Prognostic value of the baseline 18F-FDG PET/CT metabolic tumour volume (MTV) and further stratification in low-intermediate (L-I) and high-intermediate (H-I) risk NCCNIPI subgroup by MTV in DLBCL MTV predict prognosis in DLBCL

Diffuse large B-cell lymphoma (DLBCL) is the most common type of non-Hodgkin's lymphoma, accounting for approximately one-third of adult cases of non-Hodgkin's lymphoma [1]. In the past 20 years, the long-term remission rate and cure rate of this aggressive lymphoma have been improved by adding rituximab to the standard CHOP (cyclophosphamide, doxorubicin, vincristine, and prednisone) strategy [2]. However, despite the significant improvement, more than 30% of patients receiving R-CHOP or R-CHOP-like chemotherapy have become refractory or have experienced recurrent lymphoma, leading to treatment failure [3]. Therefore, the accurate risk stratification of these patients before treatment and the early detection of patients who are unlikely to be cured are important factors guiding treatment. Since 1993, the International Prognostic Index (IPI) has been used to predict the prognosis of aggressive NHL treated with doxorubicin-based therapy; this index is based on five stratified clinical characteristics: age, Ann Arbor stage, lactate dehydrogenase (LDH), number of extranodal sites, and Eastern Cooperative Oncology Group (ECOG) performance status [4]. However, the IPI was developed before the era of rituximab. After adapting to the new era, the index was refined, and a revised IPI (R-IPI) [5]and NCCNIPI [6] were produced, which improved and refined the risk stratification. However, some refractory and relapsed patients still have not been identified, and the stratified approach based on clinical characteristics and biochemical indicators does not include all the patient's features, such as tumour metabolic burden. In recent years, FDG PET/CT has been widely used and recommended by the International Lymphoma Working Group for lymphoma in terms of pretreatment staging, treatment response evaluation, and prognosis prediction [7]. For DLBCL, many studies have shown that semiquantitative volume parameters such as metabolic tumour volume (MTV) and total metabolic glycolysis (TLG) are independent risk factors for patient prognosis [8‐10]. A higher MTV is significantly associated with poor PFS and/or OS. Though the volume parameter shows its prognostic ability in overall DLBCL patients, according to our knowledge, and further stratification by MTV in subgroup has not been confirmed and studied, the aim of our study was to provide evidence to support this point.

In this article, the tumour burden metabolic parameter and the NCCNIPI score were independent risk factor for the prognosis of DLBCL and helped in the prediction of PFS and OS. The NCCNIPI score is composed of five indicators. In univariate analysis, although ECOG status and LDH level were not significant (p > 0.05) due to the small sample size, when the five indicators were combined to form the NCCNIPI score, the prediction efficiency of the score had statistical significance (P = 0.00006), which shows that the ability of comprehensive scoring is more convincing than a single indicator for predicting prognosis, avoiding random errors caused by insufficient sample size.

When we divided the patients by NCCNIPI into four subgroup as low(0–1), low-intermediate (2–3), high-intermediate (4–5), and high (6–8) risk group, the survival curve was separated clearly, p < 0.0001, showing results similar to those of Zhou et al. [6].

As the cells of origin (COO) are controversial, the Hans (CD10, BCL6, and MUM1/IRF4) method was used to classify the cells of origin (GCB/non-GCB) in our article [11]. In the univariate analysis, there was no statistical significance, which was not enough to confirm the predictive effect on PFS and OS. Previous studies have suggested that the use of immunohistochemistry (IHC) to perform cell origin typing cannot predict patient survival [15, 16], but gene expression profile (GEP)-related technologies such as rapid reverse transcriptase multiplex ligation-dependent probe amplification assay (RT-MLPA) technology [17], DASL (cDNA-mediated annealing, selection, ligation and extension) [18], etc., which classify the origin of tumour cells, have a clear role in predicting the prognosis of DLBCL. Our results also suggest that the clinical immunohistochemical classification cannot be used as an index of patient prognostic risk classification.

In DLBCL, whether the baseline PET imaging characteristic SUVmax is an independent prognostic factor is controversial; early studies have suggested that SUVmax is an independent prognostic factor [19, 20]. In recent years, with the recognition of volume metabolism parameters and the observation of large samples, it has been concluded that volume metabolism parameters can improve the accuracy of DLBCL prediction [9].Regardless of SUVmax and SUVmean, neither the treatment response [21] nor PFS and OS [17] can be predicted.

The baseline PET/CT volume parameter MTV has been confirmed as an independent prognostic factor for PFS and OS in DLBCL in numerous studies [9]. This conclusion was confirmed in a meta-analysis of 13 DLBCL studies [22] and in a recent large-phase III clinical trial (NCT01287741) [23]. In this study, the cutoff value of the MTV was 64.1 cm³. The higher metabolic volume group had a worse prognosis than the lower metabolic volume group. The definition of tumour MTV is based on the threshold. The threshold is selected by several methods: SUV ≥ 2.5, SUV ≥ 41% of the SUVmax, and SUV ≥ the average liver uptake background. All methods have similar accuracy in terms of predicting PFS and OS [24].The 41% SUVmax method selected in this study is based on recommendations from the European Association of Nuclear Medicine [14].

The advantage of this article lies on further confirmation of the prognosis value of MTV as the most reliable volume parameter in the subgroup of NCCNIPI system, which excludes the interference of different subgroups determined by clinical indicators on the prognosis stratification. In previous studies [25], a similar subgroup analysis was performed, but it did not strictly follow the NCCNIPI subgroup standard. In our 2 subgroups (L-I risk and H-I risk), MTV also distinguish their prognosis by the 64.1 cm³ as the cutoff value of higher volume and lower volume. In the L-I risk group, the p values were 0.068 for PFS and 0.078 for OS, and in the H-I risk group the P value was 0.069 and 0.036, which were all statistically significant at a p < 0.1 level. This is a small sample and retrospective study; when we stratified with a relatively small sample size in the NCCNIPI subgroup according to the MTV parameters and performed survival analysis, we could already get a clear difference in survival, and the two survival curves were separated clearly. Although part of them, the P value were not less than 0.05, we thought 0.1 could also illustrated the difference, which have provided ideas and directions for further multi-centre prospective large sample researches.

Recently, many new biochemical molecular markers, such as ctDNA [26], have appeared to predict the prognosis of DLBCL. Image-related metrics also constantly emerge. In the future, the combined prediction of clinical indicators as well as blood biochemical, pathological and functional imaging metrics or radiomics metrics [21] will become a trend.

The limitation of this article is the single-centre retrospective design. A further validation is needed to be carried out in multicentre prospective cohorts.

The baseline volume metabolic parameter MTV and the NCCNIPI score are independent prognostic risk factors for DLBCL patients treated with R-CHOP or R-CHOP-like regimens. In our study, we further confirmed the risk stratification function of MTV as an FDG metabolic parameter in the NCCNIPI L-I and H-I risk subgroups, which support that physicians make the decision not only using a traditional scoring system, but also by MTV as a supplement for the patients' individual therapy strategy.