent risk factors for DLBCL, and receiver operating characteristic (ROC) curve analysis was used to assess the diagnostic value of the different parameters.
Results:
Compared with DLBCL, FL was more likely to affect the para-iliac (62.5% vs. 32.6%, P<0.05) and inguinal (50.0% vs. 28.4%, P<0.05) regions. Significant differences (P<0.05) were observed between DLBCL and FL in lesion fusion, necrosis, surrounding fat stranding, adjacent organ involvement, nonround shape, SUVmax, SUVpeak, SUVmean, SULmax, SULpeak, SULmean, lesion SUVmax/liver SUVmax (L/L), lesion SUVmax/mediastinum SUVmax (L/M), and total lesion glycolysis (TLG). Multivariate logistic regression analysis identified SUVmax >12.35, necrosis, and surrounding fat stranding as independent risk factors of DLBCL. The diagnostic model achieved an area under the curve (AUC) of 0.882, with a sensitivity of 69.5% and specificity of 94.6% for diagnosing DLBCL.
Conclusions:
¹⁸F-FDG PET/CT may provide greater diagnostic value in distinguishing nodal DLBCL from FL and probably aid in the early detection of FL transformation and thereby guide timely clinical intervention, which could improve the prognosis.
Key Words: diffuse large B-cell lymphoma, follicular lymphoma, positron emission tomography, 18F fluorodeoxyglucose, differential diagnosis
Non-Hodgkin lymphoma (NHL) is the most prevalent form of lymphoma worldwide, and ranks 10th and 11th as the most common malignancy and cause of mortality, respectively.1 Diffuse large B-cell lymphoma (DLBCL) and follicular lymphoma (FL) are the predominant NHL subtypes and differ significantly in their clinical behavior and aggressiveness. DLBCL is categorized as a clinically aggressive lymphoma, whereas FL is considered indolent; aggressive lymphomas are generally associated with a substantially worse prognosis.2 Histologic transformation, particularly progression from FL to DLBCL, remains a significant challenge in the management of patients with NHL.3 Over time, 10% to 70% of FL cases transform into DLBCL, with an annual transformation rate of ∼2%.4 The early identification of DLBCL-transformed cases is critical for optimizing treatment strategies in patients with lymphoma.
Numerous studies have explored the diagnostic indicators of the FL-to-DLBCL transformation. These indicators include a sudden increase in serum lactate dehydrogenase (LDH) levels, development of B symptoms, hypercalcemia, and rapid lymph node enlargement, all of which suggest a high likelihood of transformation.5 18F-fluorodeoxyglucose positron emission tomography/computed tomography (18F-FDG PET/CT) is a noninvasive imaging modality that not only resembles conventional imaging in the identification of morphologic features but also quantifies tumor metabolic activity through the maximum standardized uptake value (SUVmax) to thereby reflect the tumor metabolic characteristics. As lymphoma is a systemic disease, PET/CT, a whole-body imaging tool, overcomes the regional limitations of conventional imaging techniques and provides additional imaging information for clinical assessment. The metabolic activity of a lymphoma correlates with its aggressiveness, with higher metabolic activity frequently indicating a more aggressive nature. Therefore, the presence of lesions with an elevated SUVmax in FL patients should raise the suspicion of transformation to DLBCL. Although biopsy remains the gold standard for lymphoma diagnosis, it has inherent limitations, including sampling inaccuracy, invasiveness, and the inability to perform biopsies in critically ill patients.6 In such cases, imaging-based evaluation can provide crucial and timely diagnostic insights for clinical decision-making.
Most of the studies on the initial diagnosis of aggressive and indolent lymphomas have focused on metabolic characteristics, with the majority reporting a statistically significant difference in the SUVmax between these 2 subtypes.7,8 Conversely, relatively few studies have evaluated the morphologic features. Although Imada et al9 identified necrosis and surrounding fat stranding as valuable predictors, their study was restricted to the head and neck region. To date, only a few studies have systematically integrated the qualitative CT morphologic signs with quantitative metabolic parameters for a comprehensive whole-body evaluation.
Therefore, in the present study, we aimed to systematically compare the baseline 18F-FDG metabolic parameters and CT morphologic features of DLBCL and FL to address this limitation and provide a comprehensive reference to enable a more accurate differential diagnosis.
PATIENTS AND METHODS
Patient Selection
A retrospective analysis was conducted on 141 consecutive patients with DLBCL and 56 patients with FL whose diagnosis was pathologically confirmed at our Hospital, from May 2017 to January 2024. The DLBCL group comprised 72 men and 69 women, aged 27 to 92 years (median age 66 y). The FL group included 25 men and 31 women, aged 32 to 84 years (median age 59 y). All patients were clinically staged according to the Ann Arbor Staging System. FL was histologically graded according to the World Health Organization (WHO) classifications.
The inclusion criteria were as follows: (1) pathologically confirmed diagnosis of DLBCL or FL; (2) underwent 18F-FDG PET/CT with no prior treatment before the scan; (3) a less than 2-week interval between the 18F-FDG PET/CT and pathologic examination; (4) the primary lesion was confined to nodal sites other than the spleen.
The exclusion criteria were as follows: (1) diagnosed of lymphoma subtypes other than DLBCL or FL (including FL grade 3B); (2) history of other malignancies; (3) lymph nodes with a short-axis diameter <1 cm; (4) involvement of the liver [to ensure that the liver could serve as a stable reference for lesion SUVmax/liver SUVmax (L/L) calculation]; (5) primary or secondary central nervous system involvement. This study was conducted in accordance with the principles of the Declaration of Helsinki and was approved by the Ethics Committee of our Hospital.
18F-FDG PET/CT Imaging Protocol
A Siemens Biograph 16 PET/CT scanner (Siemens, Erlangen, Germany) was used for imaging. The imaging agent 18F-FDG was automatically synthesized using an RDS III cyclotron (CTI, USA) and a PET-FDG-IT-I chemical synthesis module (Beijing PET Biotechnology Co. Ltd), while ensuring a radiochemical purity of >95%. Patients fasted for 4 to 6 h before imaging to maintain fasting blood glucose levels <11.1 mmol/L. 18F-FDG was intravenously administered at a dose of 370 to 555 MBq (3.7 to 7.4 MBq/kg), followed by a 60-min resting period, and the patients were asked to empty their bladder before the PET/CT acquisition. The PET/CT parameters included a scanning range from the skull vertex to the upper tibia (including the elbow joints). For body imaging, the CT scan was conducted with the following parameters: 120 kV, 50 mA, a rotation time of 0.5/s, a pitch of 0.75, and a matrix size of 512×512. PET acquisition was performed in 3D mode with 6 to 7 bed positions and 2 minutes/bed position.
Image Evaluation
The mage interpretation and analysis were performed by 2 experienced diagnostic physicians in the PET/CT imaging department who independently reviewed all PET/CT images while being blinded to the pathologic diagnosis. Disagreements, if any, regarding the interpretation of the results were resolved by inviting a third-party physician to review the images and discuss the final results. Lymph nodes were considered 18F-FDG-positive if their metabolic activity exceeded the liver blood pool background.10,11 The lymph node with the highest 18F-FDG uptake was selected as the target lesion using visual analysis, avoiding regions of physiological uptake. The morphologic features included nonround shape (defined as a short-to-long axis ratio of the lymph node <0.8),12 maximum short-axis diameter (cm), CT attenuation value (HU), necrosis (defined as a visually identifiable low-density area with a CT value <30 HU),13 surrounding fat stranding (defined as linear or reticular hyperattenuation and/or blurring of the perinodal fat plane adjacent to the node, visible on at least 2 consecutive axial slices), adjacent organ involvement, and lymph node fusion. For metabolic assessment, a region of interest (ROI) was manually delineated to fully encompass the lesion, and maximum standardized uptake value (SUVmax), peak standardized uptake value (SUVpeak), maximum standardized uptake value normalized to lean body mass (SULmax), and peak standardized uptake value normalized to lean body mass (SULpeak) were automatically calculated. L/L and lesion SUVmax/mediastinum SUVmax (L/M) ratios were calculated as lesion SUVmax divided by liver and mediastinal blood pool SUVmax, respectively. Specifically, SUL indices were derived using the Janmahasatian formula for lean body mass estimation,14 and SUVpeak (or SULpeak) was defined as the mean activity within a 1 cm3 spherical volume of interest that was centered on the voxel with the maximum uptake.15 In adherence to the guideline of the European Association of Nuclear Medicine (EANM),15 the 41% SUVmax threshold was used to calculate the mean standardized uptake value (SUVmean), mean standardized uptake value normalized to lean body mass (SULmean), metabolic tumor volume (MTV), and total lesion glycolysis (TLG). In the present study, MTV was calculated from the lesion demonstrating the highest FDG uptake and is therefore referred to as lesionMTV. Accordingly, TLG derived from the same lesion is referred to as lesionTLG.
Statistical Analysis
Statistical analyses were performed using IBM SPSS 27.0 and Prism 10. Quantitative variables that did not follow a normal distribution are expressed as medians (Q1, Q3), whereas categorical variables are presented as frequencies (percentages). Comparisons between groups were conducted using the χ2 test for categorical variables and the Mann-Whitney U test for continuous variables. A false discovery rate (FDR) correction was applied to adjust for multiple comparisons. Receiver operating characteristic (ROC) curve analysis was performed for significant metabolic parameters. Univariate and multivariate logistic regression analyses were used to identify the independent risk factors for DLBCL. The selection of independent variables for the multivariate model was based on univariate significance (P<0.05) and assessment of multicollinearity using the variance inflation factor (VIF). Finally, ROC analysis was performed to assess the predictive performance of these risk factors. The threshold of statistical significance was set at P<0.05, which was considered statistically significant.
RESULTS
Baseline Characteristics
A statistically significant difference in age was observed between the DLBCL and FL groups (Z=-4.12, P<0.001; Table 1). However, there was no significant difference in sex distribution or Ann Arbor stage between the 2 groups (χ2=0.66, and <0.01, P=0.416, 0.935; Table 1). In the FL group, lymphomas in 39 patients (69.6%) were classified as grades 1 to 2, whereas those in 17 patients (30.4%) were classified as grade 3A (Table 1).
TABLE 1.
General Characteristics, CT Morphologic Features, and 18F-FDG Metabolic Characteristics of Patients With DLBCL and FL
| Variables | DLBCL (n=141), n (%) or M (Q1, Q3) | FL (n=56), n (%) or M (Q1, Q3) | Z/χ2 | P |
|---|---|---|---|---|
| Age | 66.0 (59.0, 74.0) | 59.0 (48.8, 65.0) | Z=-4.12 | <0.001 |
| Sex | χ2=0.66 | 0.416 | ||
| Male | 72 (51.1) | 25 (44.6) | ||
| Female | 69 (48.9) | 31 (55.4) | ||
| Ann Arbor stage | χ2<0.01 | 0.935 | ||
| I-II | 47 (33.3) | 19 (33.9) | ||
| III-IV | 94 (66.7) | 37 (66.1) | ||
| FL histologic grade | NA | NA | ||
| 1-2 | NA | 39 (69.6) | ||
| 3A | NA | 17 (30.4) | ||
| Fusion | 87 (61.7) | 23 (41.1) | χ2=6.92 | 0.009 |
| Necrosis | 105 (74.5) | 16 (28.6) | χ2=35.63 | <0.001 |
| surrounding fat stranding | 99 (70.2) | 17 (30.4) | χ2=26.30 | <0.001 |
| adjacent organ involvement | 43 (30.5) | 4 (7.1) | χ2=12.03 | <0.001 |
| Nonround shape | 89 (63.1) | 25 (44.6) | χ2=5.61 | 0.018 |
| CT value (HU) | 44.8 (40.0, 52.7) | 45.1 (40.8, 48.7) | Z=-0.81 | 0.418 |
| Largest short diameter (cm) | 2.4 (1.6, 3.9) | 1.9 (1.5, 2.7) | Z=-1.93 | 0.053 |
| SUVmax | 18.8 (12.5, 25.8) | 9.6 (7.7, 12.0) | Z=-7.53 | <0.001 |
| SUVpeak | 16.2 (10.0, 22.1) | 7.6 (6.1, 9.6) | Z=-7.34 | <0.001 |
| SULmax | 12.8 (8.8, 17.8) | 7.0 (5.7, 8.2) | Z=-7.27 | <0.001 |
| SULpeak | 11.2 (7.2, 15.5) | 5.5 (4.5, 6.8) | Z=-7.35 | <0.001 |
| SUVmean | 11.8 (7.4, 16.1) | 6.1 (4.6, 7.5) | Z=-7.30 | <0.001 |
| SULmean | 8.1 (5.5, 11.7) | 4.3 (3.4, 5.1) | Z=-7.32 | <0.001 |
| lesionMTV | 15.0 (6.8, 48.4) | 10.1 (4.9, 23.6) | Z=-1.52 | 0.128 |
| lesionTLG | 177.9 (64.8, 592.0) | 54.8 (29.6, 167.7) | Z=-4.20 | <0.001 |
| L/L | 5.9 (3.7, 7.7) | 2.8 (2.3, 3.7) | Z=-6.60 | <0.001 |
| L/M | 8.6 (6.0, 12.5) | 4.1 (3.4, 5.5) | Z=-6.97 | <0.001 |
L/L indicates lesion SUVmax/liver SUVmax; L/M, lesion SUVmax/mediastinum SUVmax; M: median; Q1, first quartile; Q3, third quartile; Z, Mann-Whitney U test.
Differences in the Distribution of Involved Lymph Nodes Between the 2 Groups
Statistically significant differences were observed between the DLBCL and FL groups in the involvement of Waldeyer ring, para-iliac, and inguinal regions (χ2=5.120, 14.777, and 8.297, respectively; all P<0.05; Table 2). Further FDR multiple comparison correction revealed that, compared with DLBCL, FL was more likely to involve the para-iliac (P=0.001) and inguinal regions (P=0.022; Table 2).
TABLE 2.
Involvement Sites of Lymphoma in the 2 Groups
| Regions | DLBCL,n (%) | FL, n (%) | χ2 | P | FDR-P |
|---|---|---|---|---|---|
| Waldeyer ring | 54 (38.3) | 12 (21.4) | 5.120 | 0.024 | 0.087 |
| Neck | 88 (62.4) | 41 (73.2) | 2.070 | 0.150 | 0.267 |
| Subclavian | 16 (11.3) | 7 (12.5) | 0.052 | 0.820 | 0.820 |
| Axillary | 52 (36.9) | 28 (50.0) | 2.861 | 0.091 | 0.250 |
| Mediastinal | 49 (34.8) | 24 (42.9) | 1.129 | 0.288 | 0.396 |
| Hilar | 28 (19.9) | 14 (25.0) | 0.632 | 0.427 | 0.469 |
| Spleen | 34 (24.1) | 19 (33.9) | 1.964 | 0.161 | 0.267 |
| Upper abdomen | 65 (46.1) | 30 (53.6) | 0.896 | 0.344 | 0.420 |
| Lower abdomen | 88 (62.4) | 41 (73.2) | 2.070 | 0.150 | 0.267 |
| Para-iliac vessels | 46 (32.6) | 35 (62.5) | 14.777 | <0.001 | 0.001 |
| Inguinal | 40 (28.4) | 28 (50.0) | 8.297 | 0.004 | 0.022 |
CT Morphologic Differences in the Most Metabolically Active Lymph Nodes
Significant differences were observed between the DLBCL and FL groups in lesion fusion, internal necrosis, surrounding fat stranding, adjacent organ involvement, and nonround shape (χ2=6.92, 35.63, 26.30, 12.03, and 5.61, all P<0.05; Table 1). However, there were no significant differences between the 2 groups in the CT attenuation values (Z=−0.81, P=0.416) or short-axis diameter of the lesions (Z=−1.93, P=0.053; Table 1).
Differences in 18F-FDG Metabolic Characteristics of the Most Metabolically Active Lymph Nodes
Significant differences were observed between the DLBCL and FL groups in SUVmax, SUVpeak, SULmax, SULpeak, SUVmean, SULmean, L/L, L/M, and lesionTLG (Z=−4.20 to −7.53, all P<0.05; Table 1). The area under the curve (AUC) values for these parameters were 0.844, 0.836, 0.832, 0.836, 0.834, 0.835, 0.802, 0.819, and 0.692, respectively. However, no statistically significant difference was found between the 2 groups in lesionMTV (Z=−1.52, P=0.128; Table 1). Representative PET/CT images of typical cases from both groups are shown in Figures 1 and 2.
FIGURE 1.
A representative case of a 67-year-old woman with diffuse large B-cell lymphoma (DLBCL). A, Maximum intensity projection (MIP) image showing lymph node involvement in cervical para-iliac, and inguinal (arrow) regions. B, High-resolution CT (soft tissue window) showing an irregular lesion with internal necrosis (arrow), and surrounding fat stranding (arrowhead). C and D, PET and fused PET/CT images demonstrated significantly increased 18F-FDG uptake in the left inguinal lymph node (arrow), which exhibited the highest metabolic activity in the entire body (SUVmax=14.3). A subsequent biopsy confirmed the diagnosis of DLBCL.
FIGURE 2.
A representative case of a 55-year-old woman with follicular lymphoma (FL). A, Maximum intensity projection (MIP) image showing lymph node involvement in the cervical, axillary, mediastinal, upper abdominal, lower abdominal, para-iliac, and inguinal (arrow) regions. B, High-resolution CT (soft tissue window) demonstrating a homogeneous lesion (arrow) with a well-defined shape and a clear surrounding fat interface. C and D, PET and fused PET/CT images showed mildly increased 18F-FDG uptake in the right inguinal lymph node (arrow), which exhibited the highest metabolic activity in the entire body (SUVmax=7.9). Subsequent lymph node excisional biopsy confirmed the diagnosis of FL.
Logistic Regression and ROC Curve Analysis
Using FL as the reference group, SUVmax [odds ratio (OR)=1.230, 95% CI: 1.127-1.341, P<0.001; Table 3], necrosis (OR=3.506, 95% CI: 1.537-7.998, P=0.003; Table 3), and surrounding fat stranding (OR=2.497, 95% CI: 1.092-5.709, P=0.030; Table 3) were identified as independent risk factors for predicting DLBCL. The AUC for SUVmax, necrosis, and surrounding fat stranding were 0.844, 0.730, and 0.699, respectively. When these 3 factors were combined, the diagnostic model achieved an AUC of 0.882, with a sensitivity of 69.5% and a specificity of 94.6%, for diagnosing DLBCL (Fig. 3).
TABLE 3.
Univariate and Multivariate Logistic Regression Analysis of DLBCL
| Univariate Analysis | Multivariate Analysis | |||
|---|---|---|---|---|
| Variables | OR (95% CI) | P | OR (95% CI) | P |
| Fusion | 2.31 (1.23-4.35) | 0.009 | 0.96 (0.41-2.27) | 0.930 |
| Necrosis | 7.29 (3.65-14.57) | <0.001 | 3.51 (1.54-8.00) | 0.003 |
| Surrounding fat stranding | 5.41 (2.76-10.61) | <0.001 | 2.50 (1.09-5.71) | 0.030 |
| adjacent organ involvement | 5.70 (1.94-16.77) | 0.002 | 1.69 (0.43-6.64) | 0.450 |
| Nonround shape | 2.12 (1.13-3.98) | 0.019 | 1.73 (0.77-3.91) | 0.184 |
| lesionMTV | 1.00 (1.00-1.01) | 0.183 | 1.00 (1.00-1.00) | 0.948 |
| SUVmax | 1.26 (1.16-1.37) | <0.001 | 1.23 (1.13-1.34) | <0.001 |
| SUVmean * | 1.42 (1.25-1.62) | <0.001 | ||
| SUVpeak * | 1.29 (1.18-1.41) | <0.001 | ||
| SULmax * | 1.37 (1.22-1.54) | <0.001 | ||
| SULmean * | 1.67 (1.39-2.02) | <0.001 | ||
| SULpeak * | 1.45 (1.26-1.65) | <0.001 | ||
| L/L* | 1.58 (1.31-1.90) | <0.001 | ||
| L/M* | 1.37 (1.21-1.54) | <0.001 | ||
| lesionTLG* | 1.00 (1.00-1.00) | 0.090 | ||
Variables marked with an asterisk exhibited significant collinearity with SUVmax and were excluded from the multivariate logistic regression analysis.
L/L indicates lesion SUVmax/liver SUVmax; L/M, lesion SUVmax/mediastinum SUVmax.
FIGURE 3.
ROC curves for SUVmax, necrosis, surrounding fat stranding, and their combined diagnostic performance in identifying DLBCL.
DISCUSSION
Currently, 18F-FDG PET/CT is widely used for lymphoma staging before treatment, for the evaluation of therapeutic effects, and for the assessment of prognosis and recurrence. Compared with conventional imaging modalities such as CT and MRI, PET/CT combines the advantages of anatomic and functional imaging and improves the identification of early lesions by reflecting abnormal glucose metabolism within tumor cells through the tracer-18F-FDG,16 a glucose analog that enters cells through mechanisms similar to that of glucose transport. Upon entering the cell, 18F-FDG is phosphorylated by hexokinase into 6-phosphofluoroglucose. However, unlike glucose, it cannot proceed through the tricarboxylic acid cycle for further metabolism or be utilized for glycogen synthesis, and this leads to intracellular accumulation. Most malignant tumors exhibit an increased expression of glucose transporters on the cell membrane or elevated levels of hexokinase and phosphofructokinase, which results in elevated 18F-FDG uptake.17 Numerous studies have demonstrated a positive correlation between the degree of 18F-FDG uptake and the aggressiveness of lymphoma.18,19 Furthermore, 18F-FDG PET/CT can detect metabolic abnormalities before anatomic structural changes, such as identifying highly metabolic lymph nodes of normal size on CT, which improves the early detection of the disease.20 Furthermore, targeted biopsy guided by 18F-FDG PET/CT at sites with the highest metabolic activity significantly increases the likelihood of a more accurate pathologic diagnosis.21
NHL is classified into aggressive and indolent subtypes, based on the degree of invasiveness. DLBCL is the most common aggressive lymphoma, whereas FL is the most prevalent indolent lymphoma and the second most common lymphoma overall, following DLBCL. The transformation of FL into DLBCL remains a major challenge in the diagnosis, treatment, and management of lymphoma. A Spanish study involving over 1000 patients reported a 5-year cumulative transformation rate of 7.3% for FL.22 Transformed FL exhibits a biological behavior that is similar to that of DLBCL, and the treatment strategies for both conditions are comparable.23,24 Therefore, the early identification of FL transformation is crucial.
The current gold standard for lymphoma diagnosis is pathologic examination. However, in clinical practice, with improvements in living standards, an increasing number of patients are reluctant to undergo invasive examinations. In addition, some patients face challenges in obtaining pathologic results owing to difficulties in obtaining tissue samples, which makes it impossible to determine the pathologic subtype of NHL. As a noninvasive whole-body functional imaging technique, 18F-FDG PET/CT can simultaneously visualize lesion involvement and metabolic activity, which offers distinct advantages for diagnosing lymphoma aggressiveness. Therefore, analyzing the imaging characteristics of 18F-FDG PET/CT in these 2 most common lymphoma subtypes will provide valuable guidance for hematologists in evaluating the tumor aggressiveness of NHL and identifying histologic transformation.
The SUVmax in aggressive lymphomas has been previously reported as generally >10.7,18,25 When the SUVmax cutoff value was used as the demarcation in ROC analyses, the AUC was 0.80 to 0.86. In this study, the SUVmax cutoff value was 12.35, with an AUC of 0.844, which is consistent with previous findings. Beyond the SUVmax, this study further explored the diagnostic utility of SUVmean, SUVpeak, SULmax, SULmean, SULpeak, L/L, L/M, lesionTLG, and lesionMTV. Among these parameters, only lesionMTV showed no statistically significant differences between the DLBCL and FL groups. MTV reflects the tumor load by measuring the 3-dimensional volume of metabolically active regions in the lesion. However, as this study only included a single lymph node with the highest metabolic activity, the most metabolically active lymph node may not necessarily be the largest, which potentially leads to a lack of statistical significance for lesionMTV in distinguishing DLBCL from FL. As lymphoma is a systemic disease, the whole-body metabolic tumor volume (WBMTV) may differ between subtypes.26 Furthermore, WBMTV is more prominently associated with prognosis,27,28 whereas this study specifically analyzed baseline 18F-FDG PET/CT imaging differences between DLBCL and FL without considering the prognostic implications. The TLG, calculated by multiplying the MTV by the SUVmean, quantifies the total metabolic activity of the lesion by integrating the lesion volume and metabolic intensity, and this may have contributed to the significant differentiation between DLBCL and FL in this study. SUVmax, SUVmean, SUVpeak, SULmax, SULmean, SULpeak, L/L, and L/M reflect 18F-FDG metabolism from different perspectives. This study demonstrated that all 7 metabolic parameters differed significantly between DLBCL and FL with similar AUC values. However, the SUVmax remains the most widely used and clinically accepted metabolic parameter. Therefore, the SUVmax was considered the optimal choice, which aligns with the findings of Jia et al.7 Moreover, as these parameters exhibited significant collinearity with SUVmax, SUVmax was selected as the sole metabolic variable for the multivariate logistic regression analysis.
Furthermore, this study analyzed the CT morphologic characteristics and revealed statistically significant differences between DLBCL and FL in terms of necrosis, surrounding fat stranding, nonround shape, lesion fusion, and adjacent organ involvement. As an aggressive lymphoma, DLBCL exhibits a high proliferation rate, which may contribute to tumor hypoxia and necrosis. In addition, its rapid growth can compromise the integrity of the surrounding fat plane.29 Consequently, the incidences of necrosis and surrounding fat stranding are higher in patients with DLBCL. In a CT-based study of head and neck DLBCL and FL conducted by Imada et al,9 necrosis was identified as an independent risk factor for DLBCL, whereas surrounding fat stranding was not. This discrepancy may be attributed to differences in anatomic regions. The head and neck region is densely structured, which may limit the visualization of fat planes compared with other body regions and potentially affect the assessment of the surrounding fat stranding. No statistically significant difference in the short-axis diameter of the lesions was found between the 2 groups. This may be attributed to the target lesion-selection method that was used in this study, which was based on the SUVmax rather than morphologic size. The most metabolically active lymph node was not necessarily the largest. Furthermore, because only one lymph node per patient was included in the analysis, selection bias may have influenced the results.
Furthermore, we investigated differences in the lymph node distribution between patients with DLBCL and FL. According to previous studies and clinical experience, compared with aggressive lymphomas, FL typically presents with mild-to-moderate lymph node enlargement and is more likely to affect the superficial lymph nodes.30 In the present study, we categorized all 18F-FDG-positive lymph nodes based on their anatomic regions using the Ann Arbor staging system.31 The results showed significant differences between DLBCL and FL in the involvement of the Waldeyer ring, para-iliac, and inguinal regions. As multiple independent χ2 tests may increase the false-positive rate, we applied FDR correction. The corrected results indicated that FL, rather than DLBCL, was more likely to involve the para-iliac and inguinal lymph nodes.
Radiomics and deep learning have emerged as the key research areas in lymphoma imaging. A radiomics and deep-learning model that de Jesus et al32 developed to differentiate DLBCL from FL achieved an AUC of 0.86. In this study, we conducted multivariate logistic regression analyses that incorporated CT morphologic features and 18F-FDG metabolic parameters and identified SUVmax, surrounding fat stranding, and necrosis as independent risk factors for DLBCL. The combined AUC of these 3 variables was 0.882, which was higher than that of any single parameter and approached that of the aforementioned radiomics-based study. However, some methodological differences exist between the 2 studies. First, the previous study included the 5 most metabolically active lymph nodes, whereas our study analyzed only the single most metabolically active lymph node. Second, the radiomics study used machine-learning–based image-segmentation methods, which may differ from the manual selection process that we used in our study. Unlike radiomics and machine-learning models, which rely on specialized software and computational analyses, our findings are amenable to clinical interpretation using conventional imaging assessments. Given the current clinical setting, our method is more feasible for widespread clinical application and may offer greater practical value in guiding lymphoma diagnosis and management.
Although the SUVmax cutoff value identified in this study aligns with previous findings, a universal quantitative standard for distinguishing aggressive from indolent lymphomas is yet to be established. Several factors may have contributed to those findings. The differences in scanning technology potentially play a significant role; this study used 3D acquisition, which generally produces higher SUV measurements than conventional 2D acquisition. Advanced reconstruction techniques, such as time-of-flight (TOF) and point spread function (PSF), modeling further enhance contrast recovery and frequently result in higher SUVmax values. Furthermore, an intrinsic metabolic overlap exists between aggressive and indolent lymphomas because high-grade FL frequently exhibits SUVmax values similar to those of DLBCL. Taken together, these factors highlight the importance of pathologic biopsy for a definitive diagnosis of lymphoma. Nevertheless, unlike prior studies that primarily focused on metabolic PET parameters alone, the present study integrated CT morphologic features with 18F-FDG metabolic indices within a multivariable framework. This combined imaging approach allows for a more comprehensive characterization of lymphoma lesions and may provide incremental diagnostic value beyond single-parameter analysis.
This study has some limitations, which include the (1) single-center retrospective design: this retrospective study was conducted at a single center with a relatively small sample size and lacked external validation. Thus, owing to the incidence rates of the subtypes, the FL group was significantly smaller than the DLBCL group, which may have generated a selection bias. (2) Partial volume effect: the quantitative metabolic parameters for small lymph nodes may have been underestimated owing to partial volume effects. (3) Not all lymph nodes were pathologically confirmed, and the most metabolically active lymph node that was selected for analysis was not always the one that underwent pathologic confirmation, which may have increased the possibility of discordance between metabolic profiling and histologic truth. (4) Lack of transformed cases: This study constitutes preliminary research that focused on differentiating treatment-naive DLBCL from FL to establish the baseline imaging differences. However, transformed FL cases were not analyzed separately in this cohort; nonetheless, we are currently expanding our data set to validate the diagnostic utility of these findings in detecting histologic transformations. To mitigate these limitations, large, prospective multicenter studies are warranted.
CONCLUSION
Compared with FL, DLBCL was more likely to present an SUVmax >12.35, necrosis, and surrounding fat stranding; FL was more likely to involve the para-iliac and inguinal lymph nodes. Therefore, a comprehensive analysis of 18F-FDG PET/CT characteristics provides high diagnostic value for differentiating nodal DLBCL from FL. In addition, these imaging features may facilitate the early detection of FL transformation and thereby enable timely intervention and potentially improve prognosis.
Footnotes
The authors declare no conflict of interest.
Contributor Information
Hao Yu, Email: 921545109@qq.com.
Chaowei Li, Email: 773715632@qq.com.
Fei Jin, Email: jf370826@126.com.
Wenwen Jiang, Email: 406395281@qq.com.
Yudong Sui, Email: yudongs2014@163.com.
Lei Zeng, Email: zenglei0011@163.com.
Yanli Wang, Email: wangyanli1105@163.com.
Na Fang, Email: fangna7859@163.com.
REFERENCES
- 1.Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024;74:229–263. [DOI] [PubMed] [Google Scholar]
- 2.Parry EM, Roulland S, Okosun J. DLBCL arising from indolent lymphomas: how are they different? Semin Hematol. 2023;60:277–284. [DOI] [PubMed] [Google Scholar]
- 3.Carbone A, Roulland S, Gloghini A, et al. Follicular lymphoma. Nat Rev Dis Primers. 2019;5:83. [DOI] [PubMed] [Google Scholar]
- 4.Jacobsen E. Follicular lymphoma: 2023 update on diagnosis and management. Am J Hematol. 2022;97:1638–1651. [DOI] [PubMed] [Google Scholar]
- 5.Casulo C, Burack WR, Friedberg JW. Transformed follicular non-Hodgkin lymphoma. Blood. 2015;125:40–47. [DOI] [PubMed] [Google Scholar]
- 6.Lv L, Liu Y. Clinical application of liquid biopsy in non-Hodgkin lymphoma. Front Oncol. 2021;11:658234. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Jia Q, Wang A, Liu Y, et al. Diagnostic value of baseline 18 F-FDG PET/CT and peripheral blood inflammatory markers for aggressive lymphoma in non-Hodgkin’s lymphoma. Nucl Med Commun. 2025;46:60–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Alobthani G, Romanov V, Isohashi K, et al. Value of 18F-FDG PET/CT in discrimination between indolent and aggressive non-Hodgkin’s lymphoma: a study of 328 patients. Hell J Nucl Med. 2018;21:7–14. [DOI] [PubMed] [Google Scholar]
- 9.Imada H, Kato H, Ando T, et al. Computed tomography imaging findings for predicting histological subtypes and clinical outcomes in patients with head and neck nodal involvement of diffuse large B-cell lymphoma and follicular lymphoma. J Comput Assist Tomogr. 2021;45:472–476. [DOI] [PubMed] [Google Scholar]
- 10.Barrington SF, Mikhaeel NG, Kostakoglu L, et al. Role of imaging in the staging and response assessment of lymphoma: consensus of the International Conference on Malignant Lymphomas Imaging Working Group. J Clin Oncol. 2014;32:3048–3058. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Wondergem MJ, Rizvi SN, Jauw Y, et al. 18F-FDG or 3’-deoxy-3’-18F-fluorothymidine to detect transformation of follicular lymphoma. J Nucl Med. 2015;56:216–221. [DOI] [PubMed] [Google Scholar]
- 12.Kanamoto T, Matsuki M, Okuda J, et al. Preoperative evaluation of local invasion and metastatic lymph nodes of colorectal cancer and mesenteric vascular variations using multidetector-row computed tomography before laparoscopic surgery. J Comput Assist Tomogr. 2007;31:831–839. [DOI] [PubMed] [Google Scholar]
- 13.Kahle XU, Montes de Jesus FM, Kwee TC, et al. Relationship between semiquantitative 18F-fluorodeoxyglucose positron emission tomography metrics and necrosis in classical Hodgkin lymphoma. Sci Rep. 2019;9:11073. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Wahl RL, Jacene H, Kasamon Y, et al. From RECIST to PERCIST: Evolving Considerations for PET response criteria in solid tumors. J Nucl Med. 2009;(suppl 1):122S–150SS. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Boellaard R, Delgado-Bolton R, Oyen WJ, et al. FDG PET/CT: EANM procedure guidelines for tumour imaging: version 2.0. Eur J Nucl Med Mol Imaging. 2015;42:328–354. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Cheson BD, Fisher RI, Barrington SF, et al. Recommendations for initial evaluation, staging, and response assessment of Hodgkin and non-Hodgkin lymphoma: the Lugano classification. J Clin Oncol. 2014;32:3059–3068. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Zaniboni A, Savelli G, Pizzocaro C, et al. Positron emission tomography for the response evaluation following treatment with chemotherapy in patients affected by colorectal liver metastases: a selected review. Gastroenterol Res Pract. 2015;2015:706808. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Ngeow JYY, Quek RHH, Ng DCE, et al. High SUV uptake on FDG-PET/CT predicts for an aggressive B-cell lymphoma in a prospective study of primary FDG-PET/CT staging in lymphoma. Ann Oncol. 2009;20:1543–1547. [DOI] [PubMed] [Google Scholar]
- 19.Schöder H, Noy A, Gönen M, et al. Intensity of 18fluorodeoxyglucose uptake in positron emission tomography distinguishes between indolent and aggressive non-Hodgkin’s lymphoma. J Clin Oncol. 2005;23:4643–4651. [DOI] [PubMed] [Google Scholar]
- 20.Seam P, Juweid ME, Cheson BD. The role of FDG-PET scans in patients with lymphoma. Blood. 2007;110:3507–3516. [DOI] [PubMed] [Google Scholar]
- 21.Zanoni L, Mattana F, Calabrò D, et al. Overview and recent advances in PET/CT imaging in lymphoma and multiple myeloma. Eur J Radiol. 2021;141:109793. [DOI] [PubMed] [Google Scholar]
- 22.Méndez M, Torrente M, Sánchez-Beato M, et al. Transformed follicular lymphoma in the rituximab era: a report from the Spanish Lymphoma Oncology Group. Hematol Oncol. 2019;37:143–150. [DOI] [PubMed] [Google Scholar]
- 23.Fischer T, Zing NPC, Chiattone CS, et al. Transformed follicular lymphoma. Ann Hematol. 2018;97:17–29. [DOI] [PubMed] [Google Scholar]
- 24.Link BK, Maurer MJ, Nowakowski GS, et al. Rates and outcomes of follicular lymphoma transformation in the immunochemotherapy era: a report from the University of Iowa/MayoClinic Specialized Program of Research Excellence Molecular Epidemiology Resource. J Clin Oncol. 2013;31:3272–3278. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Albano D, Calabrò A, Dondi F, et al. The role of baseline 2-[18 F]-FDG-PET/CT metrics and radiomics features in predicting primary gastric lymphoma diagnosis. Hematol Oncol. 2024;42:e3266. [DOI] [PubMed] [Google Scholar]
- 26.Kim CY, Hong CM, Kim DH, et al. Prognostic value of whole-body metabolic tumour volume and total lesion glycolysis measured on ¹⁸F-FDG PET/CT in patients with extranodal NK/T-cell lymphoma. Eur J Nucl Med Mol Imaging. 2013;40:1321–1329. [DOI] [PubMed] [Google Scholar]
- 27.Cottereau AS, Nioche C, Dirand AS, et al. 18F-FDG PET dissemination features in diffuse large B-cell lymphoma are predictive of outcome. J Nucl Med. 2020;6:40–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Ehsan M, Arun S, Mahender Y, et al. Volumetric parameters on interim PET predict clinical outcomes better than SUVmax in patients with diffuse large B-cell lymphoma. Blood. 2015;126:1519.26405212 [Google Scholar]
- 29.Wu J, Meng F, Ran D, et al. The metabolism and immune environment in diffuse large B-cell lymphoma. Metabolites. 2023;13:734. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Salles GA. Clinical features, prognosis and treatment of follicular lymphoma. Hematology Am Soc Hematol Educ Program. 2007;2007:216–225. [DOI] [PubMed] [Google Scholar]
- 31.Armitage JO. Staging non-Hodgkin lymphoma. CA Cancer J Clin. 2005;55:368–376. [DOI] [PubMed] [Google Scholar]
- 32.de Jesus FM, Yin Y, Mantzorou-Kyriaki E, et al. Machine learning in the differentiation of follicular lymphoma from diffuse large B-cell lymphoma with radiomic [18F]FDG PET/CT features. Eur J Nucl Med Mol Imaging. 2022;49:1535–1543. [DOI] [PubMed] [Google Scholar]