Scientific Papers

Predictive value of Dmax and %ΔSUVmax of 18F-FDG PET/CT for the prognosis of patients with diffuse large B-cell lymphoma | BMC Medical Imaging


DLBCL is a clinically and pathologically heterogeneous disease, which poses a challenge for determining treatment efficacy and prognosis. Hence, the prognostic factors in DLBCL have been a hot topic of research. The IPI score is a clinical index often used to determine the prognosis of non-Hodgkin’s lymphoma, especially DLBCL. As rituximab therapy becomes available, the IPI has also been improved by introducing variants such as R-IPI, aa-IPI and NCCN-IPI, which can better reflect the prognosis of DLBCL patients with different chemotherapy regimens [9]. In the present study, there was a significant difference in PFS between the low and high IPI subgroups, which is consistent with the findings of previous studies, suggesting that IPI has important value in prognostic assessment. However, the multifactorial analysis indicated that IPI was not an independent predictor, similar to previous results [7, 10, 11]. This study also showed that the prognosis could vary among patients with the same IPI score. The IPI score was based on the patient’s pretreatment status and did not incorporate the patient’s treatment response to chemotherapy, so its use as a prognostic evaluation index has some limitations. In addition, Ann Arbor stage and LDH levels were also associated with prognosis, but they were not independent predictors.

The imaging agent 18F-FDG of PET/CT can accumulate in tumour cells with increased metabolism and proliferation, thus indicating lesion activity more accurately than conventional imaging [12]. Therefore, baseline and interim 18F-FDG PET/CT parameters are widely used to study the prognosis of DLBCL patients. In our study, Dmax was the only independent predictor of PFS among all baseline parameters, and the risk of disease progression in patients with high Dmax was 1.410 times higher than that in the low-value group. Patients with Dmax ≥ 53.20 cm had significantly lower PFS than those with Dmax < 53.20 cm, indicating that high Dmax was associated with poor prognosis. This result was similar to Zhou et al.‘s finding (57.4 cm) [13]. Cottereau’s study [4] also led to the conclusion that high Dmax was associated with poor prognosis and indicated that there was no significant difference in Dmax among patients with different heights. In addition, baseline parameters such as TMTV0 and STMTV0, although not independent predictors, were all associated with patient PFS. This is further evidence that baseline metabolic parameters are valuable for prognostic prediction. Some previous studies [14,15,16] found that pretreatment TMTV was an independent risk factor for prognosis in DLBCL patients. Some studies [6, 17, 18] have also conclusively indicated that TTLG is an independent predictor of prognosis. Our findings are not fully consistent with the abovementioned studies, which may be due to the inconsistent methods of outlining TMTV and the different survival endpoints selected. For example, in MIKHAEELNG [16], a fixed threshold method with SUV = 2.5 was used as the absolute limit, and all metabolic regions > 2.5 were included in the metabolic volume, yet the results were often higher than the true level for patients with high background metabolic levels as well as low overall tumour metabolic levels. In this study, the percentage threshold method was used, and the 41% recommended by the EANM guidelines [19] was chosen as the outline threshold. However, when the tumour SUVmax is too large, the outlined TMTV will underestimate the actual tumour volume and vice versa. Weiler-Sagie [2] analysed 766 DLBCL patients with 18F-FDG uptake before chemotherapy and found that more than 97% of lesions exhibited high uptake of 18F-FDG. Therefore, there are many interfering factors in the outline of TMTV, while Dmax, as the distance between the centres of the two most distant lesions, can be used to visualize the spatial distribution of the disease, which is not highly dependent on the lesion contour and is not seriously affected by PET/CT instrument performance and image outline, promoting its widespread use.

Compared to baseline parameters, interim 18F-FDG PET/CT metabolic parameters can reflect tumour sensitivity to first-line treatments such as R-CHOP, thereby identifying patients who are not sensitive to first-line treatment regimens and guiding clinical changes to improve prognosis. The Deauville score [20] is a widely used clinical method to assess the efficacy of lymphoma by interim 18F-FDG PET/CT. It measures the SUVmax of the lesion and compares it with the SUVmax of the liver and mediastinal blood pool on the current imaging. The present study showed poorer PFS in patients with an interim PET/CT Deauville score ≥ 4, which is consistent with the results of previous studies [9]. However, in this study, %ΔSUVmax was the only independent predictor among all interim metabolic parameters and the Deauville score. Patients with low %ΔSUVmax had a 1.765 times higher risk of disease progression than those with high %ΔSUVmax. The results showed that PFS was significantly higher in patients with %ΔSUVmax ≥ 87.82% than in patients with %ΔSUVmax < 87.82%, similar to the results of Zhang et al. [6] (86.02%). Casasnovas et al. [5] also found that %ΔSUVmax predicted PFS in patients after chemotherapy and found better agreement by comparing three readers using %ΔSUVmax and Deauville score to assess efficacy. Rekowski et al. [21] also concluded by comparing the two that %ΔSUVmax seems to be more appropriate to assess the early metabolic response of DLBCL patients to standard R-CHOP treatment. This result may be explained by two reasons. First, the Deauville score was used to select the SUVmax of the lesion and a comparison was made with the SUVmax of the liver and mediastinal blood pool, but this only reflected the metabolism of the local tumour tissue, not the systemic tumour load. Second, factors such as blood glucose, lipids, and age [22] may confound the SUVmax values of the liver and mediastinal blood pool. In contrast, %ΔSUVmax is a semiquantitative parameter used in mid-term PET/CT imaging which is easy to calculate and can reflect the metabolic level of the tumour more objectively. However, the optimal cutoff values of %ΔSUVmax have been reported differently; for example, some scholars [5, 23, 24] reported cutoff values of 70%, 74%, 81.54%, etc., which may be related to blood glucose levels, selection of target lesions, and different instrument specifications. However, Wang et al. [25] found that the Deauville score and %ΔSUVmax were associated with the prognosis of DLBCL patients, but only the Deauville score was an independent predictor. Ng et al. [26] also showed that compared to %ΔSUVmax, the Deauville score was able to better discriminate the prognosis of DLBCL patients. There are some differences between the results of the above studies and those of the present study, which can be explained as follows: (1) In some studies, time-to-progression (TTP) was chosne as the follow-up endpoint, while the endpoint of our study was PFS; (2) The present study included patients with 3–4 cycles of postchemotherapy, while some studies included patients with 2 or 4 cycles of postchemotherapy; (3) The present study included a large number of clinical and imaging parameters, which might influence each other; (4) Meignan et al. [27] pointed out that %ΔSUVmax may be a false-positive in the condition of a low SUVmax level before treatment. Therefore, whether %ΔSUVmax can replace the Deauville score still needs to be confirmed by large-sample, multicentre studies. In addition, the present study showed that the midterm parameters SUVmax1, TMTV1, and TTLG1 were all associated with patient PFS, similar to the results of previous studies, suggesting that the interim metabolic parameters also have good predictive value for prognosis. We extended the previous studies of baseline combined with interim parameters [6, 7] by adding SMTV0, STLG0, %ΔTMTV and %ΔTTLG as potential prognostic predictors.

Out of 77 patients in this study who had a %ΔSUVmax ≥ 87.82%, indicating a good interim treatment response, 9 patients still experienced relapse or progression. Therefore, judging the prognosis based only on the interim response to chemotherapy in clinical practice is not sufficient. We aimed to develop a combined model that integrated 18F-FDG PET/CT baseline and interim metabolic parameters which could enhance predictive efficacy and identify high-risk patients. We propose a prognostic assessment model that uses these two complementary parameters from baseline and interim PET/CT scans to characterize two distinct aspects of the disease: tumour dissemination and posttreatment response. This study showed that the predictive efficacy of the combined Dmax+%ΔSUVmax was higher than that of the single parameter, and the PFS of patients in all three combined models was significantly different, with medium- and high-risk patients having significantly lower PFS than low-risk patients. Cottereau [28] suggested that combining both Dmax and MTV could further improve the risk stratification of patients. Zhang et al. [6] combined both baseline TLG and %ΔSUVmax and showed good predictive power for recurrence or progression. Zhu et al. [7] suggested that combining the maximum diameter of the largest lesion and midterm treatment response could improve the efficacy of predicting PFS and help identify patients at high risk of recurrence. Few studies have combined baseline and intermediate metabolic parameters to build a combined model to predict prognosis, and the indicators used vary, but all suggest that PET/CT baseline and intermediate parameters should be used as a reference for patient risk stratification, thus aiding in the detection of high-risk patients and guiding clinical personalized treatment.

Due to insufficient follow-up time, only PFS was observed in this study, and adequate overall survival was not yet observed, pending continued long-term follow-up to enrich the data. In addition, this study is a retrospective study, and a prospective study is feasible at a later stage to validate the findings.

In conclusion, combining 18F-FDG PET/CT baseline and interim metabolic parameters and even further including clinical and pathological indicators to establish a combined model to comprehensively assess patient prognosis may be a future research direction. By combining Dmax, which can reflect lesion dissemination, and %ΔSUVmax, which can indicate treatment response, the predictive efficacy of PFS can be improved, and the risk stratification of patients can be facilitated. This can provide a basis for clinical individualization and precision treatment.



Source link