Follicular lymphoma (FL) is one of most frequent subtypes of non-Hodgkin lymphoma (NHL) in the United States and Western Europe, accounting for around 22% of all NHLs. In China, the incidence of FL is lower than in Western countries, constituting 2.5–6.6% of all NHL cases (1, 2). The World Health Organization (WHO) classifies FL into grades 1 and 2, 3A, and 3B (3). Grades 1 and 2 are considered as indolent (slow-growing), whereas grade 3B has an aggressive course and is managed as diffuse large B-cell lymphoma (DLBCL) (4, 5). However, the optimal treatment for grade 3A FL is controversial (6, 7). Grade 3A FL has been suggested to be on the same continuum as grades 1 and 2 (8), and the contrasting findings revealed that gene expression profiling demonstrates a close relationship between FL 3A and 3B, but distinct from grades 1 and 2 (9). Prognostic models based on histologic grade and other factors have indicated that grade 3A is associated with a poor prognosis (10, 11). However, histologic grade (grades 1 and 2 versus grade 3A) arguably does not predict the disease outcome (12, 13). Nonetheless, FL grade plays an important role in treatment choice. An accurate assessment of histologic grade is challenging because of the heterogeneity of the disease and high inter-reader variability (14).

Although the survival of FL patients has markedly improved since the introduction of rituximab combined with chemotherapy, 20% of patients experience disease recurrence within 2 years, with a 5-year overall survival (OS) of just 50% (15, 16). The most common prognostic indices in current use, including the Follicular Lymphoma International Prognostic Index (FLIPI) (17) and FLIPI2 (18), cannot accurately identify patients who are at a high risk of progression or relapse (11, 19). PRIMA-Prognostic Index (PRIMA-PI) (20), a simplified scoring system including β2-MG (β2-microglobulin) and bone marrow involvement, has recently been proposed for patients treated with immunochemotherapy, but its development time is short, and the accuracy of bone marrow biopsy is still insufficient (21), so its application value still needs further validation.

¹⁸F‐fluorodeoxyglucose positron emission tomography/computed tomography (¹⁸F-FDG PET/CT) is recommended for FL staging and treatment response monitoring (22, 23). Total metabolic tumor volume (TMTV) and total lesion glycolysis (TLG) are new PET/CT metabolic parameters reflecting whole-body tumor burden that are becoming increasingly important for the prognostic assessment of lymphomas (24, 25). High baseline TMTV or TLG is associated with a significantly shorter progression-free survival (PFS) or OS in FL patients and has improved risk stratification (26–29). However, these metabolic parameters do not provide information on the spatial distribution of lesions throughout the body. The distance between two lesions that are the furthest apart (D _max) has been recommended as a novel PET metric for describing tumor dissemination in DLBCL patients. A high D _max has been linked to an unfavorable prognosis and was shown to complement the prognostic performance of TMTV in advanced-stage DLBCL (30). We conjecture that the prognostic value of the metabolic parameters in FL might be improved by combining with the feature characterizing lesion dissemination.

Given the challenges of histological grading of FL, Major et al. (31) have focused on whether PET/CT can be used as an adjunct to biopsy grading of FL. The results showed that the maximum standardized uptake (SUV_max), TMTV, and TLG were capable of differentiating grade 3A from low-grade (grade 1/2) FL, although this was based on a small number of cases, especially in the grade 3A group (11 patients).

Hence, the purpose of this study was to explore whether the new metric reflecting tumor dissemination and metabolic parameters derived from the baseline ¹⁸F-FDG PET/CT can be used to predict prognosis and histologic grade in patients with grades 1–3A FL.

Identifying FL patients at a high risk of disease progression or recurrence and those with a high pathologic grade is essential for effective clinical management. Our study demonstrated that baseline ¹⁸F-FDG PET/CT values are useful for predicting PFS in FL. In the multivariate analysis, D _max, TLG, and LDH were independent predictors of PFS. A novel scoring system for predicting PFS, which incorporated a new baseline PET parameter D _max, reflecting lesion dissemination, along with the metabolic parameter TLG and serum LDH showed superior performance to FLIPI, FLIPI2, and PRIMA-PI. In addition, this study found that SUV_max was related to pathological grading, and PET/CT could be used as an auxiliary tool but cannot substitute for biopsy.

¹⁸F-FDG PET/CT provides important information about tumor burden. In the present study, baseline TLG was the most robust predictor of outcome in FL patients, whereas the predictive value of TMTV was limited. Consistent with our observations, a retrospective study found that TLG, rather than TMTV, was the independent prognostic factor for FL patient survival (27). A multicenter study reported that TMTV was a robust predictor of outcome in FL, and combining TMTV with FLIPI2 score showed a good performance in identifying patients at a high risk of early progression; however, these investigators did not explore TLG (26). Another study found that both TMTV and TLG were strong predictors of PFS and OS in FL (28). The discrepancy between these reports may be explained by the different cutoff values and thresholding methods used for TMTV or TLG, differences in the distribution of risk groups, and heterogeneity of the treatment strategies.

In this study, we used a new feature of PET/CT images, D _max, to represent disease dissemination. D _max is a three-dimensional feature, which is easily measured and not influenced by the patient’s height. Unlike radiomic features, which are difficult to interpret from a biological perspective, D _max is a measure of the extent of the disease. D _max had a strong predictive power for PFS, which was better than that of the Ann Arbor stage, even among patients with advanced-stage FL. In line with our results, a high D _max predicted a shorter PFS and OS in DLBCL patients (30, 34). It has also been reported that a high D _max is a poor prognostic factor of HL (36). Unlike TMTV or TLG, which are derived from lesion contours, D _max is calculated as the distance between two lesions that were furthest apart. Most importantly, the measurements of D _max appeared to have good reproducibility and is thus broadly applicable.

Given the limitations of FLIPI and FLIPI2, other prognostic models for survival have been developed based on genomic or imaging data (19, 26, 29, 37, 38). In a multicenter study, TMTV and FLIPI2—which predicted markedly different PFS—were combined into a joint score (26). A prognostic model was developed for the same cohort that integrated baseline TMTV and end-of-induction PET (29). In the present study, we extracted two features from baseline PET/CT images—tumor burden and dissemination—representing two distinct and complementary aspects of the disease. We established a novel scoring system for predicting PFS based on D _max, TLG, and LDH, which had a higher predictive accuracy than FLIPI, FLIPI2, and PRIMA-PI. Currently, most patients requiring treatment receive immunochemotherapy, which results in long-lasting remission and improved OS (5). Our prognostic scoring system was applicable and showed excellent performance in this population receiving immunochemotherapy. These results suggest that current prognostic indices can be further refined using tumor burden parameters and disease dissemination features obtained from PET/CT images. Due to the limited number of cases in this study, further research is needed to validate our results.

FDG uptake was shown to be higher in aggressive as compared to indolent NHL (39, 40). A widely varied SUV_max between low-grade and grade 3A FL was observed in many studies (31, 39). In a previous study, the cutoff value of SUV_max was 10.4, with 64% sensitivity and 74% specificity. In addition, TMTV and TLG had a comparable performance to SUV_max in making this distinction, with similar sensitivity and specificity values (31). However, TMTV and TLG could not discriminate low-grade from grade 3A FL in our study. The discrepancy between these findings and ours may be attributable to the differences in sample size and individual variability. The median SUV_max was significantly higher in the high-grade group than in the low-grade group (P = 0.013), but an extensive overlap existed, and there was a relatively low sensitivity for differentiating the grades of FL. Inevitably, some discordance in PET appearance may exist between the site of maximal FDG uptake and the actual site of tissue sampling. The histopathologic grading of different lesions in the same patients may be contrasting, resulting in an inconsistency between grading and clinical behavior. In addition, with the progression of the disease, some low-grade FL may be transformed into aggressive lymphoma (41, 42). The feasibility of using an absolute SUV_max cutoff value for grading FL is challenging. PET/CT may be a useful adjunct, but not a replacement for biopsy, to distinguish grade 3A from low-grade FL.

Platelet count was shown to be an independent prognostic indicator of outcome in peripheral T cell lymphoma and DLBCL (43, 44). However, platelet count was not a predictor of PFS in FL in the present study. The relationship between the platelet count and pathologic grade has been seldom discussed. We found that a low platelet count was an independent predictor of grade 3A FL. However, the underlying mechanism between the platelet count and the pathologic grade in FL needs further research.

Our study had some limitations. Firstly, this was a retrospective study with a relatively small sample size. Secondly, our predictive model was only for PFS. External validation in a larger population at multiple institutions is required.

TLG and D _max obtained from PET/CT data represent two complementary aspects of the disease, capturing the whole-body tumor burden and lesion dissemination. TLG, D _max, and serum LDH were independent prognostic factors of PFS. We established a novel scoring system for predicting PFS based on TLG, D _max, and LDH, which showed a superior performance and clinical benefit compared to existing indices (FLIPI, FLIPI2, and PRIMA-PI).

Additionally, our results suggest that PET/CT may be a useful adjunct, but not a replacement, for biopsy in distinguishing grade 3A from low-grade FL.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the Union Hospital, Tongji Medical College, Huazhong University of Science and Technology. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

HL: conceptualization, software, formal analysis, and writing—original draft preparation. MW: data curation, methodology, and writing—original draft preparation. YZ: visualization and investigation. FH: methodology and visualization. KW: software and validation. CW: data curation and validation. ZG: conceptualization, methodology, writing—review and editing, and supervision. All authors contributed to the article and approved the submitted version.

This work was supported in part by the National Natural Science Foundation of China (No. 81771866).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.