Diffuse large B-cell lymphoma (DLBCL) is the most common subtype of aggressive non-Hodgkin lymphoma (NHL), accounts for approximately 30% of all NHL. Most patients with DLBCL can achieve a cure through chemotherapy and targeted therapies, and the widely adopted first-line treatment include the rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) regimen. With R-CHOP or modified R-CHOP regimens, over 50% of patients attain complete remission, but up to one-third may experience relapse or refractory disease (1, 2). The main reason for this is tumor drug resistance, and it has been shown that Long non-coding RNA SNHG17 plays an important role in the progression of DLBCL (3, 4). A previous study has indicated that patients undergoing autologous stem cell transplantation had a 1-year overall survival rate (OS) of only 41.6% (5). Therefore, the early identification of patients with poor prognosis or insensitivity to first-line treatment is crucial for improving outcomes (6).

Many studies have attempted to use noninvasive tests to predict the prognosis of patients with DLBCL; for example, it has been suggested that pelvic MRI is effective in detecting bone marrow involvement (BMinv) in patients with DLBCL and that it may ultimately be used to improve the accuracy of clinical staging, guide the treatment of patients, and assess prognosis (7). Currently, ¹⁸F-FDG PET/CT is a common imaging modality for the diagnosis and treatment of DLBCL and plays a significant role in staging, treatment monitoring, and treatment response assessment (8, 9). However, the value of interim PET/CT for mid-term efficacy and prognostic assessment in patients with DLBCL remains controversial, with no established gold standard for evaluation. The First International Lymphoma PET/CT Workshop, held in Deauville, France in 2009, recommended the use of the Deauville five-point scale (5-PS) to assess different responses to lymphoma treatment at mid-term and post-treatment (10). The 5-PS scoring is assigned based on the Standardized Uptake Value max (SUVmax) of the lesion with the highest uptake, as follows: 1 point: no uptake; 2 points: uptake≤mediastinum; 3 points: mediastinum<uptake≤liver; 4 points: uptake moderately higher than liver; 5 points: uptake markedly higher than liver and/or new lesions; X: new areas of uptake unlikely to be related to lymphoma. A Deauville score of 1–3 is considered complete metabolic remission (CMR), and a score of 4–5 is categorized as follows: if the uptake is lower than the baseline PET/CT, it is considered a partial metabolic response (PMR); if there is no significant change in uptake compared to the baseline PET/CT and no new or progressing lesions, it is classified as no metabolic response (NMR); if the uptake is higher than the baseline PET/CT and/or new lesions appear, it is considered a progressive metabolic disease (PMD) (10, 11). The 5-PS score does not require complex calculations, such as metabolic tumor volume (MTV) and total lesion glycolysis (TLG), and reduces the impact of different PET/CT equipment. The reports issued by different hospitals can be efficiently compared. Owing to its simplicity and efficiency, the PET/CT performed in the interim during chemotherapy has been widely investigated for response-adapted therapy in Hodgkin’s lymphoma (HL), DLBCL, and other subsets of NHL (12).

Commonly used PET/CT metabolic parameters in clinical practice include the MTV, TLG, SUVmax, and Standardized Uptake Value mean (SUVmean). SUVmax, which represents the highest uptake intensity within a Volume of Interest (VOI), is widely used in pre-treatment assessment, mid-term efficacy evaluation, and post-treatment efficacy assessment of lymphoma treatment. However, SUVmax was measured at the site of highest uptake, reflecting the metabolic intensity of the most active tumor cells. Consequently, factors, such as different PET/CT devices, tumor heterogeneity, image algorithms, and scan intervals, can significantly affect the SUVmax (5, 13, 14). In recent years, with the widespread clinical application of ¹⁸F-FDG PET/CT, an increasing number of studies have shown that using MTV and TLG for the prognostic evaluation of patients with lymphoma can improve the accuracy of ¹⁸F-FDG PET/CT in predicting lymphoma prognosis (15–17).

Currently, there is no gold standard for determining the threshold for the MTV, and different thresholds may yield significantly different MTV and TLG measurements for the same patient. Three common methods have previously been used to determine the marginal threshold for the MTV. One method uses an absolute cutoff value of 2.5 as the SUV threshold, defining tissues with an absolute SUV value greater than 2.5 as tumor lesions (18–20). However, many factors can significantly affect the absolute SUV cutoff value, such as the time interval between tracer injection and scanning, different PET equipment, and injection malfunctions, rendering the use of the absolute SUV cutoff value as a threshold for measuring the MTV inaccurate. The second method uses a certain percentage of SUVmax in the most metabolically active lesion to determine MTV, commonly ranging from 25% to 75% (21–23). However, owing to the variations in uptake in different lesions and pathological subtypes, the ideal percentage may differ, and there is currently no unified standard for the best SUVmax ratio for measuring the MTV. When a higher percentage of SUVmax is used, there is a risk of underestimating tumor volume, whereas a lower percentage may lead to an overestimation of tumor volume. The third method calculates the patient’s own liver SUVmean and adds two to three times the standard deviation (SD) to obtain the threshold (24, 25). This method can significantly reduce the impact of different PET/CT technologies and subjective factors. The Positron Emission Tomography Response Criteria in Solid Tumors (PERCIST) also recommends using MTV and TLG to predict patient prognosis (26). However, the predictive abilities of these parameters vary across studies, and there is no unified standard for measuring the MTV or TLG.

In this study, a PET/CT Lesion Quantifier, which significantly reduced the calculation complexity and improved the calculation speed, was used to calculate the MTV and TLG. We investigated the relationship between baseline MTV, TLG, and interim ΔMTV and ΔTLG(after 3-4 cycles of treatment) with the prognosis of patients with DLBCL and attempted to compare with the 5-PS score. This study aimed to explore which parameters, MTV or TLG, are more suitable for predicting the prognosis of patients with DLBCL and investigate the value of interim PET/CT for mid-term efficacy and prognosis assessment in patients with DLBCL. Inspired by the Deauville score, which uses mediastinal and liver uptake values as scoring criteria, we used the mean, third quartile, maximum, 1.5 × the mean, and two times the mean of the mediastinal and liver uptake values as thresholds to measure the MTV and TLG, respectively. After screening, we chose the third quartile of liver uptake values as the threshold for the measurement of MTV and TLG in patients for the next step of the study (the specific screening process is shown in the Supplementary Material ).

In DLBCL, whether SUVmax is an independent prognostic factor remains controversial. However, previous studies have shown that SUVmax is an independent prognostic factor (29, 30). Recently, with an increased understanding of volumetric metabolic parameters and observations from large samples, researchers have concluded that volumetric metabolic parameters can improve the accuracy of DLBCL predictions (22). They argued that neither SUVmax nor SUVmean could effectively predict the treatment response, PFS, and OS (20). Ceriani et al. (18) extensively studied whether SUVmax could serve as a predictor of PFS and OS in a DLBCL population, including 141 patients in a training cohort and 113 patients in a validation cohort, both with similar Ann Arbor stages. In both cohorts, SUVmax was not a significant predictor of PFS or OS. In a larger study evaluating 169 patients with DLBCL treated with R-CHOP (Ann Arbor stages II and III without extranodal lesions), Song et al. (20) found that patients with an MTV <220 cm³ had significantly better PFS and OS. Even after multivariate Cox regression analysis for stages II and III disease, the correlation between MTV and prognosis remained significant. Therefore, they concluded that the MTV is an independent predictor of PFS and OS in patients with DLBCL, regardless of the Ann Arbor stage. However, in a study involving 91 patients, Zhou et al. (31) reported that despite the association of high baseline MTV and TLG with poor prognosis, only TLG was an independent predictor of PFS and OS. In this study, patients with high TLG levels were more prone to relapse during treatment, even if they achieved complete remission, compared with patients with low TLG levels (40% vs. 9%, p = 0.012). Our study utilized the third quartile of liver uptake values as the threshold to measure MTV and TLG, exploring the relationship between MTV, TLG, and the prognosis of patients with DLBCL. The results of the univariate Cox proportional hazards model analysis indicated that the SUVmax did not exhibit statistical significance in relation to either PFS or OS in the DLBCL population in this study. This could be attributed to the sensitivity of SUVmax being influenced by various factors, such as the time interval between injection and scanning, partial volume effects in small lesions, attenuation of tracer activity, technical characteristics, acquisition, and reconstruction parameters of the scanner. Additionally, SUVmax only records the intensity of ¹⁸F-FDG uptake in the most metabolically active region, making it challenging to reflect the overall tumor burden of patients, especially in cases of DLBCL where multiple lesions are common, leading to a larger tumor burden. Therefore, the accuracy of predicting prognosis based on the SUVmax measured from the most intense lesion before treatment has significant limitations.

In our study, although the MTV showed a significant correlation in the univariate analysis, the results of the multivariate analysis showed that only the TLG was an independent predictor of patient prognosis. In terms of calculation methods, MTV and TLG showed a certain correlation; however, a significant difference was found in this study between MTV and TLG in predicting the OS and PFS of patients with DLBCL. This difference may be related to the definitions of the MTV and TLG. MTV represents the volume of all pixels on PET images that exceed a preset SUV value based on the assumption of elevated metabolism in tumors than in normal tissues. On the contrary, TLG is a metabolic parameter derived by multiplying MTV with SUVmean. These MTV results, to some extent, overlook the intensity of tumor metabolism; thus, failing to accurately reflect the overall tumor burden, especially in lesions with non-uniform metabolism where differences might be more pronounced. In contrast, TLG not only reflects the metabolic activity of the tumor, but also considers the metabolic volume, aligning more closely with the principles of PET imaging and the concept of tumor burden. This more reliably reflects the patient’s tumor burden. Therefore, we believe that the TLG, relative to the MTV, is a more reliable indicator for predicting the prognosis of patients with DLBCL.

Compared with Ann Arbor staging, TLG seems to have an advantage in predicting patient prognosis. In patients with stage I-II, the 1-year PFS rate and 1-year OS rate of patients in the low TLG group were significantly higher than those in the high TLG group. The same results were found in patients with stage III-IV. This indicates that in patients with stages I–II disease, the tumor burden of some patients might be underestimated. Although these patients have fewer lymph nodes or organs involved, their tumor volume is larger, or their metabolism is more active, resulting in a larger tumor burden and relatively poorer treatment outcomes. Similarly, in patients with stage III–IV disease, the tumor burden of some patients might be overestimated. Although these patients have more lymph nodes or organs involved, their tumor volume is smaller, or their metabolism is less active, resulting in a smaller tumor burden and better treatment outcomes. By calculating TLG, it is possible to distinguish between these patients, formulate individualized treatment plans, and improve their prognosis. For example, for patients with Ann Arbor stage I or II disease but with high TLG levels, we can consider adding radiation therapy appropriately after completing all cycles of R-CHOP treatment. However, this hypothesis must be tested in a larger study.

Although some studies (32, 33) have found that interim PET/CT is not a prognostic factor for patients with DLBCL, others (34–36) have indicated that interim ¹⁸F-FDG PET/CT is an effective predictor of survival in patients with DLBCL. In our study, the high ΔTLG group had a higher 1-year OS rate than the low ΔTLG group, indicating that a higher ΔTLG is associated with a better prognosis in patients with DLBCL. The interim PET/CT examination was conducted after the third or fourth cycles of R-CHOP or R-CHOP modified regimen treatment to reduce the impact of false positives, we found that high ΔTLG is an independent predictor of favorable PFS and OS in patients with DLBCL. This implies that after 3–4 cycles of R-CHOP treatment, patients with DLBCL can be assessed for treatment effectiveness and prognosis by detecting changes in TLG parameters, allowing for timely adjustment of treatment plans, especially for patients with a possible poor prognosis. Admittedly, the baseline and interim PET/CT scans in this study were from the same group of patients, and it is possible that the same patient could draw different conclusions from the two analyses. We believe that a small number of patients with a higher tumor burden can still obtain a better prognosis if they are sensitive to treatment options. The TLG of baseline PET/CT may be more informative for the staging of DLBCL patients and the development of the initial treatment regimen, while the ΔTLG may be able to monitor the sensitivity of patients to the treatment regimen. For patients who are not responsive to first-line regimens, the treatment plan may be adjusted in advance to prolong their survival. For patients who originally needed maintenance treatment, if the baseline PET/CT shows that the tumor burden is small, and the interim PET/CT shows that they are sensitive to treatment, we can consider not performing maintenance treatment after completing the entire cycle of treatment. However, this hypothesis must be verified in larger studies.

Compared with 5-PS score, ΔTLG also has a significant correlation with a patient’s prognosis. Especially in terms of observing the long-term survival of patients, ΔTLG seems to have more advantages. Patients with a 5-PS score >3 are usually considered to have a poor prognosis. Therefore, we grouped patients with a 5-PS score >3 points again according to ΔTLG. It was found that among patients with a 5-PS score >3 points, there is still a significant difference in the prognosis between the high ΔTLG group and the low ΔTLG group. Based on these results, we considered that some patients with a 5-PS score >3 achieved long-term survival. As in the cases illustrated in Figures 7 and 8 , clinicians can often encounter situations in which a portion of patients with a 5-PS score of ≤3 have a poor prognosis, whereas a portion of patients with a 5-PS score of >3 have a long-term survival, which may be due to the fact that PET/CT has some false negative and false positive rates. Although the negative predictive value of interim PET/CT is high (>80%), the positive predictive value is significantly lower (approximately 15%), resulting in a greater prognostic variability in patients with a 5-PS score >3. This is because of the inability of interim PET/CT to distinguish between the presence of residual surviving tumor tissue and a nonspecific inflammatory host response (12, 13). In our study, the prognostic difference between the high ΔTLG group and the low ΔTLG group was not statistically significant in patients with a 5-PS <3 score. This finding also suggests that the negative predictive value of interim PET/CT is high. However, ΔTLG does not depend on a single tumor tissue, it is based on the patient’s own liver uptake value to measure TLG. This can minimize the error caused by the false positive rate of PET/CT, which can help to screen out patients with a 5-PS score >3 points but a good prognosis. Moreover, the 5-PS score is subject to human subjectivity, and different physicians may score the same patient differently, whereas the TLG based on PET/CT Lesion Quantifier measurements significantly reduces the influence of human subjective consciousness on the results. The limitations of this study are that it was a single-center retrospective study with a small number of patients included, and the conclusions drawn need to be validated in a larger study. Nevertheless, the data in this study illustrate that ΔTLG may be another reliable indicator of interim efficacy assessment, in addition to the 5-PS score.5. We did not split the baseline PETCT data into control and experimental groups, but we did so for the interim PETCT data. This is because once again there have been many pre-vious studies demonstrating the role of baseline TLG in predicting survival in patients with non-Hodgkin’s lymphoma, including DLBCL (37–39), whereas there have been few studies on the relationship between ΔTLG and survival in patients with DLBCL, which is one of our innovations. In most patients, the results of baseline TLG and ΔTLG are not contradictory, that is, patients with high baseline TLG are more likely to have a smaller ΔTLG. We do not advocate direct comparison of these two parameters because baseline TLG tends to stratify patients before treatment, while ΔTLG mainly assesses whether the patient is sensitive to first-line treatment. But as you said, there are indeed a few patients with relatively large baseline TLG and relatively large ΔTLG. This small number of cases shows that although a few patients have a large tumor burden, they are very sensitive to the treatment regimen. These patients may have a good prognosis. In this process, ΔTLG plays a role of re-evaluation. At present, the treatment of DLBCL has entered the era of R-CHOP+X (new drugs of different types). ΔTLG may be able to quantify the efficacy of new therapies, making the short-term efficacy of different innovative therapies more comparable. We provide a case in the Supplementary Material that may help understand their relationship (case.docx).

This study has several limitations. First, it is a retrospective analysis, which may introduce selection bias. Second, the sample size is relatively small, limiting the generalizability and statistical significance of the findings. Additionally, the lack of external validation may affect the reliability of our results. Despite these limitations, this study demonstrates a strong cor-relation between ΔTLG and patient's prognosis, highlighting the significant potential of in-terim PET/CT in assessing patient's prognosis. These findings provide an important basis for future clinical applications.

Baseline TLG may be able to distinguish patients with poor prognosis among those with Ann Arbor staging of stage I-II. Moreover, ΔTLG may distinguish patients with good prognosis among those with 5-PS score >3. TLG will hopefully help clinicians develop more individualized treatment plans and improve the prognosis of DLBCL patients, and we call on more scholars to devote themselves to the study of metabolic parameters of PET/CT.